{"url": "https://arxiv.org/abs/1503.03417", "title": "Upper Bounds on the Relative Entropy and Rényi Divergence as a Function of Total Variation Distance for Finite Alphabets", "abstract": "A new upper bound on the relative entropy is derived as a function of the total variation distance for probability measures defined on a common finite alphabet. The bound improves a previously reported bound by Csiszár and Talata. It is further extended to an upper bound on the Rényi divergence of an arbitrary non-negative order (including $\\infty$) as a function of the total variation distance.", "text": "\\section{Introduction}\nConsider two probability distributions $P$ and $Q$ defined on a common\nmeasurable space $(\\ensuremath{\\mathcal}{A}, \\mathscr{F})$.\nThe Csisz\\'{a}r-Kemperman-Kullback-Pinsker inequality (a.k.a. Pinsker's inequality)\nstates that\n\\begin{align}\n\\label{eq: Pinsker}\n\\tfrac12 \\, |P-Q|^2 \\, \\log e \\leq D(P \\| Q)\n\\end{align}\nwhere\n\\begin{align} \\label{eq: relative entropy}\nD(P \\| Q) = \\ensuremath{\\mathbb{E}}_P\\left[ \\log \\frac{\\text{d}P}{\\text{d}Q} \\right]\n= \\int_{\\ensuremath{\\mathcal}{A}} \\text{d}P(a) \\, \\log \\frac{\\text{d}P}{\\text{d}Q} \\, (a)\n\\end{align}\ndesignates the relative entropy (a.k.a. the Kullback-Leibler divergence)\nfrom $P$ to $Q$, and\n\\begin{align} \\label{eq: TV distance}\n|P-Q| = 2 \\, \\sup_{\\mathcal{F} \\in \\mathscr{F}} |P(\\mathcal{F}) - Q(\\mathcal{F})|\n\\end{align}\nis the total variation distance between $P$ and $Q$.\n\nA ``reverse Pinsker inequality\" providing an upper bound on the relative\nentropy in terms of the total variation distance does not exist in general\nsince we can find distributions that are arbitrarily close in total variation\nbut with arbitrarily high relative entropy.\nNevertheless, it is possible to introduce constraints under which such\nreverse Pinsker inequalities can be obtained.\nIn the case where the probability measures $P$ and $Q$ are defined\non a common discrete (i.e., finite or countable) set~$\\ensuremath{\\mathcal}{A}$,\n\\begin{align}\n\\label{eq: relative entropy - discrete}\n& D(P \\| Q) = \\sum_{a \\in \\ensuremath{\\mathcal}{A}} P(a) \\, \\log \\frac{P(a)}{Q(a)}, \\\\\n\\label{eq: total variation - discrete}\n& |P-Q| = \\sum_{a \\in \\ensuremath{\\mathcal}{A}} \\bigl|P(a) - Q(a)\\bigr|.\n\\end{align}\n\nOne of the implications of \\eqref{eq: Pinsker} is that convergence\nin relative entropy implies convergence in total variation distance. The total\nvariation distance is bounded $|P-Q| \\leq 2$, whereas the relative\nentropy is an unbounded information measure.\n\nImproved versions of Pinsker's inequality were studied, e.g., in \\cite{FedotovHT_IT03},\n\\cite{Gilardoni10}, \\cite{OrdentlichW_IT2005}, \\cite{ReidW11}, \\cite{Vajda_IT1970}.\n\nA ``reverse Pinsker inequality\" providing an upper bound on the relative\nentropy in terms of the total variation distance does not exist in general\nsince we can find distributions that are arbitrarily close in total variation\nbut with arbitrarily high relative entropy.\nNevertheless, it is possible to introduce constraints under which such\nreverse Pinsker inequalities can be obtained.\nIn the case of a finite alphabet $\\ensuremath{\\mathcal}{A}$, Csisz\\'ar and Talata\n\\cite[p.~1012]{CsiszarT_IT06} show that\n\\begin{align} \\label{eq: CsTa}\nD(P \\| Q) \\leq \\left(\\frac{\\log e}{Q_{\\min}} \\right)\n\\cdot |P-Q|^2,\n\\end{align}\nwhere\n\\begin{align} \\label{eq: Q_min}\nQ_{\\min} \\triangleq \\min_{a \\in \\ensuremath{\\mathcal}{A}} Q(a).\n\\end{align}\n\nRecent applications of \\eqref{eq: CsTa} can be found in \\cite[Appendix~D]{KostinaV15} and\n\\cite[Lemma~7]{TomamichelT_IT13} for the analysis of the third-order asymptotics\nof the discrete memoryless channel with or without cost constraints.\n\nIn addition to $Q_{\\min}$ in \\eqref{eq: Q_min}, the bounds in this paper involve\n\\begin{align}\n\\label{eq: beta1}\n& \\beta_1 = \\min_{a\\in \\ensuremath{\\mathcal}{A}} \\frac{Q(a)}{P(a)}, \\\\\n\\label{eq: beta2}\n& \\beta_2 = \\min_{a\\in \\ensuremath{\\mathcal}{A}} \\frac{P(a)}{Q(a)}\n\\end{align}\nso, $\\beta_1, \\beta_2 \\in [0,1]$.\n\nIn this paper, Section~\\ref{section: Reksnip} derives a reverse Pinsker inequality\nfor probability measures defined on a common finite set, improving the bound in \\eqref{eq: CsTa}.\nThe utility of this inequality is studied in Section~\\ref{sec: applications}, and it\nis extended in Section~\\ref{section: RD} to R\\'{e}nyi divergences\nof an arbitrary non-negative order.\n\n\n\\section{A New Reverse Pinsker Inequality for Distributions on a Finite Set}\n\\label{section: Reksnip}\nThe present section introduces a strengthened version of \\eqref{eq: CsTa},\nfollowed by some remarks and an example.\n\n\\subsection{Main Result and Proof}\n\\begin{theorem} \\label{thm: reksnip - improved}\n{\\em Let $P$ and $Q$ be probability measures defined on a common finite set $\\ensuremath{\\mathcal}{A}$, and assume\nthat $Q$ is strictly positive on $\\ensuremath{\\mathcal}{A}$. Then, the following inequality holds:\n\\begin{align}\n\\label{eq: UB-RE-FS1}\nD(P \\| Q) & \\leq \\log \\left(1 + \\frac{|P-Q|^2}{2 Q_{\\min}} \\right) - \\frac{\\beta_2 \\log e}{2} \\cdot |P-Q|^2 \\\\\n\\label{eq: UB-RE-FS2}\n& \\leq \\log \\left(1 + \\frac{|P-Q|^2}{2 Q_{\\min}} \\right)\n\\end{align}\nwhere $Q_{\\min}$ and $\\beta_2$ are given in \\eqref{eq: Q_min} and\n\\eqref{eq: beta2}, respectively.}\n\\end{theorem}\n\n\\begin{proof}\nTheorem~\\ref{thm: reksnip - improved} is proved by obtaining upper and lower bounds on\nthe $\\chi^2$-divergence from $P$ to~$Q$\n\\begin{align} \\label{eq: chi-square}\n\\chi^2(P\\|Q) \\triangleq \\sum_{a \\in \\ensuremath{\\mathcal}{A}} \\frac{(P(a)-Q(a))^2}{Q(a)}.\n\\end{align}\nA lower bound follows by invoking Jensen's inequality\n\\begin{align}\n\\label{eq1: lb chi-square divergence}\n\\chi^2(P\\|Q) & = \\sum_{a \\in \\ensuremath{\\mathcal}{A}} \\frac{P(a)^2}{Q(a)} - 1 \\\\\n\\label{eq2: lb chi-square divergence}\n& = \\sum_{a \\in \\ensuremath{\\mathcal}{A}} P(a) \\, \\exp\\left( \\log \\frac{P(a)}{Q(a)} \\right) - 1 \\\\\n\\label{eq3: lb chi-square divergence}\n& \\geq \\exp\\left(\\sum_{a \\in \\ensuremath{\\mathcal}{A}} P(a) \\, \\log \\frac{P(a)}{Q(a)} \\right) - 1 \\\\[0.1cm]\n\\label{eq4: lb chi-square divergence}\n& = \\exp\\bigl( D(P \\| Q) \\bigr) - 1.\n\\end{align}\nAlternatively, \\eqref{eq4: lb chi-square divergence} can be obtained by combining the equality\n\\begin{align}\n\\chi^2(P\\|Q) = \\exp\\bigl(D_2(P\\|Q)\\bigr) - 1\n\\end{align}\nwith the monotonicity of the R\\'{e}nyi divergence $D_{\\alpha}(P\\|Q)$ in $\\alpha$, which implies\nthat $D_2(P\\|Q) \\geq D(P\\|Q)$.\n\nA refined version of \\eqref{eq4: lb chi-square divergence} is derived in the following. The starting\npoint is a refined version of Jensen's inequality in \\cite[Lemma~1]{ISSV15}, generalizing a result from\n\\cite[Theorem~1]{Dragomir06}), which leads to (see \\cite[Theorem~7]{ISSV15})\n\\begin{align}\n& \\min_{a \\in \\ensuremath{\\mathcal}{A}} \\frac{P(a)}{Q(a)} \\cdot D(Q\\|P) \\nonumber \\\\[0.1cm]\n\\label{eq0: re and chi^2}\n& \\leq \\log \\bigl( 1 + \\chi^2(P\\|Q) \\bigr) - D(P\\|Q) \\\\[0.1cm]\n\\label{eq: re and chi^2}\n& \\leq \\max_{a \\in \\ensuremath{\\mathcal}{A}} \\frac{P(a)}{Q(a)} \\cdot D(Q\\|P).\n\\end{align}\nFrom \\eqref{eq: re and chi^2}\nand the definition of $\\beta_2$ in \\eqref{eq: beta2}, we have\n\\begin{align}\n& \\chi^2(P\\|Q) \\nonumber \\\\[0.1cm]\n\\label{eq1: refined lb chi^2 divergence}\n& \\geq \\exp \\Bigl( D(P\\|Q) + \\beta_2 \\, D(Q \\|P) \\Bigr) - 1 \\\\[0.1cm]\n\\label{eq2: refined lb chi^2 divergence}\n& \\geq \\exp \\left( D(P \\|Q) + \\frac{\\beta_2 \\, \\log e}{2} \\cdot |P-Q|^2 \\right) - 1\n\\end{align}\nwhere \\eqref{eq1: refined lb chi^2 divergence} follows from\n\\eqref{eq0: re and chi^2} and the definition of $\\beta_2$ in \\eqref{eq: beta2},\nand \\eqref{eq2: refined lb chi^2 divergence} follows from Pinsker's inequality\n\\eqref{eq: Pinsker}. Note that the lower bound in\n\\eqref{eq2: refined lb chi^2 divergence} refines the lower bound in\n\\eqref{eq4: lb chi-square divergence} since $\\beta_2 \\in [0,1]$.\n\nAn upper bound on $\\chi^2(P\\|Q)$ is derived as follows:\n\\begin{align}\n\\chi^2(P\\|Q) & = \\sum_{a \\in \\ensuremath{\\mathcal}{A}} \\frac{(P(a)-Q(a))^2}{Q(a)} \\nonumber \\\\[0.1cm]\n\\label{eq1: ub1}\n& \\leq \\frac{\\sum_{a \\in \\ensuremath{\\mathcal}{A}} \\bigl(P(a)-Q(a)\\bigr)^2}{Q_{\\min}} \\\\[0.1cm]\n\\label{eq2: ub1}\n& = \\frac{|P-Q|}{Q_{\\min}} \\cdot \\max_{a \\in \\ensuremath{\\mathcal}{A}} |P(a)-Q(a)|\n\\end{align}\nand, from \\eqref{eq: TV distance},\n\\begin{align}\n|P-Q| \\geq 2 \\max_{a \\in \\ensuremath{\\mathcal}{A}} |P(a)-Q(a)|.\n\\label{eq: ub2}\n\\end{align}\nCombining \\eqref{eq2: ub1} and \\eqref{eq: ub2} yields\n\\begin{align}\n\\chi^2(P\\|Q) \\leq \\frac{|P-Q|^2}{2Q_{\\min}}.\n\\label{eq: ub chi-square divergence}\n\\end{align}\nFinally, \\eqref{eq: UB-RE-FS1} follows by combining the upper and lower bounds\non the $\\chi^2$-divergence in \\eqref{eq2: refined lb chi^2 divergence} and\n\\eqref{eq: ub chi-square divergence}.\n\\end{proof}\n\n\\begin{remark}\n{\\em It is easy to check that Theorem~\\ref{thm: reksnip - improved} strengthens the\nbound by Csisz\\'ar and Talata in \\eqref{eq: CsTa} by at least a factor of~2\nsince upper bounding the logarithm in \\eqref{eq: UB-RE-FS1} gives\n\\begin{align} \\label{betterthanCT}\nD(P \\| Q) \\leq \\frac{(1-\\beta_2 \\, Q_{\\min}) \\log e}{2 Q_{\\min}} \\cdot {|P-Q|^2}.\n\\end{align}}\n\\end{remark}\n\n\\vspace*{0.2cm}\nIn the finite-alphabet case, we can obtain another upper bound on $D(P\\|Q)$ as a function of the $\\ell_2$ norm\n$\\| P - Q \\|_2$:\n\\begin{align} \\label{eq: UB-RE-FS}\nD(P \\| Q) \\leq \\log \\left(1 + \\frac{\\|P-Q\\|_2^2}{Q_{\\min}} \\right) - \\frac{\\beta_2 \\log e}{2} \\cdot \\|P-Q\\|_2^2\n\\end{align}\nwhich follows by combining \\eqref{eq2: refined lb chi^2 divergence}, \\eqref{eq1: ub1}, and\n$\\|P-Q\\|_2 \\leq |P-Q|$. Using the inequality $\\log (1 + x) \\leq x \\log e$ for $x \\geq 0$\nin the right side of \\eqref{eq: UB-RE-FS}, and also loosening this bound by ignoring the\nterm $\\frac{\\beta_2 \\log e}{2} \\cdot \\|P-Q\\|_2^2$, we recover the bound\n\\begin{align} \\label{eq: UB-RE-FS-looser}\nD(P \\| Q) \\leq \\frac{\\|P-Q\\|_2^2 \\, \\log e}{Q_{\\min}}\n\\end{align}\nwhich appears in the proof of Property~4 of \\cite[Lemma~7]{TomamichelT_IT13},\nand also used in \\cite[(174)]{KostinaV15}.\n\n\\vspace*{0.1cm}\n\\begin{remark}\n{\\em The lower bounds on the $\\chi^2$-divergence in \\eqref{eq4: lb chi-square divergence}\nand \\eqref{eq2: refined lb chi^2 divergence} improve the one in \\cite[Lemma~6.3]{CsiszarT_IT06}\nwhich states that $D(P \\| Q) \\leq \\chi^2(P\\|Q) \\log e$.}\n\\end{remark}\n\n\\vspace*{0.1cm}\n\\begin{remark}\n{\\em Reverse Pinsker inequalities have been also derived in quantum information theory\n(\\cite{AE1, AE2}), providing upper bounds on the relative entropy of two quantum\nstates as a function of the trace norm distance when the minimal eigenvalues of the\nstates are positive (c.f. \\cite[Theorem~6]{AE1} and \\cite[Theorem~1]{AE2}).\nThese type of bounds are akin to the weakend form in \\eqref{eq: UB-RE-FS2}. When the variational\ndistance is much smaller than the minimal eigenvalue (see \\cite[Eq.~(57)]{AE1}),\nthe latter bounds have a quadratic scaling in this distance, similarly to\n\\eqref{eq: UB-RE-FS2}; they are also inversely proportional to the minimal eigenvalue,\nsimilarly to the dependence of \\eqref{eq: UB-RE-FS2} in $Q_{\\min}$.}\n\\end{remark}\n\n\n\\section{Applications of Theorem~\\ref{thm: reksnip - improved}}\n\\label{sec: applications}\n\n\\subsection{The Exponential Decay of the Probability for a Non-Typical Sequence}\n\\label{subsection: typicality}\n\nTo exemplify the utility of Theorem~\\ref{thm: reksnip - improved}, we bound the function\n\\begin{align}\\label{def:ldeltaq}\nL_{\\delta} ( Q ) = \\min_{P \\not \\in \\mathcal{T}_\\delta (Q )} D(P\\|Q)\n\\end{align}\nwhere we have denoted the subset of probability measures on $ (\\ensuremath{\\mathcal}{A}, \\mathscr{F})$\nwhich are $\\delta$-close to $Q$ as\n\\begin{align}\n\\mathcal{T}_\\delta (Q ) = \\Bigl\\{P\n\\colon \\forall \\, a \\in \\ensuremath{\\mathcal}{A}, \\; \\; |P(a)-Q(a)| \\leq \\delta \\, Q(a) \\Bigr\\}\n\\end{align}\nNote that $(a_1, \\ldots , a_n )$ is strongly $\\delta$-typical according to $Q$ if its\nempirical distribution belongs to $\\mathcal{T}_\\delta (Q )$. According to Sanov's theorem\n(e.g. \\cite[Theorem~11.4.1]{Cover_Thomas}), if the random variables are independent\ndistributed according to $Q$, then the probability that\n$(Y_1, \\ldots , Y_n)$, is not $\\delta$-typical\nvanishes exponentially with exponent $L_{\\delta} ( Q )$.\n\\par\nTo state the next result, we invoke the following notions from \\cite{OrdentlichW_IT2005}.\nGiven a probability measure $Q$, its \\textit{balance coefficient} is given by\n\\begin{align} \\label{balanceOW}\n\\beta_Q = \\inf_{A\\in \\mathscr{F}\\colon Q(A) \\geq \\frac12} Q(A).\n\\end{align}\nThe function $\\phi\\colon(0, \\tfrac12] \\to [\\tfrac12 \\log e, \\infty)$ is given by\n\\begin{align}\n\\label{eq: phi refined pinsker}\n\\phi(p) = \\left\\{\n\\begin{array}{ll}\n\\frac1{4(1-2p)} \\, \\log \\left( \\frac{1-p}{p} \\right), & p \\in\n\\bigl(0, \\tfrac12 \\bigr), \\\\[0.2cm]\n\\tfrac12 \\log e, & p=\\tfrac12 .\n\\end{array}\n\\right.\n\\end{align}\n\\begin{theorem}\n{\\em If $Q_{\\min}>0$, then\n\\begin{align}\n\\label{potalo}\n\\phi(1 - \\beta_Q) \\, Q_{\\min}^2 \\, \\delta^2\n&\\leq L_{\\delta} ( Q ) \\\\\n&\\leq \\log \\left(1 + 2 Q_{\\min} \\, \\delta^2 \\right)\n\\label{potaup}\n\\end{align}\nwhere \\eqref{potaup} holds if $\\delta \\leq Q_{\\min}^{-1} -1$.}\n\\end{theorem}\n\n\\vspace*{0.2cm}\n\n\\begin{proof}\nOrdentlich and Weinberger \\cite[Section~4]{OrdentlichW_IT2005}\nshow the refinement of Pinsker's inequality:\n\\begin{align} \\label{eq: OrdentlichW}\n\\phi(1-\\beta_Q) \\; |P-Q|^2 \\leq D(P \\| Q).\n\\end{align}\nNote that if $Q_{\\min} > 0$ then $\\beta_Q \\leq 1 - Q_{\\min} < 1$, and\ntherefore $\\phi(1-\\beta_Q)$ is well defined and finite.\nIf $P \\not\\in \\mathcal{T}_\\delta (Q )$ the simple bound\n\\begin{align}\n|P - Q| > \\delta Q_{\\min}\n\\end{align}\ntogether with \\eqref{eq: OrdentlichW} yields \\eqref{potalo}.\n\\par\nThe upper bound \\eqref{potaup} follows from \\eqref{eq: UB-RE-FS2}\nand the fact that if $\\delta \\leq Q_{\\min}^{-1}-1$, then\n\\begin{align}\n\\min_{P \\not\\in \\mathcal{T}_\\delta (Q )} | P - Q | = 2 \\delta Q_{\\min}.\n\\end{align}\n\\end{proof}\n\nIf $\\delta \\leq Q_{\\min}^{-1} - 1$, the ratio between the upper and lower bounds\nin \\eqref{potaup}, satisfies\n\\begin{align} \\label{eq: exponents' ratio}\n \\frac1{Q_{\\min}} \\cdot \\frac{\\log e}{2 \\, \\phi(1-\\beta_Q)} \\cdot\n\\frac{\\log \\left(1 + 2 Q_{\\min} \\, \\delta^2 \\right)}{\\tfrac12 \\log e\n\\; Q_{\\min} \\, \\delta^2} \\leq \\frac{4}{Q_{\\min}}\n\\end{align}\nwhere \\eqref{eq: exponents' ratio} follows from the fact that its second\nand third factors are less than or equal to~1 and~4, respectively. Note\nthat the bounds in \\eqref{potalo} and \\eqref{potaup} scale like $\\delta^2$\nfor $\\delta \\approx 0$.\n\n\\subsection{Distance from Equiprobable}\n\\label{subs: equiprobable}\n\nIf $P$ is a distribution on a finite set $\\ensuremath{\\mathcal}{A}$, $H ( P )$ gauges the ``distance\" from\n$U$, the equiprobable distribution, since\n\\begin{align} \\label{eq: RE-equiprobable}\nH ( P ) = \\log | \\ensuremath{\\mathcal}{A} | - D(P \\| U).\n\\end{align}\nThus, it is of interest to explore the relationship between $H ( P ) $ and $|P-U|$.\nParticularizing \\eqref{eq: Pinsker}, \\cite[(2.2)]{BretagnolleH79} (see also \\cite[pp.~30--31]{Vapnik98}),\nand \\eqref{eq: UB-RE-FS2} we obtain\n\\begin{align}\n\\label{eq: 1st ubtv-uniform}\n|P-U| &\\leq \\sqrt{ \\frac{2}{\\log e} \\cdot \\bigl( \\log | \\ensuremath{\\mathcal}{A} | - H(P) \\bigr) }, \\\\\n\\label{eq: 2nd ubtv-uniform}\n|P-U| &\\leq 2 \\sqrt{1 - \\frac1{|\\ensuremath{\\mathcal}{A}|} \\cdot \\exp\\bigl( H(P) \\bigr)}, \\\\\n\\label{eq: lbtv-uniform}\n|P-U| &\\geq \\sqrt{ 2 \\left( \\exp\\bigl( -H(P) \\bigr) - \\frac1{|\\ensuremath{\\mathcal}{A}|} \\right) },\n\\end{align}\nrespectively.\n\n\\begin{figure}[here!]\n\\includegraphics[width=8.2cm]{equiprobable_alphabet_size_4-SV.pdf}\n\\includegraphics[width=8.2cm]{equiprobable_alphabet_size_16-SV.pdf}\n\\caption{\\label{figure:equiprobable}\nBounds on $|P-U|$ as a function of $H(P)$ for $|\\ensuremath{\\mathcal}{A}| = 4$, and\n$|\\ensuremath{\\mathcal}{A}| = 16$. The point $(H(P), |P-U|)=(0, 2(1-|\\ensuremath{\\mathcal}{A}|^{-1}) )$\nis depicted on the $y$-axis. In the curves of the two plots, the bounds\n(a), (b) and (c) refer, respectively, to \\eqref{eq: 1st ubtv-uniform},\n\\eqref{eq: 2nd ubtv-uniform} and \\eqref{eq: lbtv-uniform}.}\n\\end{figure}\n\nThe bounds in \\eqref{eq: 1st ubtv-uniform}--\\eqref{eq: lbtv-uniform} are illustrated\nfor $|\\ensuremath{\\mathcal}{A}| = 4, 16$ in Figure~\\ref{figure:equiprobable}.\nFor $H(P)=0$, $|P - U| = 2(1-|\\ensuremath{\\mathcal}{A}|^{-1} )$ is shown for reference in Figure~\\ref{figure:equiprobable};\nas the cardinality of the alphabet increases,\nthe gap between $|P-U|$ and its upper bound is reduced (and this gap decays\nasymptotically to zero).\n\\par\nResults on the more general problem of finding bounds on $|H( P ) - H(Q)|$ based on $|P-Q|$\ncan be found in \\cite[Theorem~17.3.3]{Cover_Thomas}, \\cite{HoY_IT2010},\n\\cite{Prelov_PPI2008}, \\cite{Sason_IT2013}, \\cite[Section~1.7]{Verdu_book} and \\cite{Zhang_IT2007}.\n\n\n\\section{Extension of Theorem~\\ref{thm: reksnip - improved} to R\\'{e}nyi Divergences}\n\\label{section: RD}\n\n\\begin{definition}\n{\\em The R\\'{e}nyi divergence of order $\\alpha \\in [0, \\infty]$ from $P$ to $Q$ is defined\nfor $\\alpha \\in (0,1) \\cup (1, \\infty)$ as\n\\begin{align}\n& D_{\\alpha}(P || Q) \\triangleq \\frac{1}{\\alpha-1} \\; \\log \\left( \\sum_{a \\in \\mathcal{A}}\nP^{\\alpha}(a) \\, Q^{1-\\alpha}(a) \\right).\n\\label{eq: Renyi divergence}\n\\end{align}\nRecall that $D_{1}(P\\|Q) \\triangleq D(P\\|Q)$ is defined to be the analytic extension\nof $D_{\\alpha}(P || Q)$ at $\\alpha=1$ (if $D(P||Q) < \\infty$, L'H\\^{o}pital's rule\ngives that $D(P||Q) = \\lim_{\\alpha \\uparrow 1} D_{\\alpha}(P || Q)$).\nThe extreme cases of $\\alpha = 0, \\infty$ are defined as follows:\n\\begin{itemize}\n\\item If $\\alpha = 0$ then $D_0(P||Q) = -\\log Q(\\text{Support}(P))$,\n\\item If $\\alpha = +\\infty$ then\n$$D_{\\infty}(P||Q) = \\log \\left(\\sup_{a \\in \\ensuremath{\\mathcal}{A}} \\frac{P(a)}{Q(a)}\\right).$$\n\\end{itemize}}\n\\end{definition}\n\nPinsker's inequality was extended by Gilardoni \\cite{Gilardoni10}\nfor a R\\'{e}nyi divergence of order $\\alpha \\in (0,1]$\n(see also \\cite[Theorem~30]{ErvenH14}), and it gets the form\n\\begin{equation*}\n\\tfrac{\\alpha}{2} \\, |P-Q|^2 \\, \\log e \\leq D_{\\alpha}(P \\| Q).\n\\end{equation*}\nA tight lower bound on the R\\'{e}nyi divergence of order\n$\\alpha > 0$ as a function of the total variation distance\nis given in \\cite{Sason_ISIT15}, which is consistent with\nVajda's tight lower bound for $f$-divergences in \\cite[Theorem~3]{Vajda_1972}.\n\nMotivated by these findings, we extend the upper bound on the relative entropy\nin Theorem~\\ref{thm: reksnip - improved} to R\\'{e}nyi divergences of an arbitrary order.\n\n\\vspace*{0.2cm}\n\\begin{theorem}\\label{thm: ub-RD-TV-FS}\n{\\em\nAssume that $P, Q$ are strictly positive with minimum masses\ndenoted by $P_{\\min}$ and $Q_{\\min}$, respectively.\nLet $\\beta_1$ and $\\beta_2$ be given in \\eqref{eq: beta1} and \\eqref{eq: beta2},\nrespectively, and abbreviate $\\delta \\triangleq \\tfrac12 |P-Q| \\in [0,1]$.\nThen, the R\\'{e}nyi divergence of order $\\alpha \\in [0,\\infty]$ satisfies\n\\begin{align} \\label{eq: ub-RD-TV-FS}\n& D_{\\alpha}(P \\| Q) \\nonumber \\\\\n& \\leq \\left\\{\n\\begin{array}{ll}\nf_1, & \\mbox{$\\alpha \\in (2, \\infty]$} \\\\[0.3cm]\nf_2, & \\mbox{$\\alpha \\in [1,2]$} \\\\[0.3cm]\n\\min \\left\\{ f_2, f_3, f_4 \\right\\},\n& \\mbox{$\\alpha \\in \\bigl(\\tfrac12, 1 \\bigr)$} \\\\[0.3cm]\n\\min \\left\\{ 2 \\log\\left(\\frac1{1-\\delta} \\right), f_2, f_3, f_4 \\right\\},\n& \\mbox{$\\alpha \\in \\bigl[0, \\tfrac12\\bigr]$}\n\\end{array}\n\\right.\n\\end{align}\nwhere, for $\\alpha \\in [0, \\infty]$,\n\\begin{align} \\label{eq: f1}\n& f_1(\\alpha, \\beta_1, \\delta) \\nonumber \\\\\n& \\triangleq \\left\\{\n\\begin{array}{ll}\n\\frac{1}{\\alpha-1} \\; \\log\\left(1 + \\frac{\\delta (\\beta_1^{1-\\alpha}-1)}{1-\\beta_1}\\right)\n& \\alpha \\in [0,1) \\cup (1, \\infty) \\\\[0.3cm]\n\\frac{\\delta}{1-\\beta_1} \\; \\log \\frac1{\\beta_1}, & \\alpha = 1, \\\\[0.3cm]\n\\log \\frac1{\\beta_1}, & \\alpha = \\infty \\\\[0.3cm]\n\\end{array}\n\\right.\n\\end{align}\nfor $\\alpha \\in [0,2]$\n\\begin{align}\n\\label{eq: f2}\n& f_2(\\alpha, \\beta_1, Q_{\\min}, \\delta) \\nonumber \\\\[0.2cm]\n& \\triangleq \\min \\left\\{f_1(\\alpha, \\beta_1, \\delta), \\; \\log\\left(1+\\frac{2\\delta^2}{Q_{\\min}}\\right) \\right\\}\n\\end{align}\nand, for $\\alpha \\in [0,1)$, $f_3$ and $f_4$ are given by\n\\begin{align}\n\\label{eq: f3}\n& f_3(\\alpha, P_{\\min}, \\beta_1, \\delta) \\nonumber \\\\[0.2cm]\n& \\triangleq\n\\left(\\frac{\\alpha}{1-\\alpha}\\right)\n\\left[ \\log \\left(1 + \\frac{2 \\delta^2}{P_{\\min}} \\right)\n- 2 \\beta_1 \\delta^2 \\, \\log e \\right], \\\\[0.3cm]\n\\label{eq: f4}\n& f_4(\\beta_2, Q_{\\min}, \\delta) \\nonumber \\\\\n& \\triangleq \\min\\left\\{ \\log \\left(1 + \\frac{2 \\delta^2}{Q_{\\min}} \\right)\n- 2 \\beta_2 \\delta^2 \\, \\log e , \\right.\n\\nonumber \\\\[0.1cm]\n& \\hspace*{1.2cm} \\left. \\log \\left(1 + \\frac{\\min\\{\\delta, 2\\delta^2\\}}{Q_{\\min}} \\right) \\right\\}.\n\\end{align}}\n\\end{theorem}\n\n\\vspace*{0.3cm}\n\\begin{proof}\nSee \\cite[Section~7.C]{ISSV15}.\n\\end{proof}\n\n\\begin{remark}\n{\\em A simple bound, albeit looser than the one in Theorem~\\ref{thm: ub-RD-TV-FS} is\n\\begin{align} \\label{eq: simple ub}\nD_{\\alpha}(P\\|Q) \\leq \\log \\left(1 + \\frac{|P-Q|}{2 Q_{\\min}} \\right)\n\\end{align}\nwhich is asymptotically tight as $\\alpha \\to \\infty$ in the case of a binary alphabet\nwith equiprobable $Q$.}\n\\end{remark}\n\n\\begin{example}\n{\\em Figure~\\ref{figure:compare_ub_RD_finite_graph1} illustrates the bound in\n\\eqref{eq: f1}, which is valid for all $\\alpha \\in [0, \\infty]$ (see\n\\cite[Theorem~23]{ISSV15}), and the upper bounds of Theorem~\\ref{thm: ub-RD-TV-FS} \nin the case of binary alphabets.\n\\begin{figure}[here!]\n\\centerline{\\includegraphics[width=9.8cm]{jagged.pdf}}\n\\caption{\\label{figure:compare_ub_RD_finite_graph1}\nThe R\\'{e}nyi divergence $D_{\\alpha}(P\\|Q)$ for $P$ and $Q$\nwhich are defined on a binary alphabet with $P(0)=Q(1)=0.65$,\ncompared to (a) its upper bound in \\eqref{eq: ub-RD-TV-FS},\nand (b) its upper bound in \\eqref{eq: f1} (see \\cite[Theorem~23]{ISSV15}).\nThe two bounds coincide here when $\\alpha \\in (1, 1.291) \\cup (2, \\infty)$.}\n\\end{figure}}\n\\end{example}\n\n\n\\section{Summary}\nWe derive in this paper some ``reverse Pinsker inequalities\" for probability measures\n$P \\ll Q$ defined on a common finite set, which provide lower bounds on the total\nvariation distance $P-Q$ as a function of the relative entropy $D(P\\|Q)$ under the\nassumption of a bounded relative information or $Q_{\\min} > 0$. More general results\nfor an arbitrary alphabet are available in \\cite[Section~5]{ISSV15}.\n\nIn \\cite{ISSV15}, we study bounds among various $f$-divergences, dealing with arbitrary\nalphabets and deriving bounds on the ratios of various distance measures.\nNew expressions of the R\\'{e}nyi divergence in terms of the relative information\nspectrum are derived, leading to upper and lower bounds on the R\\'{e}nyi divergence\nin terms of the variational distance.\n\n\\section*{Acknowledgment}\nThe work of I. Sason has been supported by the Israeli Science Foundation (ISF) under\nGrant 12/12, and the work of S. Verd\\'{u} has been supported by the US National Science\nFoundation under Grant CCF-1016625, and in part by the Center for Science of Information,\nan NSF Science and Technology Center under Grant CCF-0939370.\n\n\n", "meta": {"timestamp": "2015-10-20T02:09:11", "yymm": "1503", "arxiv_id": "1503.03417", "language": "en", "url": "https://arxiv.org/abs/1503.03417", "abstract": "A new upper bound on the relative entropy is derived as a function of the total variation distance for probability measures defined on a common finite alphabet. The bound improves a previously reported bound by Csiszár and Talata. It is further extended to an upper bound on the Rényi divergence of an arbitrary non-negative order (including $\\infty$) as a function of the total variation distance.", "subjects": "Information Theory (cs.IT); Probability (math.PR)", "title": "Upper Bounds on the Relative Entropy and Rényi Divergence as a Function of Total Variation Distance for Finite Alphabets", "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.985496423290417, "lm_q2_score": 0.8723473779969194, "lm_q1q2_score": 0.8596952208827375}} {"url": "https://arxiv.org/abs/1702.06271", "title": "Jacobson's lemma via Groebner-Shirshov bases", "abstract": "Let $R$ be a ring with identity $1$. Jacobson's lemma states that for any $a,b\\in R$, if $1-ab$ is invertible then so is $1-ba$. Jacobson's lemma has suitable analogues for several types of generalized inverses, e.g., Drazin inverse, generalized Drazin inverse, and inner inverse. In this note we give a constructive way via Groebner-Shirshov basis theory to obtain the inverse of $1-ab$ in terms of $(1-ba)^{-1}$, assuming the latter exists.", "text": "\\section{Introduction}\nLet $R$ be a ring with identity $1$.\nRecall that an element $x\\in R$ is \\emph{invertible (in $R$)} if there exists $y\\in R$ such that $xy=yx=1$.\n{Jacobson's lemma} \\cite{lam2014jacobson,corach2013extensions} states that for any $a,b\\in R$, if $1-ab$ is invertible then so is $1-ba$.\nJacobson's lemma has suitable analogues for several types of generalized inverses, e.g., Drazin inverse \\cite{castro2010generalized, cvetkovic2010jacobson},\ngeneralized Drazin inverse \\cite{zhuang2012jacobson}, and inner inverse \\cite{chen2010unit}.\n\nSuppose $a, b\\in R$ and $1-ab$ is invertible.\nIt is easy to verify that\n\\begin{align*}\n(1 + b (1 - ab)^{-1}a)(1-ba)=1-b(1-(1-ab)^{-1}+(1-ab)^{-1}ab)a=1-b0a=1\\\\\n(1-ba)(1 + b (1 - ab)^{-1}a)=1+b((1-ab)^{-1}-1-ab(1-ab)^{-1}a)a=1+b0a=1\n\\end{align*}\n and then Jacobson's lemma follows.\nBut, how is this formula constructed?\nAs Halmos \\cite{halmos1981does} observed, $(1-ba)^{-1}$ can be ``constructed'' by using the following ``geometric series trick'',\nwhich is ``usually ascribed to Jacobson'' \\cite{lam2014jacobson},\n\\begin{align}\\label{formula_Jacabson}\n (1-ba)^{-1}&=1 + ba + baba + bababa + \\cdots\\nonumber \\\\\n &=1 + b (1 + ab + abab + \\cdots) a\\nonumber \\\\\n &= 1 + b (1 - ab)^{-1}a.\n\\end{align}\nOnce the inverse is ``constructed'', it is easy to verify that Formula (\\ref{formula_Jacabson}) works as above, despite its unlawful derivation.\n\nNow it is natural to ask: Does there exist a legal way to construct the inverse of $1-ba$?\nWe shall answer this question in this note.\n\n\nIn this note we give a constructive way via Gr\\\"obner-Shirshov basis\\ theory to obtain the inverse of $1-ab$ in terms of $(1-ba)^{-1}$, assuming the latter exists.\nA similar method is used in \\cite{herfort2014grobner} to obtain Hua's identity.\n\n\\section{Gr\\\"obner-Shirshov bases\\ for algebras over a unitary ring}\nIn this section, we briefly review several properties of Gr\\\"obner-Shirshov basis\\ for algebras over a unitary commutative ring.\n\nLet $X\\neq\\emptyset$ and $k$ be a unitary ring.\nLet $X^*$ be the free monoid generated by $X$ and $A=k\\langle X\\rangle =kX^*$ be the free algebra over $k$ with generator set $X$.\nSimilar to the case of associative algebras over a field (see \\cite{bokut2014grobner}, c.f., \\cite{cox2005using,zhangzhao2013Modules}),\nwe have concepts of \\emph{monomial ordering},\n\\emph{leading term/coefficient}, \\emph{composition}, and \\emph{Gr\\\"obner-Shirshov basis}.\nParticularly, a Gr\\\"obner-Shirshov basis\\ consists of only \\emph{monic polynomials} (i.e., polynomials with leading coefficient $1$).\nFor convenience, we define the leading term of a nonzero element from $k$ to be $1$ and the leading term of $0$ to be $0$.\nFixing a monomial ordering, we denote the leading term and leading coefficient of $f\\in A$ by $\\overline{f}$ and $\\operatorname{lc}(f)$ respectively.\n\nGiven $f=\\alpha_1u_1+\\cdots\\alpha_nu_n\\in A$, where $n\\in \\mathbb{N}$, each $\\alpha_i\\in k$, $u_i\\in X^*$, and $\\alpha_i\\neq0$,\nthe \\emph{support} of $f$ is $\\operatorname{Supp} (f)=\\{u_1,\\ldots,u_n\\}$.\nLet $f,h, g\\in A$, where $g$ is a monic polynomial.\nThen \\emph{$f$ reduces to $h$ modulo $g$},\ndenoted by $f\\rightarrow_g h$, if $\\overline{f}=\\overline{agb}$ for some $a,b\\in X^*$,\nand $h=f-\\operatorname{lc}(f) agb$.\nFor a set $G$ consisting of monic polynomials,\nwe say that \\emph{$f$ reduces to $h$ modulo $G$},\ndenoted by $f\\rightarrow_G {h}$, if there exists a finite chain of reductions\n\\[\nf\\rightarrow_{g_1} f_1\\rightarrow_{g_2} f_2\\rightarrow_{g_3}\\cdots \\rightarrow_{g_t} f_t=h,\n\\]\nwhere each $g_i\\in G$ and $t\\in \\mathbb{N}$.\nDenote the set of words that are {\\it irreducible} w.r.t. $G$ by $\\operatorname{Irr}(G)$, i.e.,\n\\[\n\\operatorname{Irr}(G)=\\{w\\in X^*:w\\not=a\\overline{g}b \\mbox{ for any }g\\in G, a,b\\in X^*\\}.\n\\]\nIf {${\\operatorname{Supp} (h)}\\subseteq \\operatorname{Irr}(G)$}, then we say $h$ is \\emph{irreducible} w.r.t. $G$.\n\nThe following lemma follows from the composition-diamond lemma for associative algebras over a commutative ring (\\cite{bokut2014grobner}, Theorem 1 and Remark 2).\n\\begin{lemma}\\label{lemma-GSB-reduction}\n Suppose $G$ is a Gr\\\"obner-Shirshov basis\\ for $I\\unlhd A$, where $I$ is the ideal of $A$ generated by $G$.\n Then, for any $f\\in A$, $f\\in I$ if and only if\\ $f\\to_G0$.\n\\end{lemma}\n\\section{A constructive method to Jacobson's lemma}\n\n\nLet $R$ be a unitary ring.\nGiven $a,b\\in R$ with $1-ab$ invertible (say, $c=(1-ab)^{-1}$),\nwe give a constructive method in this section to find the inverse of $1-ba$.\n\nLet $S$ be the $\\mathbb{Z}$-algebra generated by $\\{a,b,c\\}$ with defining relations\n$\\{(1-ab)c=1, c(1-ab)=1\\}$, i.e.,\n\\(\nS=\n\\mathbb{Z}\\langle a,b,c: (1-ab)c=1, c(1-ab)=1\\rangle.\n\\)\nWe\nwant to solve the following system for $x$ in $S$:\n\\begin{align}\\label{system}\n (1-ba)x=1\\nonumber\\\\\n x(1-ba)=1.\n\\end{align}\nIt is obvious that the subring of $R$ generated by $\\{a,b,c\\}$ is a ring homomorphic image of $S$. Thus the image of $x$ is the inverse of $1-ba$ in $R$.\n\nLet $<$ be the degree-lexicographic ordering on $\\{a,b,c\\}^*$ with $av_2>\\cdots>v_n$.\n Then $uv_1\\not\\in \\operatorname{Irr}(G)$.\n\\end{lemma}\n\\begin{proof}\n If $uv_1\\in \\operatorname{Irr}(G)$, then $uv_1\\neq 0$ is the leading term of $\\sum_{i=1}^n\\alpha_iuv_i$,\n contradicting the assumption that $\\sum_{i=1}^n\\alpha_iuv_i\\to_G0$.\n\\end{proof}\n\nThe following theorem gives a new proof for Jacobson's lemma.\n\\begin{theorem}\\label{thm_main}\n System (\\ref{system}) has a unique solution in $S$.\n\\end{theorem}\n\\begin{proof}\n(Uniqueness) It follows from the uniqueness of an inverse in a ring.\n\n(Existence)\nSuppose a solution $x\\in S$ of the system exists and let $x=\\sum_{i=1}^m\\alpha_iu_i$ where $0\\neq\\alpha_i\\in\\mathbb{Z}$, $m\\in \\mathbb{N}$ and all $u_i\\in{\\operatorname{Irr}(G)}$ with $u_1>u_2>\\cdots>u_m$. Then it follows from $(1-ba)x=1$ and Lemma \\ref{lemma-GSB-reduction} that\n\\begin{align*}\n bax-x+1=\\sum_{i=1}^m \\alpha_ibau_i-\\sum_{i=1}^m\\alpha_iu_i+1\\to_G0.\n\\end{align*}\nIf $bau_1\\in\\operatorname{Irr}(G)$ then $bau_1\\neq0$ is the leading term of $bax-x+1$,\ncontradicting the fact that $bax-x+1\\to_G0$.\nThus $bau_1\\not\\in\\operatorname{Irr}(G)$.\nHence $u_1=bcv$ for some $v\\in \\{a,b,c\\}^*$,\nwhere $\\{a,b,c\\}^*$ denotes the free monoid generated by $\\{a,b,c\\}$.\nSimilarly, the assumption $1-x(1-ba)=0$ in $S$ implies that $u_1=wca$ for some $w\\in \\{a,b,c\\}^*$.\n\nNow there are two cases for $u_1$: (i) $u_1=bca$ and (ii) $u_1=bczca$ for some $z\\in \\{a,b,c\\}^*$.\nWe begin with Case (i).\nSuppose $u_1=bca$.\nFirst we notice that\n\\[\nba(\\alpha_1bca)-\\alpha_1bca+1\\to_f \\alpha_1b(c-1)a-\\alpha_1bca+1=-\\alpha_1ba+1,\n\\]\nwhere $-\\alpha_1ba+1$ is irreducible modulo $G$.\nThus $ba(\\alpha_1bca)-\\alpha_1bca+1\\not\\to_G 0$ and $x\\neq \\alpha_1 bca$.\nHence $m\\geq 2$.\nNow we may suppose $x=\\alpha_1bca+\\sum_{i=2}^m\\alpha_iu_i$.\nThen\n\\begin{align}\n bax-x+1&= \\alpha_1babca+\\sum_{i=2}^m\\alpha_ibau_i-\\alpha_1bca-\\sum_{i=2}^m\\alpha_iu_i+1\\nonumber\\\\\n &\\rightarrow_f \\alpha_1b(c-1)a+\\sum_{i=2}^m\\alpha_ibau_i-\\alpha_1bca-\\sum_{i=2}^m\\alpha_iu_i+1\\nonumber\\\\\n &= \\sum_{i=2}^m\\alpha_ibau_i -\\alpha_1ba -\\sum_{i=2}^m\\alpha_iu_i+1\\label{eqn_m}.\n\\end{align}\nWe claim that $m=2$.\nOtherwise, suppose $m\\geq 3$.\nThen $u_2>u_3\\geq1$, $bau_2>bau_i$, $bau_2>ba$,\nand $bau_2>u_j$ for all $i\\geq 3$ and $j\\geq 2$.\nNamely, $bau_2$ is the leading monomial of (\\ref{eqn_m}).\nThus the fact $bax-x+1\\to_G0$ implies $bau_2\\not\\in\\operatorname{Irr}(G)$.\nSo $u_2=bcu'$ for some $u'\\in \\{a,b,c\\}^*$.\nBut $u_2u_i$.\nTherefore $bau_i\\in\\operatorname{Irr}(G)$ for $i\\geq 3$.\nThus we may reduce polynomial (\\ref{eqn_m}) as follows\n\\begin{align*}\n&\\sum_{i=2}^m\\alpha_ibau_i -\\alpha_1ba -\\sum_{i=2}^m\\alpha_iu_i+1\\\\\n=&\\ \\alpha_2babc+\\sum_{i=3}^m\\alpha_ibau_i -\\alpha_1ba -\\alpha_2bc-\\sum_{i=3}^m\\alpha_iu_i+1\\\\\n\\to_f&\\ \\alpha_2b(c-1)+\\sum_{i=3}^m\\alpha_ibau_i -\\alpha_1ba -\\alpha_2bc-\\sum_{i=3}^m\\alpha_iu_i+1\\\\\n=&\\ -\\alpha_2b+\\sum_{i=3}^m\\alpha_ibau_i -\\alpha_1ba -\\sum_{i=3}^m\\alpha_iu_i+1.\n\\end{align*}\nThe last polynomial, denoted by $h$, is irreducible and its leading term is\n(note that if $u_3=1$ then $m=3$)\n\\[\n\\overline{h}=\\begin{cases}\n bau_3 &\\text{ if } u_3>1\\\\\n ba &\\text{ if } u_3=1 \\text{ and } \\alpha_3\\neq \\alpha_1\\\\\n b &\\text{ if } u_3=1 \\text{ and } \\alpha_3= \\alpha_1\n\\end{cases}.\n\\]\nIn each of these three cases, $\\overline{h}\\neq0$ and thus $h\\neq 0$,\nwhich contradicts the assumption that $bax-x+1\\rightarrow_G 0$.\nThis proves our claim that $m=2$.\n\nNow we may write $x=\\alpha_1bca+\\alpha_2u_2$ and from (\\ref{eqn_m}) we have\n\\begin{align*}\n bax-x+1\\to_f\\alpha_2bau_2-\\alpha_1ba-\\alpha_2u_2+1.\n\\end{align*}\nIf $bau_2\\not\\in \\operatorname{Irr}(G)$, then $u_2=bc$ as we proved in the last paragraph.\nThen we have\n\\begin{align*}\n bax-x+1&\\to_f\\alpha_2babc-\\alpha_1ba-\\alpha_2bc+1\\\\\n &\\to_f\\alpha_2b(c-1)-\\alpha_1ba-\\alpha_2bc+1\\\\\n&=-\\alpha_2b-\\alpha_1ba+1,\n\\end{align*}\nwhich is irreducible modulo $G$ and nonzero, contradicting that $bax-x+1\\to_G0$.\nTherefore, $bau_2\\in\\operatorname{Irr}(G)$.\nNow the only possibility in which $\\alpha_2bau_2-\\alpha_1ba-\\alpha_2u_2+1\\to_G0$ is that $u_2=1$ and\n$\\alpha_1=\\alpha_2=1$.\nThat is, $x=bca+1$.\n\n\nNow we check by reduction whether $x=bca+1$ is a solution of (\\ref{system}).\n\\begin{align*}\n bax-x+1&=ba(bca+1)-(bca+1)+1\\\\\n &=babca+ba-bca\\\\\n &\\to_f b(c-1)a+ba-bca\\\\\n &=0\\\\\n\\end{align*}\nSimilarly, we have\n\\begin{align*}\n xba-x+1&=(bca+1)ba-(bca+1)+1\\\\\n &=bcaba+ba-bca\\\\\n &\\to_g b(c-1)a+ba-bca\\\\\n &=0\\\\\n\\end{align*}\nThus, $x=bca+1$ is a solution of (\\ref{system}) in $S$.\nThis implies that $x$ is an inverse of $1-ba$ in $S$.\nSince an inverse is unique if it exists in $S$,\nwe do not need to consider Case (ii) for $x$.\n\\end{proof}\nIt is clear that Theorem \\ref{thm_main} gives a new proof for Jacobson's lemma.\n\n\\\n\n{\\bf Acknowledgements.} This work was motivated by a talk by R. Padmanabhan in the Rings and Modules Seminar at the University of Manitoba.\n", "meta": {"timestamp": "2017-02-22T02:02:45", "yymm": "1702", "arxiv_id": "1702.06271", "language": "en", "url": "https://arxiv.org/abs/1702.06271", "abstract": "Let $R$ be a ring with identity $1$. Jacobson's lemma states that for any $a,b\\in R$, if $1-ab$ is invertible then so is $1-ba$. Jacobson's lemma has suitable analogues for several types of generalized inverses, e.g., Drazin inverse, generalized Drazin inverse, and inner inverse. In this note we give a constructive way via Groebner-Shirshov basis theory to obtain the inverse of $1-ab$ in terms of $(1-ba)^{-1}$, assuming the latter exists.", "subjects": "Rings and Algebras (math.RA)", "title": "Jacobson's lemma via Groebner-Shirshov bases", "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.9865717440735736, "lm_q2_score": 0.868826784729373, "lm_q1q2_score": 0.8571599563082929}} {"url": "https://arxiv.org/abs/1906.12085", "title": "Tutorial: Complexity analysis of Singular Value Decomposition and its variants", "abstract": "We compared the regular Singular Value Decomposition (SVD), truncated SVD, Krylov method and Randomized PCA, in terms of time and space complexity. It is well-known that Krylov method and Randomized PCA only performs well when k << n, i.e. the number of eigenpair needed is far less than that of matrix size. We compared them for calculating all the eigenpairs. We also discussed the relationship between Principal Component Analysis and SVD.", "text": "\\section{Introduction}\nDimensionality reduction has always been a trendy topic in machine learning. Linear subspace method for reduction, e.g., Principal Component Analysis and its variation have been widely studied\\cite{Xu1995Robust, Partridge2002Robust, Zou2006Sparse}, and some pieces of literature introduce probability and randomness to realize PCA\\cite{Tipping2010Probabilistic, Halko2010Finding, Halko2011An, J1989Estimating, Woolfe2008A}. However, linear subspace is not applicable when the data lies in a non-linear manifold\\cite{Belkin2014Laplacian, Kai2010Clustered}. Due to the direct connection with PCA, Singular Value Decomposition (SVD) is one of the most well-known algorithms for low-rank approximation\\cite{Woodruff2014Low, Frieze1998Fast}, and it has been widely used throughout the machine learning and statistics community. \nSome implementations of SVD are solving least squares\\cite{Golub1970Singular, Zhang2010Regularized}, latent semantic analysis\\cite{deerwester1990indexing, hofmann2001unsupervised}, genetic analysis, matrix completion\\cite{Recht2012Exact, Cai2008A, Candes2010Matrix, Cand2010The}, data mining\\cite{Chong2016Feature, Belabbas2009On} etc. However, when it comes to a large scale matrix, the runtime of traditional SVD is intolerable and the memory usage could be enormously consuming. \n\n$\\textbf{Notations}$: we have a matrix $A$ with size $m\\times n$, usually $m \\gg n$. Our goal is to find the Principal Components (PCs) given the cumulative explained variance threshold $t$.\n\n$\\textbf{Assumptions}$: In this tutorial, every entry of matrix $A$ is real-valued; W.l.o.g., assume $m \\gg n$ and $A$ has zero mean over each feature. \n\nFor your information, either each column or row of $A$ could represent an example, and the definition will be specified when necessary.\n\n\nIn the traditional approach of PCA, we need to compute the covariance matrix $S$, then perform the eigen-decomposition on $S$. By selecting the top $K$ largest eigenvalues and corresponding eigenvectors, we get our Principal Components (PCs). Nevertheless, if each column of $A$ is an example and the row size of $A$ is tremendously large, saving even larger covariance matrix into memory is expensive, let alone the eigen-decomposition process.\n\n\\section{Preliminary knowledge}\\label{sec:pre}\n\\subsection{Singular Value Decomposition (SVD)}\nAny real or complex matrix can be approximated over the summation of a series of rank-1 matrix. In SVD, we have\n\\begin{align*}\nA = U\\Sigma V^T\n\\end{align*}\nwhere\n\\begin{align*}\nU &= [u_1 \\cdots u_n]\\in\\mathbb{R}^{m \\times n}\\\\\n\\Sigma &= diag(\\sigma_1 \\cdots \\sigma_n)\\in\\mathbb{R}^{n\\times n}\\\\\nV &= [v_1 \\cdots v_n]\\in\\mathbb{R}^{n \\times n}\n\\end{align*}\nHere $U$ and $V$ are orthogonal matrices, i.e.\n\\begin{align*}\nU^T U &= I_n\\\\\nUU^T &= I_m\\\\\nV^TV&=VV^T=I_n\n\\end{align*}\nWe could also rewrite SVD as following\n\\begin{align}\\label{AV}\nAv_i = \\sigma_i u_i, i = 1 \\cdots n \n\\end{align}\nGeometrically speaking, the matrix $A$ rotates the unit vector $v_i$ to $u_i$ and then stretches the Euclidean norm of $u_i$ with a factor of $\\sigma_i$.\n\nThe orthogonal matrices $U$ and $V$ can be obtained by eigen-decomposition of matrix $AA^T$ and $A^TA$, and the singular values $\\sigma_i's$ are the square root of the eigenvalues of $AA^T$ or $AA^T$.\n\nProof:\n\\begin{align} \nAA^T &= U\\Sigma V^T V\\Sigma^T U^T = U\\Sigma^2 U^T \\label{AA}\\\\\nA^TA &= V\\Sigma^T U^T U\\Sigma V^T = V\\Sigma^2 V^T \\label{AB}\n\\end{align}\n\nSet $\\Lambda = \\Sigma^2= diag(\\sigma_1^2 \\cdots \\sigma_n^2)$, i.e. \n\n\\begin{align}\\label{LambdaSigma}\n\\lambda_i = \\sigma_i^2\n\\end{align}\n\nWe could rewrite Eq. (\\ref{AA}) and Eq. (\\ref{AB}) as\n\n\\begin{align} \nAA^Tu_i &= \\lambda_i u_i \\\\\nA^TAv_i &= \\lambda_i v_i\n\\end{align}\n\nTherefore, the column vectors of $U$ and $V$ are the eigenvectors (with the unit norm) of $AA^T$ and $A^T A$, respectively. Moreover, the eigenvalues $\\lambda_i$ are square of singular values $\\sigma_i$, as in Eq. (\\ref{LambdaSigma}). In other words, the square root of eigenvalues are singular values. \n\\subsection{Relations with PCA}\n\\textbf{If each column of $A$ represents an example or data point}, set $Y=U^TA$ to our transformed data points, where $U$ is the left singular matrix in SVD. The covariance matrix of $Y$ is\n\\begin{align}\ncov(Y) = YY^T = U^TAA^TU = U^TU\\Sigma^2U^TU = \\Sigma^2\n\\end{align}\n\n\\textbf{If each row of $A$ represents an example or data point}, set $Y=AV$ to our transformed data points, where $V$ is the right singular matrix in SVD. The covariance matrix of $Y$ is\n\\begin{align}\ncov(Y) = Y^TY = V^TA^TAV = V^TV\\Sigma^2V^TV = \\Sigma^2\n\\end{align}\nIt means the transformed data $Y$ are uncorrelated. Therefore, the column vectors of orthogonal matrix $U$ in SVD are the projection bases for Principal Components.\n\n\n\n\\subsection{Truncated SVD}\nAlthough the derivation of SVD is clear theoretically, practically speaking, however, it is unwise to do eigen-decomposition on matrix $AA^T$, as it has a tremendous size of $m \\times m$, which will deplete memory and cost a great amount of time. On the contrary, the matrix $A^TA$ only has a size of $n \\times n$, thus it is plausible that we compute orthogonal matrix $V$ first.\nThen we can plug the Eq. (\\ref{AV}) into Eq. (\\ref{LambdaSigma}) and get\n\\begin{align}\nu_i = \\dfrac{Av_i}{\\sqrt{\\lambda_i}}, i = 1 \\cdots n\n\\end{align}\n\nThis equation is the key to improving time and space efficiency because we do not perform eigen-decomposition on huge matrix $AA^T\\in\\mathbb{R}^{m\\times m}$, which takes $\\mathcal{O}(m^3)$ time and $\\mathcal{O}(m^2)$ space.\n\nThen we column-wisely combine $u_i$ to get $U$, and the same for $v_i$ to form $V$. For $\\Sigma$, it is $diag(\\sqrt{\\sigma_1} \\cdots \\sqrt{\\sigma_n})$.\n\n\\subsection{PCA Evaluation}\nIn Section \\ref{sec:pre}, we have proved that the column vectors of orthogonal matrix $U$ or $V$ in SVD are the projection bases for PCA. In the literature of PCA, there are many criteria for evaluating the residual error, e.g., Frobenius norm and induced $L_2$ norm of the difference matrix (original matrix minus approximated matrix), explained variance and cumulative explained variance. \n\nUsually, we use the cumulative explained variance criterion for evaluation.\n\n\\textbf{cumulative explained variance criterion:} Given the threshold $t$, find the minimal integer $K$ such that\n\\begin{align}\n\\frac{\\Sigma_{i=1}^{K} \\lambda_i }{\\Sigma_{i=1}^n \\lambda_i}\\geq t\n\\end{align}\nwhere each $\\lambda_i$ is the eigenvalue of matrix $A^TA$ or $AA^T$. Every $\\lambda_i=\\sigma_i^2$ indicates the variance in principal axis, this is why the criterion is named cumulative explained variance.\n\nFor those SVD or PCA algorithms who do not obtain all the eigenvalues or can not get accurate them accurately, it seems that the denominator term $\\Sigma_{i=1}^n \\lambda_i$ can not be calculated. Actually, the sum of all eigenvalues can be done by\n\\begin{align}\\label{sumeigen}\n\\Sigma_{i=1}^n \\lambda_i = tr(A^TA) = tr(AA^T) = \\|A\\|_F^2 = \\Sigma_{i, j} a_{ij}^2\n\\end{align}\n\\textbf{Therefore, we do not need to implement eigen-decomposition on either large matrix $AA^T\\in\\mathbb{R}^{m\\times m}$ or small matrix $A^TA\\in\\mathbb{R}^{n\\times n}$.} Eq. (\\ref{sumeigen}) saves us $\\mathcal{O}(n^3)$ time and $\\mathcal{O}(n^2)$ space.\n\n\\section{Complexity Analysis}\nIn this section, we compare the time complexity and space complexity of Krylov method, Randomized PCA and truncated SVD. Due to the copyrights issue, the mechanisms of MatLab and Python built-in econSVD are not available, whose complexity analysis will not be conducted.\n\nTo restate again, our matrix $A$ has size $m\\times n$ and $m>n$.\n\n\n\\subsection{Time Complexity}\nFor time complexity, we use the number of FLoating-point OPerations (FLOP) as a quantification metric. \n\nFor matrix $A \\in \\mathbb{R}^{m\\times n}$ and $B \\in \\mathbb{R}^{n\\times l}$, the time complexity of matrix multiplication $AB$ takes $mnl$ FLOP of products and $ml(n-1)$ FLOP of summations. Therefore, the multiplication of two matrices takes $\\mathcal{O}(mnl+ml(n-1))=\\mathcal{O}(2mnl-ml) = \\mathcal{O}(mnl)$ FLOP. We could ignore the coefficient here for it will not bring bias to our analysis.\n\nFor your information, the coefficient of time complexity in Big-O notation will not be ignored when comparing different SVD algorithms as it is of importance in our analysis.\n\\subsubsection{Krylov method}\nFor the Krylov method, we discuss the time complexity of each step.\n\\begin{enumerate}\n\t\\item Forming standard normal distribution matrix $G$ of size $n \\times l$ takes $\\mathcal{O}(nl)$ FLOP, where in practice $l=0.5n$.\n\t\\item Forming matrix $H^{(0)}=AG \\in \\mathbb{R}^{m\\times l}$ takes $\\mathcal{O}(mnl)$ FLOP.\n\t\\item Forming matrix $H^{(i)}=A(A^T H^{(i-1)})\\in \\mathbb{R}^{m\\times l}$ takes $\\mathcal{O}(2imnl)$ FLOP, as the matrix multiplication in bracket $A^T H^{(i-1)}$ takes $\\mathcal{O}(nml)$ FLOP, then multiplying by $A$ takes $\\mathcal{O}(mnl)$ FLOP. In total, it takes $\\mathcal{O}{(2imnl)}$ FLOP to generate $i$ matrices. \n\t\\item Forming matrix $H=\\left(H^{(0)}\\left|H^{(1)}\\right| \\ldots\\left|H^{(i-1)}\\right| H^{(i)}\\right) \\in \\mathbb{R}^{m\\times (i+1)l}$ by concatenating each $H^{(i)}$ takes $\\mathcal{O}(1)$ FLOP.\n\t\\item Performing QR decomposition on $H$ takes $\\mathcal{O}(2m[(i+1)l]^2 - 2[(i+1)l]^3/3)$ FLOP.\n\t\\item Forming $T = A^T Q \\in \\mathbb{R}^{n\\times (i+1)l}$ takes $\\mathcal{O}((i+1)mnl)$ FLOP.\n\t\\item Performing SVD on $T=\\tilde{V}\\tilde{\\Sigma}W^T$ takes $\\mathcal{O}(n[(i+1)l]^2)$ FLOP.\n\t\\item Forming $\\tilde{U}=QW$ takes $\\mathcal{O}(m[(i+1)l]^2)$ FLOP.\n\\end{enumerate}\n\nIn total, the time complexity of Krylov method is \n\\begin{align}\n\\mathcal{O}(nl+(3i+2)mnl+(i+1)^2 l^2 (m^2+n^2+2m-\\dfrac{2}{3}(i+1)l)) \\text{ FLOP}\n\\end{align}\n\nIn practice, $l = 0.5n, i=1$, then the time complexity will be\n\\begin{align}\n\\mathcal{O}(\\dfrac{n^2}{2}+\\dfrac{5}{2}mn^2+4 (m^2 n^2 +2mn^2+n^4 -\\dfrac{2}{3}n^3)) \\text{ FLOP}\n\\end{align}\n\n\\subsubsection{Randomized PCA}\nFor Randomized PCA, we discuss the time complexity of each step. It is very similar to the Krylov method.\n\\begin{enumerate}\n\t\\item Forming standard normal distribution matrix $G$ of size $n \\times l$ takes $\\mathcal{O}(nl)$ FLOP, where in practice $l=0.5n$.\n\t\\item Forming matrix $H^{(0)}=AG \\in \\mathbb{R}^{m\\times l}$ takes $\\mathcal{O}(mnl)$ FLOP.\n\t\\item Forming matrix $H^{(i)}=A(A^T H^{(i-1)})\\in \\mathbb{R}^{m\\times l}$ takes $\\mathcal{O}(2imnl)$ FLOP, as the matrix multiplication in bracket $A^T H^{(i-1)}$ takes $\\mathcal{O}(nml)$ FLOP, then multiplying by $A$ takes $\\mathcal{O}(mnl)$ FLOP. In total, it takes $\\mathcal{O}{(2imnl)}$ FLOP. \n\t\\item Performing QR decomposition on $H$ takes $\\mathcal{O}(2ml^2 - 2l^3/3)$ FLOP.\n\t\\item Forming $T = A^T Q \\in \\mathbb{R}^{n\\times l}$ takes $\\mathcal{O}(mnl)$ FLOP.\n\t\\item Performing SVD on $T=\\tilde{V}\\tilde{\\Sigma}W^T$ takes $\\mathcal{O}(nl^2)$ FLOP.\n\t\\item Forming $\\tilde{U}=QW$ takes $\\mathcal{O}(ml^2)$ FLOP.\n\\end{enumerate}\n\nIn total, the time complexity of Randomized PCA is \n\\begin{align}\n\\mathcal{O}(nl+(2i+2)mnl+l^2 (3m+n-\\dfrac{2}{3}l)) \\text{ FLOP}\n\\end{align}\n\nIn practice, $l = 0.5n, i = 1$, then the time complexity will be\n\\begin{align}\n\\mathcal{O}(\\dfrac{n^2}{2}+\\dfrac{11}{4}mn^2+\\dfrac{n^3}{6}) \\text{ FLOP}\n\\end{align}\n\n\\subsubsection{Truncated SVD}\nWe discuss the time complexity of Truncated SVD for each step.\n\\begin{enumerate}\n\t\\item Forming matrix $A^T A\\in\\mathbb{R}^{n\\times n}$ takes $\\mathcal{O}(mn^2)$ FLOP.\n\t\\item Performing eigen-decomposition on $A^TA \\in\\mathbb{R}^{n\\times n}$ takes $\\mathcal{O}(n^3)$ FLOP.\n\t\\item Taking the square root of each eigenvalue of $A^T A$ takes $\\mathcal{O}(n)$ FLOP.\n\t\\item Forming $u_i = \\dfrac{Av_i}{\\sigma_i}$ takes $\\mathcal{O}(n(mn+m))$ FLOP, as $Av_i$ takes $\\mathcal{O}(mn)$ FLOP while divided by $\\sigma_i$ takes $\\mathcal{O}(m)$ FLOP. In total, we have $n$ equations like this, thus it takes $\\mathcal{O}(n(mn+m))$ FLOP.\n\\end{enumerate}\n\nIn total, the time complexity of Truncated SVD is\n\\begin{align}\n\\mathcal{O}(2mn^2+n^3+n+mn)\n\\end{align}\n\n\n\\subsection{Space Complexity}\nWe evaluate the space complexity by the number of matrix entries. For a matrix $A\\in\\mathbb{R}^{m\\times n}$, its space complexity is $\\mathcal{O}(mn)$. In MatLab or Python programming language, each entry takes 8 bytes memory.\n\\subsubsection{Krylov method}\n\\begin{enumerate}\n\t\\item Forming standard normal distribution matrix $G$ of size $n \\times l$ takes $\\mathcal{O}(nl)$, where in practice $l=0.5n$.\n\t\\item Forming matrix $H^{(0)}=AG \\in \\mathbb{R}^{m\\times l}$ takes $\\mathcal{O}(ml)$.\n\t\\item Forming matrix $H^{(i)}=A(A^T H^{(i-1)})\\in \\mathbb{R}^{m\\times l}$ takes $\\mathcal{O}(ml)$. \n\t\\item Forming matrix $H=\\left(H^{(0)}\\left|H^{(1)}\\right| \\ldots\\left|H^{(i-1)}\\right| H^{(i)}\\right) \\in \\mathbb{R}^{m\\times (i+1)l}$ by concatenating each $H^{(i)}$ takes $\\mathcal{O}((i+1)ml)$.\n\t\\item Performing QR decomposition on $H$ takes $\\mathcal{O}((i+1)ml)$. Note that we discard matrix $R$, only $Q$ is saved.\n\t\\item Forming $T = A^T Q \\in \\mathbb{R}^{n\\times (i+1)l}$ takes $\\mathcal{O}((i+1)nl)$.\n\t\\item Performing SVD on $T=\\tilde{V}\\tilde{\\Sigma}W^T$ takes $\\mathcal{O}((i+1)nl + 2[(i+1)l]^2)$, for matrix $\\tilde{V}\\in\\mathbb{R}^{n\\times (i+1)l}$ takes $\\mathcal{O}((i+1)nl)$ and $\\tilde{\\Sigma}\\in\\mathbb{R}^{(i+1)l\\times (i+1)l}$ takes $\\mathcal{O}([(i+1)l]^2)$, and $W\\in\\mathbb{R}^{(i+1)l\\times (i+1)l}$ takes $\\mathcal{O}([(i+1)l]^2)$. In total, it takes $\\mathcal{O}((i+1)nl + 2[(i+1)l]^2)$.\n\t\\item Forming $\\tilde{U}=QW$ takes $\\mathcal{O}((i+1)ml)$.\n\\end{enumerate}\nIn total with $A$ taking $\\mathcal{O}(mn)$, the space complexity of Krylov method is\n\\begin{align}\n\\mathcal{O}(mn+(3i+4)ml + (2i+3)nl + 2[(i+1)l]^2)\n\\end{align}\n\nIn practice, $l = 0.5n, i = 1$, then the space complexity of Krylov method will be\n\\begin{align}\n\\mathcal{O}(\\dfrac{9}{2}mn + \\dfrac{9}{2}n^2)\n\\end{align}\n\n\n\n\n\\subsubsection{Randomized PCA}\n\\begin{enumerate}\n\t\\item Forming standard normal distribution matrix $G$ of size $n \\times l$ takes $\\mathcal{O}(nl)$, where in practice $l=0.5n$.\n\t\\item Forming matrix $H^{(0)}=AG \\in \\mathbb{R}^{m\\times l}$ takes $\\mathcal{O}(ml)$.\n\t\\item Forming matrix $H^{(i)}=A(A^T H^{(i-1)})\\in \\mathbb{R}^{m\\times l}$ takes $\\mathcal{O}(ml)$.\n\t\\item Performing QR decomposition on $H$ takes $\\mathcal{O}(ml)$. Note that we discard matrix $R$, only $Q$ is saved.\n\t\\item Forming $T = A^T Q \\in \\mathbb{R}^{n\\times l}$ takes $\\mathcal{O}(nl)$.\n\t\\item Performing SVD on $T=\\tilde{V}\\tilde{\\Sigma}W^T$ takes $\\mathcal{O}(nl + 2l^2)$, for matrix $\\tilde{V}\\in\\mathbb{R}^{n\\times l}$ takes $\\mathcal{O}(nl)$ and $\\tilde{\\Sigma}\\in\\mathbb{R}^{l\\times l}$ takes $\\mathcal{O}(l^2)$, and $W\\in\\mathbb{R}^{l\\times l}$ takes $\\mathcal{O}(l^2)$. In total, it takes $\\mathcal{O}(nl + 2l^2)$.\n\t\\item Forming $\\tilde{U}=QW$ takes $\\mathcal{O}(ml)$.\n\\end{enumerate}\nIn total with $A$ taking $\\mathcal{O}(mn)$, the space complexity of Randomized PCA is\n\\begin{align}\n\\mathcal{O}(mn+3ml+3nl+2l^2)\n\\end{align}\n\nIn practice, $l = 0.5n, i = 1$, then the space complexity of Randomized PCA will be\n\\begin{align}\n\\mathcal{O}(\\dfrac{9}{2}mn + 3n^2)\n\\end{align}\n\n\\subsubsection{Truncated SVD}\nWe discuss the space complexity of Truncated SVD for each step.\n\\begin{enumerate}\n\t\\item Forming matrix $A^T A\\in\\mathbb{R}^{n\\times n}$ takes $\\mathcal{O}(n^2)$.\n\t\\item Performing eigen-decomposition on $A^TA \\in\\mathbb{R}^{n\\times n}$ takes $\\mathcal{O}(n^2+n)$.\n\t\\item Taking the square root of each eigenvalue of $A^T A$ takes $\\mathcal{O}(n)$.\n\t\\item Forming $u_i = \\dfrac{Av_i}{\\sigma_i}$ takes $\\mathcal{O}(mn)$, as each $u_i$ takes $\\mathcal{O}(m)$ and we have $n$ equations like this, thus in total it takes $\\mathcal{O}(mn)$.\n\t\\item Forming $V$ takes $\\mathcal{O}(n^2)$.\n\t\\item Storing $n$ singular values takes $\\mathcal{O}(n)$.\n\\end{enumerate}\n\nIn total with $A$ taking $\\mathcal{O}(mn)$, the space complexity of Truncated SVD is\n\\begin{align}\n\\mathcal{O}(3n^2+3n+2mn)\n\\end{align}\n\n\n\\subsection{Summary of Complexity Analysis}\n\n\\begin{table}[h]\n\t\\caption{Comparison of complexity}\n\t\\label{table-complexity}\n\t\\centering\n\t\\begin{tabular}{lll}\n\t\t\\toprule\n\t\n\t\n\t\tMethod & Time complexity & Space complexity \\\\\n\t\t\\midrule\n\t\tKrylov method & \n\t\t$\\mathcal{O}(\\dfrac{n^2}{2}+\\dfrac{5}{2}mn^2+4 (m^2 n^2 +2mn^2+n^4 -\\dfrac{2}{3}n^3))$ & $\\mathcal{O}(\\dfrac{9}{2}mn + \\dfrac{9}{2}n^2)$ \\\\\n\t\t& & \\\\\n\t\tRandomized PCA & $\\mathcal{O}(\\dfrac{n^2}{2}+\\dfrac{11}{4}mn^2+\\dfrac{n^3}{6})$ & $\\mathcal{O}(\\dfrac{9}{2}mn + 3n^2)$ \\\\\n\t\t& & \\\\\n\t\tTruncated SVD & $\\mathcal{O}(2mn^2+n^3+n+mn)$\t\t\t\t\t\t& $\\mathcal{O}(3n^2+3n+2mn)$ \\\\\n\t\t\\bottomrule\n\t\\end{tabular}\n\\end{table}\n\nWe summarized the time complexity and space complexity in Table \\ref{table-complexity}.\n\nUnder the assumptions that $m \\gg n$, for time complexity, by keeping the highest order term and its coefficient, we could see that for Krylov method, it takes $\\mathcal{O}(4m^2n^2)$ FLOP while Randomized PCA takes $\\mathcal{O}(\\dfrac{11}{4}mn^2)$ FLOP and Truncated SVD only takes $\\mathcal{O}(2mn^2)$ FLOP. Therefore, Truncated SVD is the fastest SVD algorithm among the aforementioned. Furthermore, Truncated SVD keeps all the eigenpairs rather than only first $k$ pairs as Krylov method and Randomized PCA do. \n\nFor space complexity, we could see that Truncated SVD needs the least memory usage as $\\mathcal{O}(3n^2+3n+2mn)$ and Krylov method needs the most memory space as $\\mathcal{O}(\\dfrac{9}{2}mn + \\dfrac{9}{2}n^2)$. Randomized PCA holds the space complexity in between.\n\n\n\n\\section{Experiments}\nWe generate a matrix $A$ whose entry obeys standard normal distribution, i.e., $a_{ij} \\sim \\mathcal{N}(0,1)$, with 5 row sizes in list $[2000, 4000, 6000, 8000, 10000]$ and 12 column sizes in $[100, 200, 300, 400, 500, 600, 800, 900, 1000, 1200, 1500, 2000]$, 60 matrices of $A$ in total. The experiment is repeated 10 times to get an average runtime. On evaluating the residual error, the rate of Frobenius norm is used\n\\begin{align}\n\\delta = \\dfrac{\\|A-U\\Sigma V^T \\|_F}{\\|A\\|_F}\n\\end{align}\n\nIn Fig. \\ref{runtimefixcol}, we compare the runtime of 4 SVD methods: Truncated SVD (FameSVD), Krylov method, Randomized PCA, econ SVD (MatLab built-in economic SVD). The matrix column size is fixed at 2000, and we increase the row size gradually. We could observe that all 4 methods follow a linear runtime pattern when row size increases. Of these 4 methods, Truncated SVD method outperforms the other 3 approaches.\n\n\\graphicspath{{E:/LXC/Graduate2Spring/FastSVD/experiment/MatLab/Fig/}}\n\\begin{figure}\n\t\\centering\n\t\\includegraphics[width=0.65\\linewidth]{runtime_columnSize_2000.png}\n\n\t\\caption{Runtime comparisons between 4 methods of SVD. The matrix column size is fixed at 2000, but row size varies. Truncated SVD method outperforms the rest.}\n\t\\label{runtimefixcol}\n\\end{figure}\n\n\\begin{figure}[hbt]\n\t\\centering\n\t\\includegraphics[width=0.65\\linewidth]{runtime_rowSize_10000.png}\n\n\t\\caption{Runtime comparisons between 4 methods of SVD. The matrix row size is fixed at 10000, but column size varies. Truncated SVD method outperforms the rest.}\n\t\\label{runtimefixrow}\n\\end{figure}\nIn Fig. \\ref{runtimefixrow}, we fix the row size of matrix at 10000, and we increase the column size gradually. We could observe that all 4 methods behave as non-linear runtime pattern when row size increases. Out of all 4 methods, truncated SVD method takes the least runtime in every scenario.\n\n\\begin{figure}[!ht]\n\t\\centering\n\t\\includegraphics[width=0.65\\linewidth]{memory_rowSize_10000.png}\n\n\t\\caption{Memory usage between 3 methods of SVD. The matrix row size is fixed at 10000, but column size varies. Truncated SVD needs the least memory.}\n\t\\label{memoryusagefig}\n\\end{figure}\n\nIn Fig. \\ref{memoryusagefig}, the row size of matrix is fixed at 10000, but column size varies. We could see that truncated SVD uses the minimal amount of memory while Randomized PCA needs the most.\n\n\\begin{figure}[h]\n\t\\centering\n\t\\includegraphics[width=0.8\\linewidth]{fourinone2.png}\n\n\t\\caption{Left to right: Original image, Randomized PCA, Krylov method and truncated SVD. The first 392 (784/2) principal components are reserved.}\n\t\\label{mnistfig}\n\\end{figure}\n\nWe also evaluate our algorithm on handwritten digit dataset MNIST\\cite{Deng2012The}. We form our matrix $A$ with size $60000\\times 784$ by concatenating $60000$ of vectorized $28\\times 28$ intensity image. For runtime, it takes 4.54$s$ and 10.79$s$ for Randomized PCA and Krylov method respectively to obtain the first 392 (784/2) principal components. However, it only takes Truncated SVD \\textbf{3.12s} to get all the 784 eigenvalues and eigenvectors; For memory usage, 1629.1MB for Randomized PCA and 1636.1MB for Krylov method. In the meanwhile, only \\textbf{731.9MB} is used for truncated SVD. \n\nOur experiments are conducted on MatLab R2013a and Python 3.7 with NumPy 1.15.4, with Intel(R) Core(TM) i7-6700 CPU @ 3.40GHz 3.40GHz, 8.00GB RAM, and Windows 7. The truncated SVD is faster than the built-in economic SVD of both MatLab and NumPy.\n\n\n\n\\section{Conclusion}\nthe regular SVD performs the worst because it needs the most of memory usage and time. When all eigenpairs are needed ($k = n$), truncated SVD outperforms other SVD variants mentioned in this tutorial. The memory usage of these SVD methods grows linearly with matrix column size, but truncated SVD has the lowest growth rate. The runtime of truncated SVD grows sublinearly while Krylov method grows exponentially.\n\n\n\n{\\small\n\t\\bibliographystyle{ieee}\n\t", "meta": {"timestamp": "2019-10-15T02:24:05", "yymm": "1906", "arxiv_id": "1906.12085", "language": "en", "url": "https://arxiv.org/abs/1906.12085", "abstract": "We compared the regular Singular Value Decomposition (SVD), truncated SVD, Krylov method and Randomized PCA, in terms of time and space complexity. It is well-known that Krylov method and Randomized PCA only performs well when k << n, i.e. the number of eigenpair needed is far less than that of matrix size. We compared them for calculating all the eigenpairs. We also discussed the relationship between Principal Component Analysis and SVD.", "subjects": "Numerical Analysis (math.NA); Machine Learning (cs.LG); Machine Learning (stat.ML)", "title": "Tutorial: Complexity analysis of Singular Value Decomposition and its variants", "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.9799765546169712, "lm_q2_score": 0.8740772433654401, "lm_q1q2_score": 0.8565752054223639}} {"url": "https://arxiv.org/abs/1210.6582", "title": "Minimal Periods for Ordinary Differential Equations in Strictly Convex Banach Spaces and Explicit Bounds for some l^p-Spaces", "abstract": "Let x(t) be a non-constant T-periodic solution to the ordinary differential equation x'= f(x) in a Banach space X where f is assumed to be Lipschitz continuous with constant L. Then there exists a constant c such that T L >= c, with c only depending on X. It is known that c >= 6 in any Banach space and that c = 2{\\pi} in any Hilbert space, but whereas the bound of c = 2 pi is sharp in any Hilbert space, there exists only one known example of a Banach space such that c = 6 is optimal. In this paper, we show that the inequality is in fact strict in any strictly convex Banach space. Moreover, we improve the lower bound for l^p(R^n) and L^p(M, {\\mu}) for a range of p close to p = 2 by using a form of Wirtinger's inequality for functions in W^{1,p}([0, T ], L^p(M, {\\mu})).", "text": "\\section{Introduction}\nConsider the ordinary differential equation $\\dot{x}=f(x)$ in a Banach space $X$, where $f$ is Lipschitz continuous with constant $L$, that is for any $x,y\\in X$\n$$\n\\|f(x)-f(y)\\|_X\\leq L\\|x-y\\|_X.\n$$\nIn this case one can relate the period $T$ of any non-constant periodic orbit to the Lipschitz constant $L$ via the inequality $TL\\ge c$.\nIn 1969, Yorke \\cite{yor} proved that $c=2\\pi$ when $X=\\mathbb{R}^n$ with its usual norm. Lasota \\& Yorke \\cite{lasyor} showed that the proof extends to arbitrary Hilbert spaces and they\nproved the bound $c = 4$ for any Banach space. This was improved to $c=4.5$ by Busenberg \\& Martelli \\cite{busmar} and finally to $c=6$ by Busenberg, Fisher \\& Martelli \\cite{busfishmar}\nwho also gave another proof for $c=2\\pi$ in any Hilbert space using Wirtinger's inequality. An obvious extension of the simple two-dimensional example\n$$\n\\dot x=Ly\\qquad\\dot y=-Lx\n$$\nshows that $c=2\\pi$ is sharp in any Hilbert space. Busenberg, Fisher \\& Martelli \\cite{busfishmar2} also constructed an example of an ODE on a periodic orbit of period $1$, which when viewed as a subset of\n$L^1([0,1]^2)$ has Lipschitz constant $L=6$, showing that $c=6$ is sharp for general Banach spaces.\\\\\n\nHowever, some interesting questions about minimal periods remain unanswered. Does there exist an ODE in a finite-dimensional Banach space such\nthat the lower bound of $TL=6$ is obtained? Does $TL\\geq 2\\pi$ characterise Hilbert spaces? Is there a (non-Hilbert) Banach space for which $c> 6$? \n\nThe results in this paper address this last question. First we show that in strictly convex Banach spaces necessarily $TL>6$. For these normed topological vector spaces the unit ball is a strictly\nconvex set. It is easy to see that the unit balls in $\\ell^1$ and $\\ell^\\infty$ contain a line segment and are therefore not strictly convex\nsets whereas the unit balls for all $1 6.$$\n\\end{thm}\n\nIn fact the proof of this statement is a refinement of an integral inequality originally introduced by Busenberg, Martelli \\& Fisher \\cite{busfishmar}. The revised\nversion of the result is summarised in the following lemma.\n\\begin{lem}\n\\label{BanachLemma}\nLet $X$ be a normed space and $y:\\mathbb{R}\\rightarrow X$ be a continuous, $T$-periodic function such that $ \\left\\Vert\\dot{y}(t)\\right\\Vert$ is integrable. Then $$\\int_0^T \\int_0^T \\left\\Vert y(t)-y(s) \\right\\Vert {\\rm d} s\\; {\\rm d} t \\leq \\frac{T}{6} \\int_0^T \\int_0^T \\left\\Vert\\dot{y}(t)-\\dot{y}(s)\\right\\Vert {\\rm d} s \\;{\\rm d} t.$$\nIf $X$ is a strictly convex Banach space, then the above inequality is in fact strict.\n\\end{lem}\n\nBefore we go into details of the proof, we show how Busenberg, Fisher \\& Martelli used it to establish $TL\\geq 6$ for any Banach space.\n\n\\begin{proof}[Proof of Theorem \\ref{main result}]\nApplying Lemma \\ref{BanachLemma} and using the Lipschitz continuity of $f$, it follows that\n\\begin{eqnarray*}\n\\int_0^T\\int_0^T || x(t)- x(s)||{\\rm d} s\\;{\\rm d} t &\\leq & \\frac{T}{6} \\int_0^T\\int_0^T ||\\dot{ x}(t)-\\dot{ x}(s)||{\\rm d} s\\;{\\rm d} t\\\\\n&=& \\frac{T}{6}\\int_0^T\\int_0^T ||f( x(t))-f( x(s))||{\\rm d} s\\;{\\rm d} t\\\\ &\\leq & \\frac{LT}{6}\\int_0^T \\int_0^T || x(t)- x(s)||{\\rm d} s\\;{\\rm d} t.\n\\end{eqnarray*}\nDividing both sides of the inequality by $\\int_0^T\\int_0^T || x(t)- x(s)||{\\rm d} s\\;{\\rm d} t$ yields the result.\n\\end{proof}\n\nWe now turn to the main proof of this section.\n\n\\begin{proof}[Proof of Lemma \\ref{BanachLemma}]\\label{Integral Lemma Martelli}\nWe know that $y$ is periodic with period $T$. Hence its integral over one period is shift invariant and thus $$\\int_0^T \\int_0^T \\left\\Vert y(t+s)-y(s)\\right\\Vert {\\rm d} s{\\rm d} t=\\int_0^T \\int_0^T \\left\\Vert y(t)-y(s)\\right\\Vert {\\rm d} s{\\rm d} t.$$\nUsing the above observation, we can derive the following integral expression\n\n\\begin{eqnarray}\\label{inequality martelli}\n\\int_0^T \\hspace{-0.2cm} \\int_0^T \\hspace{-0.1cm}\\left\\Vert y(t)-y(s)\\right\\Vert {\\rm d} s\\;{\\rm d} t \\hspace{-0.2cm} \\nonumber\n&=&\\hspace{-0.3cm}\\int_0^T \\hspace{-0.2cm} \\int_0^T \\hspace{-0.05cm}\\left\\Vert y(t+s)-y(s)\\right\\Vert {\\rm d} s\\; {\\rm d} t\\nonumber\\\\\n&=&\\hspace{-0.3cm}\\int_0^T \\hspace{-0.2cm}\\int_0^T \\hspace{-0.05cm}\\frac{(T-t)t}{T} \\left\\Vert\\frac{y(t+s)-y(s)}{t}-\\frac{y(s)-y(s+t-T)}{T-t}\\right\\Vert {\\rm d} s\\; {\\rm d} t\\nonumber\\\\\n&=&\\hspace{-0.3cm}\\int_0^T \\hspace{-0.2cm}\\int_0^T \\frac{(T-t)t}{T^2}\\left\\Vert \\int_0^{T} \\dot{y}\\left(s+\\frac{tr}{T}\\right)- \\dot{y}\\left(s+\\frac{tr}{T}-r\\right){\\rm d} r\\right\\Vert {\\rm d} s\\; {\\rm d} t\\nonumber\\\\\n&\\leq &\\hspace{-0.3cm}\\int_0^T \\hspace{-0.2cm} \\int_0^T \\frac{(T-t)t}{T^2}\\int_0^T \\left\\Vert \\dot{y}\\left(s+\\frac{tr}{T}\\right)-\\dot{y}\\left(s+\\frac{tr}{T}-r\\right)\\right\\Vert {\\rm d} r\\;{\\rm d} s\\; {\\rm d} t\\\\\n&= &\\hspace{-0.3cm}\\int_0^T \\hspace{-0.05cm} \\frac{(T-t)t}{T^2}\\int_0^T \\hspace{-0.2cm}\\int_0^T \\left\\Vert \\dot{y}\\left(s+\\frac{tr}{T}\\right)-\\dot{y}\\left(s+\\frac{tr}{T}-r\\right)\\right\\Vert {\\rm d} s\\; {\\rm d} r\\; {\\rm d} t\\nonumber\n\\end{eqnarray}\nThe last inner integral has been taken over one period, so we may shift it by $tr/T$ in order to obtain\n\\begin{eqnarray*}\n\\int_0^T\\int_0^T \\left\\Vert y(r)-y(s)\\right\\Vert {\\rm d} r\\;{\\rm d} s &\\leq & \\int_0^T \\frac{(T-t)t}{T^2}dt \\int_0^T \\int_0^T \\left \\Vert \\dot{y}(s+r)-\\dot{y}(s)\\right \\Vert {\\rm d} s\\; {\\rm d} r\\\\\n&=& \\frac{T}{6} \\int_0^T \\int_0^T \\left \\Vert \\dot{y}(r)-\\dot{y}(s)\\right \\Vert {\\rm d} s\\; {\\rm d} r\n\\end{eqnarray*}\ngiving us the desired inequality for arbitrary Banach spaces.\\\\\n\nFrom now on we consider the case when $X$ is in fact a strictly convex Banach space. The only actual inequality in the above argument occurs in line (\\ref{inequality martelli})\nwhere we use the triangle inequality for the Banach space $X$. Note that in doing so, we have a weight\n$$\\frac{(T-t)t}{T^2}$$\nin front of the inner integral which vanishes at $t = 0,T$. In particular, if we show that this inequality actually has to be strict for some $s$ and some $0 < t < T$, our statement follows. Additionally, because of the weight, these conditions are tight as the triangle inequality could fail to be strict at $t = 0,T$ without causing the chain of inequalities to become strict. Note that from the continuity of $\\dot{y}(t)$ we obtain that the functions\n$$ (s,t) \\rightarrow \\left\\|\\int_{0}^{T}\\dot{y}\\left(s+\\frac{tr}{T}\\right) -\\dot{y}\\left(s+\\frac{tr}{T}-r\\right){\\rm d} r\\right\\|$$\n$$ (s,t) \\rightarrow \\int_{0}^{T}\\left\\|\\dot{y}\\left(s+\\frac{tr}{T}\\right) -\\dot{y}\\left(s+\\frac{tr}{T}-r\\right)\\right\\| {\\rm d} r$$\nare continuous as well. Fix $s$ and $0 < t < T$, fix an arbitrarily fine decomposition $0 = a_0 < a_1 < \\dots < a_n = T$ and abbreviate\n$$ b_i := \\int_{a_i}^{a_{i+1}}\\dot{y}\\left(s+\\frac{tr}{T}\\right) \\qquad \\mbox{and} \\qquad c_i := \\int_{a_i}^{a_{i+1}}\\dot{y}\\left(s+\\frac{tr}{T}-r\\right).$$\n\nIf there is in fact equality in (1), then we need to have equality in every step of iteratively applying the triangle inequality and thus\n\\begin{eqnarray*}\n\\left\\|\\sum_{i=0}^{n-1}{b_i-c_i}\\right\\| &=& \\left\\|b_0 - c_0\\right\\| + \\left\\|\\sum_{i=1}^{n-1}{b_i-c_i}\\right\\| \\\\\n&=& \\left\\|b_0 - c_0\\right\\| + \\left\\|b_1 - c_1\\right\\| + \\left\\|\\sum_{i=2}^{n-1}{b_i-c_i}\\right\\| \\\\\n&=& \\dots \\\\\n&=& \\sum_{i=0}^{n-1}{\\left\\|b_i-c_i\\right\\|}.\n\\end{eqnarray*}\nW.l.o.g. we assume that all the terms satisfy $b_{i}-c_{i} \\neq \\textbf{0}$. Strict convexity implies in the last line of this argument that\n$b_{n-2} - c_{n-2}$ and $b_{n-1} - c_{n-1}$ are collinear. By the same reasoning $b_{n-3} - c_{n-3}$ and $(b_{n-2} - c_{n-2}) + (b_{n-1} - c_{n-1})$\nare collinear, however, the last expression itself is collinear to $b_{n-2}-c_{n-2}$ as well as $b_{n-1}- c_{n-1}$. Iterating this argument shows\nthat all $b_{i}-c_{i}$ are necessarily collinear. Using the continuity of $\\dot{y}(t)$, making the partition sufficiently small and applying the\nfundamental theorem of calculus, we can deduce that for every fixed $s$ and $0 < t < T$ there exists a vector $\\textbf{v} \\in X$ and a function\n$g:[0,T] \\rightarrow \\mathbb{R}$ such that for all $0 \\leq r \\leq T$\n\\begin{equation}\\dot{y}\\left(s+\\frac{tr}{T}\\right) - \\dot{y}\\left(s+\\frac{tr}{T}-r\\right) = g(r)\\textbf{v}. \\label{fund}\\end{equation}\n Note, however, that both $g$ and $\\textbf{v}$ depend on the previously fixed $s,t$. Since $y$ is not constant, it is possible to find and fix an $s$ such that\n$$ \\dot y(s) \\neq \\textbf{0}.$$\nWe now claim that this already implies that for all $0\\leq r\\leq T$\n$$\\dot{y}(s+r) = \\tilde{g}(r)\\textbf{v} + \\dot{y}(s).$$\nSuppose this was false, then there is an $r$ such that\n$$ \\dot{y}(s+r) \\notin \\left\\{\\dot y(s) + \\lambda \\textbf{v}| \\lambda \\in \\mathbb{R}\\right\\}.$$\nIn particular,\n$$ \\min_{\\lambda \\in \\mathbb{R}}{\\|\\dot{y}(s+r) - \\dot y(s) + \\lambda \\textbf{v}\\|} > 0.$$\nThis, however, can be seen to contradict (\\ref{fund}) by taking $t$ sufficiently small.\\\\\n\n\\noindent Since $y$ is periodic with period $T$,\n$$ \\int_{0}^{T}{\\dot{y}(s+r){\\rm d} r} = \\textbf{0} = \\left(\\int_{0}^{T}{\\tilde{g}(r){\\rm d} r}\\right)\\textbf{v} + T\\dot{y}(s).$$\nThis implies that $\\dot{y}(s)$ is a scalar multiple of $\\textbf{v}$, in which case\n$$ \\dot{y}(s+r) = \\left(\\tilde{g}(r) - \\frac{1}{T}\\int_{0}^{T}{\\tilde{g}(r){\\rm d} r}\\right)\\textbf{v}.$$\nThis establishes that $\\dot y(t)$ is one-dimensional, that is\n$$ \\dot y(t) = h(t) \\textbf{v}$$\nfor some $\\textbf{v} \\neq \\textbf{0}$ and a continuous, $T-$periodic function $h:[0,T] \\rightarrow \\mathbb{R}$.\\\\\n\nGoing back to an earlier stage of the argument, we had that for any fixed $s$ and $00$.\n\\end{proof}\n\n\n\\section{Improved lower bounds in $\\ell^p(\\mathbb{R}^n)$ and $L^p(M,\\mu)$}\n\\label{Better Bounds for lp}\n\nHaving established Wirtinger's inequality for $W^{1,p}_{\\rm per}([0,T],X)$ where $X$ is either $\\ell^p(\\mathbb{R}^n)$ or $L^p(M,\\mu)$, we can now prove\n the second contribution of this paper. The simple proof is essentially that for $p=2$ due to \\cite{busfishmar} which is a particular case of this result if one notes that $C_2^{-1}=2\\pi$.\n\n\\begin{thm}\\label{Minimal bound for L^p using Poincare}\nLet $x$ be a non-constant $T$-periodic solution to $\\dot{ x}=f( x)$ in either $X=\\ell^p(\\mathbb{R}^n)$ or $X=L^p(M,\\mu)$. Further, suppose that $f$ is Lipschitz continuous from $X$ into $X$ with Lipschitz constant $L$. Then\n\\begin{equation}\\label{TLC}\nTL\\ge C_p^{-1}.\n\\end{equation}\n\\end{thm}\n\n\\begin{proof}\nAs the function $x$ is a solution to the ODE, it is differentiable by definition. Moreover, a simple calculation shows that\n$$\n\\int_0^Tx(t+h)-x(t)\\,{\\rm d} t=0.\n$$\nHence Wirtinger's inequality for $W^{1,p}_{per}((0,T),X)$ is applicable to $x(t+h)-x(t)$ and thus\n\\begin{eqnarray*}\n\\int_0^T\\| x(t+h)- x(t)\\|^p_X\\,{\\rm d} t&\\le& C_p^pT^p \\int_0^T \\|\\dot{ x}(t+h)-\\dot{ x}(t)\\|^p_X\\,{\\rm d} t\\\\\n&=& C_p^p T^p\\int_0^T \\|f( x(t+h))-f( x(t))\\|^p_X\\,{\\rm d} t\\\\\n&\\le& L^pC_p^pT^p \\int_0^T \\| x(t+h)- x(t)\\|^p_X\\,{\\rm d} t.\n\\end{eqnarray*}\nDividing both sides by $\\int_0^T \\| x(t+h)- x(t)\\|^p_X\\,{\\rm d} t$, which is non-zero as $x$ is non-constant, yields (\\ref{TLC}). \n\\end{proof}\n\n Theorem \\ref{Minimal bound for L^p using Poincare} gives an improved lower bound on the product of Lipschitz constant $L$ and period $T$ for the spaces $\\ell^p(\\mathbb{R}^n)$ and $L^p(M,\\mu)$ for a range of $p$ around $p=2$. Figure \\ref{fig:BestConstant from generalised Wirtinger} plots $C_p^{-1}$ against $p$ for $1\\leq p\\leq 4$, and shows that $C_p^{-1}>6$ for $1.43 \\leq p\\leq 3.35$.\\\\\n\n\\begin{figure}[h!]\n \\begin{center}\n\\includegraphics[width=0.7\\columnwidth]{bestconstant2.png}\n\\caption{Improved lower bound near $p=2$ using Wirtinger's inequality}\n\\label{fig:BestConstant from generalised Wirtinger}\n\\end{center}\n\\end{figure}\n\n\\textbf{Remark 1.} For $1\\leq p<\\infty$ one can construct an example of an ODE in $L^{p}(M,\\mu)$ satisfying Lipschitz conditions on its derivative with period $2\\pi$.\nSuppose there are two sets $A\\cap B= \\emptyset$ such that\n\\[\n0 < \\mu(A) =\\mu(B).\n\\]\nand consider the ODE\n$$\\dot{z}=f(z)$$\nwith $f:L^{p}(M,\\mu) \\rightarrow L^{p}(M,\\mu)$ given by\n\\[\nf(z)= -\\frac{\\chi_{B}}{\\mu(A)}\\int_{A}{z d\\mu} + \\frac{\\chi_{A}}{\\mu(B)}\\int_{B}{z d\\mu}.\n\\]\nThen H\\\"older's inequality gives that for $L=1$ the quantity\n$$ I = \\|f(z)-f(w)\\|^p_{L^{p}(M,\\mu)}$$\nsatisfies\n\\begin{eqnarray*}\n\\hspace{-1cm} I \\hspace{-0.2cm}&=&\\hspace{-0.2cm} \\left\\| - \\chi_B \\frac{1}{\\mu(A)}\\int_{A}{z-w d\\mu} + \\chi_{A}\\frac{1}{\\mu(A)}\\int_{B}{z-w d\\mu} \\right\\|_{L^p}^p \\\\\n\\hspace{-0.2cm}&=&\\hspace{-0.2cm} \\left(\\frac{1}{\\mu(A)}\\int_{A}{z-w d\\mu}\\right)^p \\mu(B) + \\left(\\frac{1}{\\mu(B)}\\int_{B}{z-w d\\mu}\\right)^p \\mu(A) \\\\\n\\hspace{-0.2cm}&\\leq&\\hspace{-0.2cm} \\frac{1}{\\mu(A)^p} \\left(\\int_{A}{|z-w|^p d\\mu}\\right) \\mu(A)^{p-1}\\mu(B) + \\frac{1}{\\mu(B)^p}\\left(\\int_{B}{|z-w|^p d\\mu}\\right) \\mu(B)^{p-1}\\mu(A) \\\\\n\\hspace{-0.2cm}&\\leq&\\hspace{-0.2cm} \\|z-w\\|_{L^p}^{p}\n\\end{eqnarray*}\nand one explicit $2\\pi-$periodic solution is given by\n$$ z(t) = -(\\cos{t})\\chi_A + (\\sin{t})\\chi_B.$$\nNotice that this example can be generalised further to the case when $0 < \\mu(A) \\neq \\mu(B)$.\\\\\n\n\\textbf{Remark 2.} Let $X$ be a Banach space which obeys `almost' a Hilbert space structure in the sense of the norm, that is there exists a $\\varepsilon>0$ such that\n$$(1-\\varepsilon) \\| x \\|_H \\leq \\| x \\|_X \\leq (1+\\varepsilon) \\| x \\|_H.$$\nLet $x:\\mathbb{R}\\rightarrow X$ be a $T$-periodic solution to the ODE $\\dot{x}=f(x)$ with $f$ being Lipschitz continuous with Lipschitz constant $L$. Since \\begin{eqnarray*}\n\\|f(x) - f(y)\\|_H \\leq \\frac{1}{1-\\varepsilon} \\| f(x) - f(y) \\|_X \\leq \\frac{1}{1 - \\varepsilon} L \\| x - y\\|_X \\leq \\frac{1+\\varepsilon}{1-\\varepsilon} L \\|x - y\\|_H,\n\\end{eqnarray*}\nit follows that $f$ is also Lipschitz continuous with respect to the Euclidean norm with Lipschitz constant $L'=\\frac{1+\\varepsilon}{1-\\varepsilon}L$. At the same time,\nthe length of the curve $x$ as measured in the Hilbert space is smaller than $(1+\\varepsilon)T$ and using the fact that $c=2\\pi$ in any Hilbert space we may conclude\nthat $$TL\\geq 2\\pi\\frac{1-\\varepsilon}{(1+\\varepsilon)^2}.$$\nHowever, this approximation lags behind the numerical results for $\\ell^p$ obtained at the beginning of this section, especially for high dimensions.\\\\\n\n\\textbf{Remark 3.} Dvoretzky's theorem in \\cite{dvor} guarantees that for any $\\varepsilon > 0$ there exists $n \\in \\mathbb{N}$ sufficiently large such that any Banach space\nwith $\\dim X \\geq n$ contains a two-dimensional subspace with Banach-Mazur distance to $\\ell_2^2$ at most $1+\\varepsilon$. The example of a simple circle in $\\ell^2_2$ realizes\n$TL = 2\\pi$. This means that in any Banach space $X$ it is possible to construct an ODE satisfying $TL \\leq 2\\pi + \\varepsilon$, where $\\varepsilon$ depends only on\nthe dimension of $X$. We do not know whether there is always an ODE for which $TL \\leq 2\\pi$.\n\n\n\\section{Conclusion}\nAs discussed in the introduction, the key question is what intrinsic property of a space $X$ determines the largest (and hence best) constant $C_X$ such that $LT\\ge C_X$. One of\n these intrinsic properties is strict convexity for which we have shown that the constant must be strictly larger than $6$. A natural question that arises is whether there exists\n a Banach space in which the optimal constant is neither $6$ nor $2\\pi$.\\\\\n\nHowever, explicit bounds are difficult to obtain. Even in the simple case $X=\\ell^p({\\mathbb R}^n)$ this is not known, although our simple argument gives an explicit lower bound for\n$p$ around $p=2$. It is interesting that a simple calculation shows that $C_p=C_{p'}$ when $p$ and $p'$ are conjugates; but it is not known whether the optimal constants\n in $\\ell^p$ and $\\ell^{p'}$ do in fact coincide (this interesting question was suggested to one of us in a personal communication from Mario Martelli).\\\\\n\nWhile the use of an $L^p$-based Wirtinger inequality suits the $\\ell^p$-spaces, there is no reason why these exponents should match. Given a Banach space $X$ it would be interesting to determine the optimal constants in the family of inequalities\n$$\n\\left(\\int_0^1 \\| u(t)\\|^p_X\\,{\\rm d} t\\right)^{1/p}\\leq C_p(X)\\left(\\int_0^T \\|\\dot{ u}(t)\\|^p_X\\,{\\rm d} t\\right)^{1/p},\n$$\nnoting that as a consequence of such a family of inequalities and the argument of Theorem \\ref{Minimal bound for L^p using Poincare} one would obtain\n$$\nTL\\ge\\sup_p C_p(X)^{-1}.\n$$\n\n\n\n\n\\section*{Acknowledgement}\nJCR is supported by an EPSRC Leadership Fellowship, grant number EP/G007470/1. MACN gratefully acknowledges funding from the European Research Council under the European\nUnion's Seventh Framework Programme (FP7/2007-2013) / ERC grant agreement n$^{\\circ}$ 291053. SS is supported by a Hausdorff scholarship of the Bonn International\nGraduate School.\n\n\n", "meta": {"timestamp": "2013-07-24T02:08:42", "yymm": "1210", "arxiv_id": "1210.6582", "language": "en", "url": "https://arxiv.org/abs/1210.6582", "abstract": "Let x(t) be a non-constant T-periodic solution to the ordinary differential equation x'= f(x) in a Banach space X where f is assumed to be Lipschitz continuous with constant L. Then there exists a constant c such that T L >= c, with c only depending on X. It is known that c >= 6 in any Banach space and that c = 2{\\pi} in any Hilbert space, but whereas the bound of c = 2 pi is sharp in any Hilbert space, there exists only one known example of a Banach space such that c = 6 is optimal. In this paper, we show that the inequality is in fact strict in any strictly convex Banach space. Moreover, we improve the lower bound for l^p(R^n) and L^p(M, {\\mu}) for a range of p close to p = 2 by using a form of Wirtinger's inequality for functions in W^{1,p}([0, T ], L^p(M, {\\mu})).", "subjects": "Classical Analysis and ODEs (math.CA)", "title": "Minimal Periods for Ordinary Differential Equations in Strictly Convex Banach Spaces and Explicit Bounds for some l^p-Spaces", "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.9896718474806948, "lm_q2_score": 0.8652240912652671, "lm_q1q2_score": 0.8562879248873022}} {"url": "https://arxiv.org/abs/1101.0388", "title": "Demystifying a divisibility property of the Kostant partition function", "abstract": "We study a family of identities regarding a divisibility property of the Kostant partition function which first appeared in a paper of Baldoni and Vergne. To prove the identities, Baldoni and Vergne used techniques of residues and called the resulting divisibility property \"mysterious.\" We prove these identities entirely combinatorially and provide a natural explanation of why the divisibility occurs. We also point out several ways to generalize the identities.", "text": "\\section{Introduction}\n\\label{sec:in}\n\nThe objective of this paper is to provide a natural combinatorial explanation of a divisibility property of the Kostant partition function. The question of evaluating Kostant partition functions has been subject of much interest, without a satisfactory combinatorial answer. To mention perhaps the most famous such case: it is known that $$K_{A_n^+}(1, 2, \\ldots, n, -{n+1 \\choose 2})=\\prod_{k=1}^{n}C_k,$$ where $C_k=\\frac{1}{k+1}{2k \\choose k}$ is the Catalan number, yet there is no combinatorial proof of the above identity! \n\nGiven such lack of understanding of the evaluation of the Kostant partition function, it seems a worthy proposition to provide a simple explanation for its certain divisibility properties. \nWe explore divisibility properties of Kostant partition functions of types $A_n$ and $C_{n+1}$, noting that such properties in types $B_{n+1}$ and $D_{n+1}$ are easy consequences of the type $C_{n+1}$ case. The type $A_n$ family of identities we study first appeared in a paper by Baldoni and Vergne \\cite{BV}, where the authors prove the identities using residues, and where they call the divisibility property ``mysterious.\" It is our hope that the combinatorial argument we provide successfully demystifies the divisibility property of the Kostant partition function and provides a natural explanation why things happen the way they do. \n\n\n\nThe outline of the paper is as follows. \nIn Section \\ref{sec:bv} we define Kostant partition functions of type $A_n$ and prove the Baldoni-Vergne identities combinatorially. Our proof also yields an affirmative answer to a question of Stanley \\cite{S} regarding a possible bijective proof of a special case of the Baldoni-Vergne identities.\nIn Section \\ref{sec:c} we define Kostant partition functions of type $C_{n+1}$, relate them to flows, and show how to modify our proof for the Baldoni-Vergne identities to obtain their analogues for type $C_{n+1}$. We also point out other possible variations of these identities in the type $C_{n+1}$ case. \n\n\\section{The Baldoni-Vergne identities}\n\\label{sec:bv}\n\nBefore stating the Baldoni-Vergne identities, we need a few definitions. Throughout this section the graphs $G$ we consider are on the vertex set $[n+1]$ with possible multiple edges, but no loops. Denote by $m_{ij}$ the multiplicity of edge $(i, j)$, $i0$\n(see~\\cite{dauge}).\n\\begin{figure}\n \\centering\n \\begin{tabular}{cc}\n \\begin{overpic}[scale=0.35]{ll0}\n \\put(-5,15){\\begin{sideways}\\textbf{Relative approximation error}\\end{sideways}}\n \\put(32,-5) {\\textbf{Mesh size $\\mathbf{h}$}}\n \\put(62,69){\\textbf{4/3}}\n \\end{overpic} \n &\\qquad\n \\begin{overpic}[scale=0.35]{ll2}\n \\put(-5,15){\\begin{sideways}\\textbf{Relative approximation error}\\end{sideways}}\n \\put(32,-5) {\\textbf{Mesh size $\\mathbf{h}$}}\n \\put(51,76){\\textbf{2}}\n \\put(51,49){\\textbf{4}}\n \\put(51,22){\\textbf{6}}\n \\end{overpic}\n \\end{tabular}\n \\medskip\n \\caption{Test Case~2: convergence curves of the first eigenvalue\n (left panel) and the third eigenvalue (right panel) using the\n nonconforming VEM space $\\VS{h}_{k}$, $k=1,2,3$,}\n \\label{fig:lshape:rates}\n\\end{figure}\nInstead, the third eigenvalue is analytical and the optimal order of\nconvergence is obtained, which can be seen by comparing the slopes of\nthe error curves and the corresponding theoretical slopes reported on\nthe plot.\nThese results confirm the convergence analysis of the previous section\nand the optimality of the method also on nonconvex domains using\npolygonal meshes.\n\n\\section{Introduction}\\label{sec:Introduction}\n\\input{introduction.tex}\n\n\\section{The continuous eigenvalue problem}\n\\label{sec:Continuous:problem}\n\nIn this section we describe the continuous eigenvalue problem and its\nassociated source problem.\nThroughout the paper, we use the notation of Sobolev spaces, norms and\nseminorms detailed in~\\cite{Adams:1975}.\nIn particular, the symbols $\\snorm{\\cdot}{s,\\omega}$ and\n$\\norm{\\cdot}{s,\\omega}$ are the seminorm and the norm of the Sobolev\nspace $\\HS{s}(\\omega)$ defined on the open bounded subset $\\omega$ of\n$\\mathbbm{R}^{\\DIM}$, and $(\\cdot,\\cdot)_{\\omega}$ is the $\\LTWO$-inner\nproduct.\nIf $\\omega$ is the whole computational domain $\\Omega$, the subscript may\nbe omitted and we may denote the Sobolev seminorm and norm by\n$\\snorm{\\cdot}{s}$ and $\\norm{\\cdot}{s}$, and the $L^2$-inner product\nby $(\\cdot,\\cdot)$.\n\n\\medskip\nLet $\\Omega\\subset\\mathbbm{R}^{\\DIM}$ for $\\DIM=2,3$ be an open\npolytopal domain with Lipschitz boundary $\\Gamma$.\nConsider the eigenvalue problem: \\emph{Find\n $\\lambda \\in\\mathbbm{R}$, such that there exists $u \\neq 0$}:\n\\begin{align}\n -\\Deltau &= \\lambdau\\phantom{0} \\quad\\textrm{in~}\\Omega,\\label{eq:mod00:A}\\\\\n u &= 0\\phantom{\\lambdau} \\quad\\textrm{on~}\\Gamma.\\label{eq:mod00:B}\n\\end{align}\nIts variational formulation\nreads as:\n\\emph{Find $(\\lambda,u)\\in\\mathbbm{R}\\times\\VS{}$, $\\NORM{u}{0} = 1$, such\n that}\n\\begin{equation}\\label{eq:eigPbm}\n a(u,v)=\\lambdab(u,v)\\qquad\\forallv\\in\\VS{},\n\\end{equation}\nwhere $\\VS{}=H^1_0(\\Omega)$, the bilinear form\n$a:\\VS{}\\times\\VS{}\\to\\mathbbm{R}$ is given by\n\\begin{align}\n a(u,v)=\\int_{\\Omega}\\nablau\\cdot\\nablav\\, {\\rm d}\\Omega\\qquad\\forallu,\\,v\\in\\VS{},\n \\label{eq:exact:a}\n\\end{align}\nand the bilinear form $b:\\VS{}\\times\\VS{}\\to\\mathbbm{R}$ is the\n$\\LTWO$-inner product on $\\Omega$, i.e., \n\\begin{align}\n\tb(u,v)=(u,v) \\qquad\\forallu,\\,v\\in\\VS{} \\,.\n\t\\label{eq:exact:b}\n\\end{align}\nThe eigenvalues of problem~\\eqref{eq:eigPbm} form a positive\nincreasing divergent sequence and the corresponding eigenfunctions are\nan orthonormal basis of $\\VS{}$ with respect both to the $\\LTWO$-inner\nproduct and the scalar product associated with the bilinear form\n$a(\\cdot,\\cdot)$. Moreover, each eigenspace has finite dimension ~\\cite{Boffi:2010,Babuska-Osborn:1991}.\n\n\\medskip\nIn the spectral convergence analysis, we will need the approximation\nresults of the source problem associated with~\\eqref{eq:eigPbm}, which\nwe state as follows:\n\\emph{Find $\\uss\\in\\VS{}$ such that}\n\\begin{equation}\\label{eq:sourcePbm}\n a(\\uss,v)=b(f,v)\\quad\\forallv\\in\\VS{},\n\\end{equation}\nwhere we assume the forcing term $f$ (at least) in $\\LTWO(\\Omega)$.\nDue to regularity results~\\cite{Agmon:1965,Grisvars:1992}, there exists a constant\n$r>1/2$ that depends only on $\\Omega$ such that the solution $\\uss$\nbelongs to the space $\\HS{1+r}(\\Omega)$.\nIf $\\Omega$ is a convex polytopal domain, then $r\\geq\n1$.\nInstead, $r\\geq\\pi/\\omega-\\varepsilon$ for any $\\varepsilon >0$ if\n$\\Omega$ is a two-dimensional non-convex polygonal domain with maximum\ninterior angle $\\omega<2\\pi$.\nA similar result holds in the three-dimensional case, $\\omega$ being the maximum\nreentrant wedge angle.\nEventually, there exists a positive constant $C$ such that\n\\begin{equation}\n \\snorm{ \\uss }{1+r}\\leq C\\norm{f}{0} \\, .\n \\label{eq:regularity}\n\\end{equation}\n\n\\section{The nonconforming virtual element method} \n\\label{sc:spaces}\n\nIn this section, we introduce the family of mesh decompositions of the\ncomputational domain and the mesh regularity assumptions and, then, we\ndefine the non-conforming virtual element space (and related\napproximation properties) that we need for the proper formulation of\nthe virtual element method, cf.~\\cite{Ayuso-Lipnikov-Manzini:2016}.\n\n\\subsection{Mesh definition and regularity assumptions}\nLet $\\mathcal{T}=\\{\\Omega_{\\hh}\\}_{h}$ be a family of decompositions of\n$\\Omega$ into nonoverlapping polytopal elements $\\P$ with\nnonintersecting boundary $\\partial\\P$, center of gravity $\\xvP$,\n$\\DIM$-dimensional measure $\\ABS{\\P}$, and diameter\n$\\hh_{\\P}=\\sup_{\\mathbf{x},\\mathbf{y}\\in\\P}\\abs{\\mathbf{x}-\\mathbf{y}}$.\nThe subindex $h$ that labels each mesh $\\Omega_{\\hh}$ is the maximum of the\ndiameters $\\hh_{\\P}$ of the elements of that mesh.\nThe boundary of $\\P$ is formed by straight edges when $\\DIM=2$ and\nflat faces when $\\DIM=3$.\nThe midpoint and length of each edge $\\E$ are denoted by $\\xvE$ and\n$\\hh_{\\E}$, respectively.\nThe center of gravity, diameter and area of each face $\\F$ are denoted\nby $\\xvF$, $\\hh_{\\F}$, and $\\ABS{\\F}$, respectively.\nSometimes we may refer to the geometric objects forming the elemental\nboundary $\\partial\\P$ by the term \\emph{side} instead of\n\\emph{edge}/\\emph{face}, and adopt a unified notation by using the\nsymbol $\\S$ instead of $\\E$ or $\\F$ regardless of the number of\nspatial dimensions.\nAccordingly, $\\xvS$, $\\hh_{\\S}$, and $\\ABS{\\S}$ denote the center of gravity,\ndiameter, and measure of side $\\S$.\n\nWe denote the unit normal vector to the elemental boundary\n$\\partial\\P$ by $\\norP$, and the unit normal vector to edge $\\E$, face\n$\\F$ and side $\\S$ by $\\norE$, $\\norF$, $\\norS$, respectively.\nEach vector $\\norP$ points out of $\\P$ and the orientation of $\\norE$,\n$\\norF$, and $\\norS$ is fixed \\emph{once and for all} in every mesh\n$\\Omega_{\\hh}$.\nFinally, $\\Eset_{h}$, $\\Fset_{h}$, and $\\Sset_{h}$ denote the\nset of edges, faces, and sides of the skeleton of $\\Omega_{\\hh}$.\nWe may distinguish between \\emph{internal} and \\emph{boundary} sides\nby using the superscript $\\INT$ and $\\BND$.\nTherefore, $\\Sset_{h}^{\\INT}$ is the set of the internal sides,\n$\\Sset_{h}^{\\BND}$ the set of the boundary sides, and, obviously,\n$\\Sset_{h}^{\\INT}\\cap\\Sset_{h}^{\\BND}=\\emptyset$ and\n$\\Sset_{h}=\\Sset_{h}^{\\INT}\\cup\\Sset_{h}^{\\BND}$.\n\n\\medskip \nNow, we state the mesh regularity assumptions that are required for\nthe convergence analysis.\nSince the method cannot be used simultaneously for $\\DIM=2$ and\n$\\DIM=3$ and, hence, no ambiguity is possible in such sense, we may\nrefer to the two- and three-dimensional case using the same label\n\\ASSUM{A0} and the same symbol $\\rho$ to denote the \\emph{mesh\n regularity constant}.\nNote that the assumptions for $\\DIM=2$ can be derived from those for\n$\\DIM=3$ by reducing the spatial dimension.\n\n\\medskip\n\\noindent\n\\ASSUM{A0}~\\textbf{Mesh regularity assumptions}.\n\n\\begin{itemize}\n \\medskip\n\\item $\\mathbf{\\DIM=3}$.\n \n There exists a positive constant $\\varrho$ independent of $h$ (and, hence, of\n $\\Omega_{\\hh}$) such that for every polyhedral element $\\P\\in\\Omega_{\\hh}$ it\n holds that\n \\begin{description}\n \\item[$(i)$] $\\P$ is star-shaped with respect to a ball with radius $\\ge\\varrho\\hh_{\\P}$;\n \\item[$(ii)$] every face $\\F\\in\\P$ is star-shaped with respect to a\n disk with radius $\\ge\\varrho\\hh_{\\F}$;\n \\item[$(iii)$] for every edge $\\E\\in\\partial\\F$ of every face\n $\\F\\in\\partial\\P$ it holds that $\\hh_{\\E}\\geq\\varrho\\hh_{\\F}\\geq\\varrho^2\\hh_{\\P}$.\n \\end{description}\n\n \\medskip\n\\item $\\mathbf{\\DIM=2}$.\n \n There exists a positive constant $\\varrho$ independent of $h$ (and, hence, of\n $\\Omega_{\\hh}$) such that for every polygonal element $\\P\\in\\Omega_{\\hh}$ it\n holds that\n \n \\begin{description}\n \\item[$(i)$] $\\P$ is star-shaped with respect to a disk with radius $\\ge\\varrho\\hh_{\\P}$;\n \\item[$(ii)$] for every edge $\\E\\in\\partial\\P$ it holds that\n $\\hh_{\\E}\\geq\\varrho\\hh_{\\P}$.\n \\end{description}\n\\end{itemize}\n\n\\medskip\n\\begin{remark}\n The star-shapedness property implies that elements and faces are\n \\emph{simply connected} subsets of $\\mathbbm{R}^{\\DIM}$ and\n $\\mathbbm{R}^{\\DIM-1}$, respectively.\n \n The scaling assumption implies that the number of edges and faces in\n each elemental boundary is uniformly bounded over the whole mesh\n family $\\mathcal{T}$.\n\\end{remark}\n\n\\subsection{Basic setting}\nWe introduce the broken Sobolev space for any $s>0$\n\\begin{align*}\n \\HS{s}(\\Omega_{\\hh}) \n = \\prod_{\\P\\in\\Omega_{\\hh}}\\HS{s}(\\P) \n = \\big\\{\\,v\\in\\LTWO(\\Omega)\\,:\\,\\restrict{v}{\\P}\\in\\HS{s}(\\P)\\,\\big\\}, \n\\end{align*}\nand define the broken $\\HS{s}$-norm\n\\begin{align}\n \\label{eq:Hs:norm-broken}\n \\norm{v}{s,h}^2 = \\sum_{\\P\\in\\Omega_{\\hh}}\\norm{v}{s,\\P}^{2}\n \\qquad\\forall\\,v\\in\\HS{s}(\\Omega_{\\hh}),\n\\end{align}\nand for $s=1$ the broken $H^{1}$-seminorm\n\\begin{align}\n \\label{eq:norm-broken}\n \\snorm{v}{1,h}^2 = \\sum_{\\P\\in\\Omega_{\\hh}}\\norm{\\nablav}{0,\\P}^{2} \n \\qquad\\forall\\,v\\in\\HONE(\\Omega_{\\hh}).\n\\end{align}\n\nLet $\\sigma\\subset\\partial\\Psgm^{+}\\cap\\partial\\Psgm^{-}$ be the internal\nside (edge/face) shared by elements $\\Psgm^{+}$ and $\\Psgm^{-}$, and $v$\na function that belongs to $\\HONE(\\Omega_{\\hh})$.\nWe denote \nthe traces of $v$ on $\\S$ from the interior of elements\n$\\Psgm^{\\pm}$ by $\\vs_{\\S}^{\\pm}$,\nand the unit normal vectors to $\\S$ pointing from $\\Psgm^{\\pm}$ to\n$\\Psgm^{\\mp}$ by $\\norS^{\\pm}$.\nThen, we introduce the jump operator\n$\\jump{v}=\\vs_{\\S}^{+}\\norS^{+}+\\vs_{\\S}^{-}\\norS^{-}$ at each internal side\n$\\S\\in\\Sset_{h}^{\\INT}$, and $\\jump{v}=\\vs_{\\S}\\norS$ at each boundary\nside $\\S\\in\\Sset_{h}^{\\BND}$.\nThe nonconforming space $H^{1,nc}(\\Omega_{\\hh};k)$ for any integer $k\\geq1$ is\nthe subspace of the broken Sobolev space $\\HONE(\\Omega_{\\hh})$ defined as\n\\begin{equation}\n \\label{eq:H1-nc:def}\n H^{1,nc}(\\Omega_{\\hh};k) = \\left\\{\\,\n v\\in\\HONE(\\Omega_{\\hh})\\,:\\,\\int_{\\S}\\jump{v}\\cdot\\norS\\,q\\,{\\rm d}\\S=0\\,\n \\,\\,\\forall\\,q\\in\\PS{k-1}(\\S),\n \\,\\,\\forall\\S\\in\\Sset_{h}\n \\,\\right\\}.\n\\end{equation}\nSince $\\jump{v}=0$ on any internal mesh side whenever $v$ belongs to\n$\\HONE(\\Omega)$, it is trivial to show that\n$H^1_0(\\Omega)\\subsetH^{1,nc}(\\Omega_{\\hh};k)$.\n\n\\medskip\nHereafter, we consider the extension of the bilinear form\n$a(\\cdot,\\cdot)$ to the broken Sobolev space $\\HONE(\\Omega_{\\hh})$, which is\ngiven by splitting it as sum of local terms:\n\\begin{align}\n &a:\\HS{1}(\\Omega_{\\hh})\\times\\HS{1}(\\Omega_{\\hh})\\to\\mathbbm{R}\\textrm{~~with~~}\\nonumber\\\\[0.5em]\n &a(u,v) = \\sum_{\\P\\in\\Omega_{\\hh}}\\as^{\\P}(u,v)\n \\textrm{~~where~~}\\as^{\\P}(u,v) = \\int_{\\P}\\nablau\\cdot\\nablav\\, {\\rm d}\\P,\n \\quad\\forallu,v\\in\\HS{1}(\\Omega_{\\hh}).\n \\label{eq:extension}\n\\end{align}\nClearly, the same definition applies when at least one entry of\n$a(\\cdot,\\cdot)$ belongs to the nonconforming space\n$H^{1,nc}(\\Omega_{\\hh};k)$, which is a subspace of $\\HONE(\\Omega_{\\hh})$, and the\nnonconforming virtual element space, which will be defined in the next\nsection as a subspace of $H^{1,nc}(\\Omega_{\\hh};k)$. \n\n\n\\medskip\nThe nonconforming space with $k=1$ has the minimal regularity required\nfor the VEM formulation and the convergence analysis.\nIt is straightforward to show that $\\snorm{\\cdot}{1,h}$ is a norm on\n$H^{1,nc}(\\Omega_{\\hh};k)$, although it is only a seminorm for the discontinuous\nfunctions of $\\HONE(\\Omega_{\\hh})$.\nMoreover, using the Poincar\\'e-Friedrichs inequality\n$\\norm{v}{0}^{2}\\leq C_{PF}\\snorm{v}{1,h}^{2}$, which\nholds for every $v\\inH^{1,nc}(\\Omega_{\\hh};k)$ and some positive constant\n$C_{PF}$ independent of $h$, cf.~\\cite{Brenner:2003}, we can show\nthat $\\snorm{\\cdot}{1,h}$ is equivalent to $\\norm{\\cdot}{1,h}$.\nTherefore, we may refer to the seminorm $\\snorm{\\cdot}{1,h}$ as a norm\nin $H^{1,nc}(\\Omega_{\\hh};k)$.\n\nAccording to~\\cite{Ayuso-Lipnikov-Manzini:2016}, for \n$u\\in\\HS{s}(\\Omega)$ with $s\\geq3\\slash{2}$ solution to \n\\eqref{eq:sourcePbm}\nand $v\\inH^{1,nc}(\\Omega_{\\hh};k)$\nwe find that\n\\begin{align}\n \\label{eq:def:Nh}\n \\mathcal{N}_{\\hh}(u,v) \n := a(u,v) - b(f,v)\n = \\sum_{\\S\\in\\Sset_{h}}\\int_{\\S}\\nablau\\cdot\\jump{v}\\, {\\rm d}\\S.\n\\end{align}\nThe quantity $\\mathcal{N}_{\\hh}(u,v)$ is called the \\emph{conformity error}.\n\n\\medskip\nWe now recall an estimate for the term measuring the nonconformity.\n\\begin{lemma}\\label{le:Nh}\n Assume \\ASSUM{A0} is satisfied. \n \n Let $u\\in\\HS{1+r}(\\Omega)$ with $r\\geq 1$ be the solution to\n \\eqref{eq:sourcePbm}.\n \n Let $v\\inH^{1,nc}(\\Omega_{\\hh};k)$, $k\\geq1$, as defined in\n \\eqref{eq:H1-nc:def}.\n \n Then, there exists a constant $C>0$ depending only on the polynomial\n degree and the mesh regularity such that\n \\begin{align}\n \\abs{\\mathcal{N}_{\\hh}(u,v)} \\leq Ch^{t}\\snorm{u}{1+r}\\snorm{v}{1,h}\n \\end{align}\n where $t=\\min\\{k,r\\}$ and \n $\\mathcal{N}_{\\hh}(u,v)$ is defined in \\eqref{eq:def:Nh}.\n\\end{lemma}\n\n\\medskip\nThroughout the paper, $\\PS{\\ell}(D)$ denotes the space of polynomials\nof degree up to $\\ell$ for any integer number $\\ell\\geq0$ on the\nbounded connected subset $D$ of $\\mathbbm{R}^{\\nu}$ with $\\nu=1,2,3$.\nThe polynomial space $\\PS{\\ell}(D)$ is finite dimensional and we\ndenote its dimension by $\\pi_{\\ell,\\nu}$.\nIt holds that $\\pi_{\\ell,1}=\\ell+1$ for $\\nu=1$;\n$\\pi_{\\ell,2}=(\\ell+1)(\\ell+2)\\slash{2}$ for $\\nu=2$;\n$\\pi_{\\ell,3}=(\\ell+1)(\\ell+2)(\\ell+3)\\slash{6}$ for $\\nu=3$.\nWe also conventionally take $\\PS{-1}(D)=\\{0\\}$ and $\\pi_{-1,\\nu}=0$.\nLet $\\mathbf{x}_{D}$ denote the center of gravity of $D$ and $h_{D}$ its\ncharacteristic length, as, for instance, the edge length for $\\nu=1$,\nthe face diameter for $\\nu=2$ and the cell diameter for $\\nu=3$.\nA basis for $\\PS{\\ell}(D)$ is provided by $\\calM{\\ell}(D) =\n\\left\\{\\,\\left(({\\mathbf{x}-\\mathbf{x}_{D})}\\slash{h_{D}}\\right)^{\\alpha}\\textrm{~with~}\\abs{\\alpha}\\leq\\ell\\,\\right\\}$,\nthe set of the \\emph{scaled monomials of degree up to $\\ell$}, where\n$\\alpha=(\\alpha_1,\\ldots\\alpha_{\\nu})$ is a $\\nu$-dimensional\nmulti-index of nonnegative integers $\\alpha_i$ with degree\n$\\abs{\\alpha}=\\alpha_1+\\ldots+\\alpha_{\\nu}$ and, with obvious notation,\n$\\mathbf{x}^{\\alpha}=x_1^{\\alpha_1}\\cdotsx_{\\nu}^{\\alpha_{\\nu}}$ for any\n$\\mathbf{x}\\in\\mathbbm{R}^{\\nu}$.\nWe will also use the set of \\emph{scaled monomials of degree exactly\n equal to $\\ell$}, denoted by $\\calM{\\ell}^{*}(D)$ and obtained by\nsetting $\\abs{\\alpha}=\\ell$ in the definition of $\\calM{\\ell}(D)$.\nFinally, we denote by $\\Piz{\\ell}\\,:\\,\\LTWO(\\P)\\to\\PS{\\ell}(\\P)$ for\n$\\ell\\geq0$ the $\\LTWO$-orthogonal projection onto the polynomial\nspace $\\PS{\\ell}(\\P)$, and by\n$\\PiSz{\\ell}\\,:\\,\\LTWO(\\S)\\to\\PS{\\ell}(\\S)$ for $\\ell\\geq0$ the\n$\\LTWO$-orthogonal projection onto the polynomial space\n$\\PS{\\ell}(\\S)$.\n\n\\subsection{Local and global nonconforming virtual element space}\nWe construct the local nonconforming virtual element space by\nresorting to the so-called \\emph{enhancement strategy} originally\ndevised in~\\cite{Ahmad-Alsaedi-Brezzi-Marini-Russo:2013} for the\nconforming \\text{VEM}{} and later extended to the nonconforming \\text{VEM}{}\nin~\\cite{Cangiani-Manzini-Sutton:2017}.\nTo this end, on every polytopal cell $\\P\\in\\Omega_{\\hh}$ and for any integer\nnumber $k\\geq 1$ we first define the finite dimensional functional\nspace\n\\begin{equation}\\label{eq:def:Vhkt0}\n {\\widetilde{V}}^{\\hh}_{k}(\\P) = \\left\\{\\,\n \\vs_{\\hh}\\in\\HONE(\\P)\\,:\\,\n \\frac{\\partial\\vs_{\\hh}}{\\partial\\mathbf{n}}\\in\\PS{k-1}(\\S)\\, \\, \\forall\\S\\subset\\partial\\P,\\,\n \\Delta\\vs_{\\hh}\\in\\PS{k}(\\P)\\,\n \\right\\}.\n\\end{equation}\nWe notice that the space ${\\widetilde{V}}^{\\hh}_{k}(\\P)$ clearly contains the polynomials of degree $k$.\n\nThen, we introduce the set of continuous linear functionals from\n${\\widetilde{V}}^{\\hh}_{k}(\\P)$ to $\\mathbbm{R}$ that for every virtual function $\\vs_{\\hh}$ of\n${\\widetilde{V}}^{\\hh}_{k}(\\P)$ provide:\n\\begin{description}\n\\item[]\\textbf{(D1)} the moments of $\\vs_{\\hh}$ of order up to $k-1$ on\n each $(\\DIM-1)$-dimensional side $\\S\\in\\partial\\P$:\n \\begin{equation}\\label{eq:dofs:01}\n \\frac{1}{\\ABS{\\S}}\\int_{\\S}\\vs_{\\hh}\\,m\\, {\\rm d}\\S,\n \\,\\,\\forallm\\in\\calM{k-1}(\\S),\\,\n \\forall\\S\\in\\partial\\P;\n \\end{equation}\n \n\\item[]\\textbf{(D2)} the moments of $\\vs_{\\hh}$ of order up to $k-2$ on\n $\\P$:\n \\begin{equation}\\label{eq:dofs:02}\n \\frac{1}{\\ABS{\\P}}\\int_{\\P}\\vs_{\\hh}\\,m\\, {\\rm d}\\P,\n \\quad\\forallm\\in\\calM{k-2}(\\P).\n \\end{equation}\n\\end{description}\nFinally, we introduce the elliptic projection operator\n$\\Pin{k}:{\\widetilde{V}}^{\\hh}_{k}(\\P)\\to\\PS{k}(\\P)$ that for any $\\vs_{\\hh}\\in{\\widetilde{V}}^{\\hh}_{k}(\\P)$ is\ndefined by:\n\\begin{align}\n \\int_{\\P}\\nabla\\Pin{k}\\vs_{\\hh}\\cdot\\nablaq\\, {\\rm d}\\P =\n \\int_{\\P}\\nabla\\vs_{\\hh}\\cdot\\nablaq\\, {\\rm d}\\P\n \\qquad \\text{$\\forall q \\in \\PS{k}(\\P)$}\n\\end{align}\ntogether with the additional conditions:\n\\begin{align}\n \\int_{\\partial\\P}(\\Pin{k}\\vs_{\\hh}-\\vs_{\\hh})\\, {\\rm d} \\S&=0 \\qquad\\textrm{if~}k=1, \\label{eq:def:Pib:1}\\\\[0.5em]\n \\int_{\\P} (\\Pin{k}\\vs_{\\hh}-\\vs_{\\hh})\\,d\\xv&=0 \\qquad\\textrm{if~}k\\geq 2.\\label{eq:def:Pib:k}\n\\end{align}\nAs proved in~\\cite{Cangiani-Manzini-Sutton:2017}, the polynomial\nprojection $\\Pin{k}\\vs_{\\hh}$ is exactly computable using only the values\nfrom the linear functionals \\textbf{(D1)}-\\textbf{(D2)}.\nFurthermore, $\\Pin{k}$ is a polynomial-preserving operator, i.e.,\n$\\Pin{k}q=q$ for every $q\\in\\PS{k}(\\P)$.\n\n\\medskip\nWe are now ready to introduce the \\emph{local nonconforming virtual\n element space of order $k$} on the polytopal element $\\P$, which is\nthe subspace of ${\\widetilde{V}}^{\\hh}_{k}(\\P)$ defined as follow:\n\\begin{equation}\\label{eq:def:Vhk0}\n \\Vhk(\\P) = \\Big\\{\\,\n v\\in{\\widetilde{V}}^{\\hh}_{k}(\\P)\\quad \\textrm{~such~that~} \\quad\n (\\vs_{\\hh}-\\Pin{k}\\vs_{\\hh},m)_{\\P}=0\n \\, \\,\\,\\forallm\\in\\calMs{k-1}(\\P)\\cup\\calMs{k}(\\P)\n \\,\\Big\\}.\n\\end{equation}\nThe space $\\Vhk(\\P)$ has the two important properties that we outline\nbelow:\n\\begin{description}\n\\item[$(i)$] it still contains the space of polynomials of degree at\n most $k$;\n\\item[$(ii)$] the values provided by the set of continuous linear\n functionals \\textbf{(D1)}-\\textbf{(D2)} uniquely determine every\n function $\\vs_{\\hh}$ of $\\Vhk(\\P)$ and can be taken as the \\emph{degrees\n of freedom} of $\\Vhk(\\P)$.\n\\end{description}\n\n\\medskip\nProperty $(i)$ above is a direct consequence of space definitions \n\\eqref{eq:def:Vhkt0} and \\eqref{eq:def:Vhk0}, and guarantees the \noptimal order of approximation, \nwhile property $(ii)$ follows from the unisolvency of the degrees of\nfreedom \\textbf{(D1)}-\\textbf{(D2)} that was proved\nin~\\cite{Ayuso-Lipnikov-Manzini:2016,Cangiani-Manzini-Sutton:2017}).\nAdditionally, the $\\LTWO$-orthogonal projection $\\Piz{k}\\vs_{\\hh}$ is\n\\emph{exactly} computable using only the degrees of freedom of $\\vs_{\\hh}$,\ncf.~\\cite{Cangiani-Manzini-Sutton:2017}, and\n$\\Piz{k}\\vs_{\\hh}=\\Pin{k}\\vs_{\\hh}$ for $k=1,2$ as for the conforming \\text{VEM}{},\ncf.~\\cite{Ahmad-Alsaedi-Brezzi-Marini-Russo:2013}.\n\n\\medskip\nFinally, the \\emph{global nonconforming virtual element space $\\Vhk$\n of order $k\\geq1$ subordinate to the mesh $\\Omega_{\\hh}$} is obtained by\ngluing together the elemental spaces $\\Vhk(\\P)$ to form a subspace of\nthe nonconforming space $H^{1,nc}(\\Omega_{\\hh};k)$.\nThe formal definition reads as:\n\\begin{align}\n \\Vhk:=\\Big\\{\\,\\vs_{\\hh}\\inH^{1,nc}(\\Omega_{\\hh};k)\\,:\\,\\restrict{\\vs_{\\hh}}{\\P}\\in\\Vhk(\\P)\n \\,\\,\\,\\forall\\P\\in\\Omega_{\\hh}\\,\\Big\\}.\n \\label{eq:def:nvem0}\n\\end{align}\nA set of degrees of freedom for $\\Vhk$ is given by collecting the\nvalues from the linear functionals \\textbf{(D1)} for all the mesh\nsides and \\textbf{(D2)} for all the mesh elements.\nThe unisolvence of such degrees of freedom\n\\eqref{eq:dofs:01}-\\eqref{eq:dofs:02} for $\\Vhk$ is an immediate\nconsequence of their unisolvence on each local space $\\Vhk(\\P)$.\nThus, the dimension of $\\Vhk$ is equal to\n$\\NS\\times\\pi_{k-1,\\DIM-1}+\\NP\\times\\pi_{k-2,\\DIM}$, where $\\NS$ is\nthe total number of sides, and $\\NP$ the total number of elemnts, \nand we recall that $\\pi_{\\ell,\\nu}$ is the dimension of\nthe space of polynomials of degree up to $\\ell$ in $\\mathbbm{R}^{\\nu}$.\n\n\\medskip\n\\begin{remark}\n The set of degrees of freedom can be properly redefined by excluding\n the moments on the $N^{\\Sset,\\BND}$ boundary sides, i.e., for\n $\\S\\in\\Sset_{h}^{\\BND}$, which are set to zero to impose the\n homogeneous Dirichlet boundary condition~\\eqref{eq:mod00:B}.\n \n This reduces the dimension of $\\Vhk$ to\n $N^{\\Sset,\\INT}\\times\\pi_{k-1,\\DIM-1}+\\NP\\times\\pi_{k-2,\\DIM}$.\n\\end{remark}\n\\begin{comment}\nTherefore, any function $\\vs_{\\hh}$ in the global space $\\Vhk$ is uniquely\ndetermined by the following set of degrees of freedom:\n\\begin{description}\n\\item[$(i)$] the moments of $\\vs_{\\hh}$ of order up to $k-1$ on each\n $(\\DIM-1)$-dimensional side $\\S\\in\\Sset_{h}$:\n \\begin{equation}\\label{eq:dofs:01g}\n \\frac{1}{\\ABS{\\S}}\\int_{\\S}\\vs_{\\hh}\\,m\\, {\\rm d}\\S,\n \\quad\\forallm\\in\\calM{k-1}(\\S),\n \n \\end{equation}\n\\item[$(ii)$] the moments of $\\vs_{\\hh}$ of order up to $k-1$ on each\n $\\DIM$-dimensional element $\\P\\in\\Omega_{\\hh}$:\n \\begin{equation}\\label{eq:dofs:02g}\n \\frac{1}{\\ABS{\\P}}\\int_{\\P}\\vs_{\\hh}\\,m\\, {\\rm d}\\P,\n \\quad\\forallm\\in\\calM{k-2}(\\P),\n \\,\\,\\forall\\P\\in\\Omega_{\\hh}.\n \\end{equation}\n\\end{description}\n\\end{comment}\n\n\\subsection{Approximation properties}\nBoth for completeness of exposition and future reference in the paper,\nwe briefly summarize the local approximation properties by polynomial\nfunctions and functions in the virtual nonconforming space.\nWe omit here any details about the derivation of these estimate and\nrefer the interested readers to\nReferences~\\cite{%\nCiarlet:1991,%\nBrenner-Scott:2008,%\nBeiraodaVeiga-Brezzi-Cangiani-Manzini-Marini-Russo:2013,%\nAhmad-Alsaedi-Brezzi-Marini-Russo:2013,%\nAyuso-Lipnikov-Manzini:2016,%\n2016stability,%\nBrenner:2017}.\n\n\\medskip\n\\noindent\n\\emph{Local polynomial approximations.}\nOn a given element $\\P\\in\\Omega_{\\hh}$, let\n$v\\in\\HS{s}(\\P)$ with $1 \\leq s\\leq k+1$.\nUnder mesh assumptions~\\ASSUM{A0}{}, there exists a piecewise polynomial approximation $v_{\\pi}$\nthat is of degree $k$ on each element, such that\n\\begin{equation}\\label{eq:approx:pi}\n \\norm{v-v_{\\pi}}{0,\\P} + \\hh_{\\P}\\snorm{v-v_{\\pi}}{1,\\P}\n \\leq C\\hh_{\\P}^{s}\\snorm{v}{s,\\P},\n\\end{equation} \nfor some constant $C>0$ that depends only on the polynomial degree\n$k$ and the mesh regularity constant $\\varrho$.\n\n\\medskip\n\\noindent\nAn instance of such a local polynomial approximation is provided by\nthe $\\LTWO$-projection $\\Piz{k}v$ onto the local polynomial space\n$\\PS{k}(\\P)$, which satisfies the (optimal) error bound \\eqref{eq:approx:pi}.\n\n\\medskip\n\\noindent\nFurthermore, consider the internal side $\\S\\in\\Sset_{h}^{\\INT}$\nand let $\\P^{\\pm}_{\\S}$ be the two elements sharing $\\S$, so that\n$\\S \\subset \\partial\\P^{+}_{\\S}\\cap\\partial\\P^{-}_{\\S}$.\nLet $\\Omega_{\\S}=\\P^{+}_{\\S}\\cup\\P^{-}_{\\S}$.\nThen, the following trace inequality~\\cite{Brenner-Scott:2008} holds \nfor every $v\\in\\HS{1}(\\Omega_\\S)$\n\\begin{equation}\\label{eq:trace:inequality}\n \\norm{v}{0,\\S} \n \\leq C\\hh_{\\S}^{-\\frac{1}{2}}\\norm{v}{0,\\Omega_{\\S}} \n + C\\hh_{\\S}^{\\frac{1}{2}}\\snorm{v}{1,\\Omega_{\\S}}.\n\\end{equation}\nMoreover, for every $v\\in\\HS{s}(\\Omega_{\\hh})$ with $1\\leq s\\leq k+1$, \ncombining the $\\LTWO$-projection approximation property with \\eqref{eq:trace:inequality}\nthe following useful error\nestimate holds \\cite{DiPietro-Ern:2012}\n\\begin{equation}\\label{eq:approx:pitrace}\n \\norm{v-\\PiSz{k}v}{0,\\S} \n \n \\leq C\\hh_{\\S}^{s- \\frac{1}{2}}\\snorm{v}{s,\\Omega_{\\S}} \\,.\n\\end{equation}\nThe same estimate holds also for the boundary sides\n$\\S\\in\\Sset_{h}^{\\BND}$ by taking $\\Omega_{\\S}=\\P$, the element to\nwhich $\\S$ belongs.\n\n\\medskip\n\\noindent\n\\emph{Interpolation error.}\nSimilarly, under mesh regularity assumptions~\\ASSUM{A0}{}, we can\ndefine an interpolation operator in $\\Vhk$ having optimal\napproximation properties.\nTherefore, for every $v\\in\\HS{s}(\\P)$ with $1\\leq s\\leq k+1$ we can\nfind the local interpolate $\\vs^{\\INTP}\\in\\Vhk(\\P)$ such that\n\\begin{align}\\label{eq:interp:00}\n \\norm{v-\\vs^{\\INTP}}{0,\\P} + \\hh_{\\P}\\snorm{v-\\vs^{\\INTP}}{1,\\P}\\leq C\\hh_{\\P}^{s}\\snorm{v}{s,\\P},\n\\end{align}\nwhere $C>0$ is a positive constant independent of $h$.\n\n\n\\section{Virtual element discretization}\n\\label{sc:approx}\n\nThis section briefly reviews the nonconforming virtual element\ndiscretization of the source problem and its extension to the\neigenvalue problem.\n\n\\subsection{The virtual element discretization of the source problem}\nThe goal of the present section is to introduce the virtual element\ndiscretization of the source problem \\eqref{eq:sourcePbm}. \nAccording to~\\cite{Ayuso-Lipnikov-Manzini:2016,Gardini-Vacca:2017}, we\ndefine a suitable discrete bilinear form $\\as_{\\hh}(\\cdot,\\cdot)$\napproximating the continuous gradient-gradient form $a(\\cdot,\n\\cdot)$, whereas for what concerns the bilinear form $b(\\cdot,\n\\cdot)$ we propose two possible discretizations, hereafter denoted by\n$\\bs_{\\hh}(\\cdot,\\cdot)$ and $\\bsht(\\cdot,\\cdot)$.\n\nThe discrete bilinear form $\\as_{\\hh}(\\cdot,\\cdot)$ is the sum of elemental\ncontributions\n\\begin{align}\n \\label{eq:a:global}\n \\as_{\\hh}(\\us_{\\hh},\\vs_{\\hh})=\\sum_{\\P\\in\\Omega_{\\hh}}\\as^P_{\\hh}(\\us_{\\hh},\\vs_{\\hh}),\n\\end{align}\nwhere\n\\begin{align}\n \\label{eq:discreteforms}\n \\as^P_{\\hh}(\\us_{\\hh},\\vs_{\\hh}) = \\as^{\\P} (\\Pin{k}\\us_{\\hh},\\Pin{k}\\vs_{\\hh}) \n + \\SP\\Big( (I-\\Pin{k})\\us_{\\hh}, (I-\\Pin{k})\\vs_{\\hh} \\Big),\n\\end{align}\nand $\\SP(\\cdot,\\cdot)$ denotes any symmetric positive definite\nbilinear form on the element $\\P$ for which there exist two\npositive uniform constants $c_*$ and $c^*$ such that\n\\begin{align}\n c_*\\as^{\\P}(\\vs_{\\hh},\\vs_{\\hh})\n \\leq\\SP(\\vs_{\\hh},\\vs_{\\hh})\n \\leqc^*\\as^{\\P}(\\vs_{\\hh},\\vs_{\\hh})\n \\quad\\forall\\vs_{\\hh}\\in\\Vhk(\\P)\\cap\n \n \\ker(\\Pin{k}) \\,.\n\\end{align}\nThis requirement implies that $\\SP(\\cdot,\\cdot)$ scales like\n$\\as^{\\P}(\\cdot,\\cdot)$, namely $\\SP(\\cdot,\\cdot)\\simeq\\hh_{\\P}^{d-2}$.\n\n\\medskip\nThe choice of the discrete form $\\as_{\\hh}(\\cdot,\\cdot)$ is driven by the\nneed to satisfy the following properties:\n\\begin{description}\n\\item[-] {\\emph{$k$-consistency}}: for all $\\vs_{\\hh}\\in\\Vhk(\\P)$ and for all\n $q\\in\\PS{k}(\\P)$ it holds\n \\begin{align}\n \\label{eq:k-consistency}\n \\as^P_{\\hh}(\\vs_{\\hh},q) = \\as^{\\P}(\\vs_{\\hh},q);\n \\end{align}\n\\item[-] {\\emph{stability}}: there exists two positive constants\n $\\alpha_*,\\,\\alpha^*$, independent of $h$ and of $\\P$, such that\n \\begin{align}\n \\label{eq:stability}\n \\alpha_*\\as^{\\P}(\\vs_{\\hh},\\vs_{\\hh})\n \\leq\\as^P_{\\hh}(\\vs_{\\hh},\\vs_{\\hh})\n \\leq\\alpha^*\\as^{\\P}(\\vs_{\\hh},\\vs_{\\hh})\\quad\\forall\\vs_{\\hh}\\in\\Vhk(\\P) \\,.\n \\end{align}\n\\end{description} \nIn particular, the first term in~\\eqref{eq:discreteforms} ensures the\n$k$-consistency of the method and the second one its stability,\ncf.~\\cite{Ayuso-Lipnikov-Manzini:2016}.\n\nFor what concerns the right hand side $b(\\cdot,\\cdot)$,\nfollowing the setting in \\cite{Gardini-Vacca:2017}, we introduce two\npossible approximated bilinear forms.\n\n\\medskip\n\\noindent\n\\emph{Non--stabilized bilinear form.} In the first choice, we\nconsider the bilinear form $\\bs_{\\hh}(\\cdot,\\cdot)$, which satisfies the\n$k$-consistency property but not the stability property \n(extending to $\\bs_{\\hh}(\\cdot,\\cdot)$ the definitions in \n\\eqref{eq:k-consistency} and \\eqref{eq:stability}).\nLet us split the right-hand side $\\bs_{\\hh}(\\cdot,\\cdot)$ with the sum of\nlocal contributions:\n\\begin{align}\n \\label{eq:bscal:globdef}\n \\bs_{\\hh}(f,\\vs_{\\hh}) = \\sum_{P\\in\\Omega_{\\hh}}\\bs^P_{\\hh}(f,\\vs_{\\hh}).\n\\end{align}\nThen, we define \n\\begin{align}\n \\label{eq:bscal:locdef}\n \\bs^P_{\\hh}(f,\\vs_{\\hh}) = \\bs^{\\P}(\\Piz{k} f, \\vs_{\\hh}) \\qquad \\forall \\vs_{\\hh} \\in \\Vhk(\\P).\n\\end{align}\nWe observe that each local term is fully computable for any functions\n$f\\in\\LTWO(\\Omega)$ and $\\vs_{\\hh}$ in $\\Vhk$ since\n\\begin{equation*}\n \\bs_{\\hh}(f,\\vs_{\\hh}) \n = \\sum_{\\P\\in\\Omega_{\\hh}}\\int_{\\P}(\\Piz{k}f)\\vs_{\\hh}\\, {\\rm d}\\P\n = \\sum_{\\P\\in\\Omega_{\\hh}}\\int_{\\P}f (\\Piz{k}\\vs_{\\hh})\\, {\\rm d}\\P,\n\\end{equation*}\nand \n$\\Piz{k}\\vs_{\\hh}$ is computable (exactly) from the degrees of\nfreedom of $\\vs_{\\hh}$ (cf. Property $(i)$ ).\nMoreover, by definition of $L^2$-projection $\\Piz{k}$, it is straightforward to check that\n\\begin{align}\n \\label{eq:bscal:locdef:equivalence}\n \\bs^P_{\\hh}(f,\\vs_{\\hh}) \n \n = \\bs^{\\P}( \\Piz{k} f, \\Piz{k}\\vs_{\\hh})\\qquad \\forall \\vs_{\\hh} \\in \\Vhk(\\P).\n\\end{align}\nWe estimate the approximation error of the right-hand side by using\nthe Cauchy-Schwarz inequality twice and estimate~\\eqref{eq:approx:pi},\nafter noting that $\\restrict{\\vs_{\\hh}}{\\P}\\in\\HONE(\\P)$ and $\\Piz{0}\\vs_{\\hh}$\nis orthogonal to $f-\\Piz{k}f$:\n\\begin{equation}\n\\label{eq:bs-bsh}\n\\begin{split}\n \\abs{b(f,\\vs_{\\hh}) - \\bs_{\\hh}(f, \\vs_{\\hh})} \n &\\leq\\sum_{\\P\\in\\Omega_{\\hh}}\\big|\\bs^{\\P}(f-\\Piz{k}f,\\vs_{\\hh})\\big| \n =\\sum_{\\P\\in\\Omega_{\\hh}}\\big|\\bs^{\\P}((I-\\Piz{k})f,(I-\\Piz{0})\\vs_{\\hh})\\big| \\\\\n & \\leq\n \\sum_{\\P\\in\\Omega_{\\hh}}\\norm{(I-\\Piz{k})f}{0,P}\\norm{(I-\\Piz{0})\\vs_{\\hh}}{0,P}\n \\\\[0.5em]\n &\\leq\\sum_{\\P\\in\\Omega_{\\hh}}C\\hh_{\\P}\\norm{(I-\\Piz{k})f}{0,P}\\snorm{\\vs_{\\hh}}{1,P}\n \\leq Ch\\norm{(I-\\Piz{k})f}{0}\\snorm{\\vs_{\\hh}}{1,h}.\n\\end{split}\n\\end{equation}\nIn the light of \\eqref{eq:a:global} and \\eqref{eq:bscal:globdef}, the virtual element discretization of source\nproblem~\\eqref{eq:sourcePbm} reads\nas\n: \\emph{Find $\\us_{\\hh}^{s}\\in\\Vhk$ such\nthat}\n\\begin{equation}\n \\label{eq:discreteSource}\n \\as_{\\hh}(\\us_{\\hh}^{s},\\vs_{\\hh}) = \\bs_{\\hh}(f,\\vs_{\\hh}) \\quad\\forall\\vs_{\\hh}\\in\\Vhk \\,.\n\\end{equation}\n\n\\medskip\n\\noindent\n\\emph{Stabilized bilinear form.} The second approximation of the\nbilinear form in \\eqref{eq:exact:b} is inspired by the definition of\nthe virtual bilinear form $\\as_{\\hh}$. \nIn order to distinguish the two formulations, we denote the stabilized\nbilinear form by $\\bsht(\\cdot,\\cdot)$.\nAs usual, we first decompose $\\bsht(\\cdot,\\cdot)$ into the sum of\nlocal contributions:\n\\begin{align}\n\\label{eq:btilde:globdef}\n \\bsht(f,\\vs_{\\hh}) = \\sum_{P\\in\\Omega_{\\hh}}\\widetilde{\\bs}^{\\P}_{\\hh}(f,\\vs_{\\hh}) \\, ,\n\\end{align}\nand, then, we define\n\\begin{equation}\n \\label{eq:discretebstab}\n \\widetilde{\\bs}^{\\P}_{\\hh}(f,\\vs_{\\hh}) \n = \\bs^{\\P}(\\Piz{k} f, \\Piz{k}\\vs_{\\hh}) \n + \\widetilde{S}^{\\P}\\Big( (I-\\Piz{k})f, (I-\\Piz{k})\\vs_{\\hh} \\Big),\n\\end{equation}\nwhere $\\widetilde{S}^{\\P}$ is any positive definite bilinear form on the element\n$\\P$ such that there exist two uniform positive constants $\\widetilde{c}_*$ and\n$\\widetilde{c}^*$ such that\n\\begin{align}\n \\label{eq:ctbot}\n \\widetilde{c}_*\\bs^{\\P}(v,v)\\leq\\widetilde{S}^{\\P}(v,v)\\leq\\cttopb(v,v)\n \\quad\\forallv\\in\\Vhk(\\P)\\cap\\KER(\\Piz{k}).\n\\end{align}\nWe notice that the bilinear form $\\widetilde{\\bs}^{\\P}_{\\hh}$ defined above satisfies both\nthe $k$-consistency and the stability property.\n\\medskip\n\\begin{remark}\n \\label{eq:scale2}\n In analogy with the condition on the form $\\SP(\\cdot,\\cdot)$, we\n require that the form $\\widetilde{S}^{\\P}(\\cdot,\\cdot)$ scales like\n $\\bs^{\\P}(\\cdot,\\cdot)$, that is $\\widetilde{S}^{\\P}(\\cdot,\\cdot)\\simeqh^{\\DIM}$.\n\\end{remark}\n\nBy definition \\eqref{eq:discretebstab} and from inequalities in\n\\eqref{eq:ctbot}, using similar computations as in \\eqref{eq:bs-bsh},\nfor all $f \\in \\LTWO(\\P)$ and for all $v \\in \\Vhk(\\P)$ it holds\nthat\n\\begin{equation}\n\\label{eq:bs-bsht}\n\\begin{split}\n \\abs{b(f,\\vs_{\\hh}) - \\bsht(f, \\vs_{\\hh})} \n &\\leq\\sum_{\\P\\in\\Omega_{\\hh}} \\left( \\big|\\bs^{\\P}(f-\\Piz{k}f,\\vs_{\\hh})\\big| + \n \\widetilde{S}^{\\P}\\Big( (I-\\Piz{k})f, (I-\\Piz{k})\\vs_{\\hh} \\Big)\\right) \\\\\n &\\leq\\sum_{\\P\\in\\Omega_{\\hh}} \\left( \\big|\\bs^{\\P}((I-\\Piz{k})f,(I-\\Piz{0})\\vs_{\\hh})\\big| +\n \\widetilde{c}^* \\norm{(I-\\Piz{k})f}{0} \\norm{(I-\\Piz{k})\\vs_{\\hh}}{0} \n \\right)\\\\\n & \\leq\n \\sum_{\\P\\in\\Omega_{\\hh}}(1 + \\widetilde{c}^*)\\norm{(I-\\Piz{k})f}{0,P}\\norm{(I-\\Piz{0})\\vs_{\\hh}}{0,P}\n \\\\[0.5em]\n &\\leq\\sum_{\\P\\in\\Omega_{\\hh}}C\\hh_{\\P}\\norm{(I-\\Piz{k})f}{0,P}\\snorm{\\vs_{\\hh}}{1,P}\n \\leq Ch \\, \\sum_{\\P\\in\\Omega_{\\hh}}\\norm{(I-\\Piz{k})f}{0} \\, \\snorm{\\vs_{\\hh}}{1,h} \\,.\n\\end{split}\n\\end{equation}\nFrom \\eqref{eq:a:global} and \\eqref{eq:btilde:globdef}, the second formulation reads as:\n\\emph{Find $\\widetilde{\\us}_{\\hh}^{s}\\in\\Vhk$ such that}\n\\begin{align}\n \\label{eq:discreteSource2}\n \\as_{\\hh}(\\widetilde{\\us}_{\\hh}^{s},\\vs_{\\hh}) = \\bsht(f,\\vs_{\\hh})\\quad\\forall\\vs_{\\hh}\\in\\Vhk.\n\\end{align}\nThe well-posedness of the discrete source problems \\eqref{eq:discreteSource} and ~\\eqref{eq:discreteSource2} stem from the the coercivity and the continuity of \nthe bilinear form $a_h(\\cdot,\\cdot)$ (cf. \\eqref{eq:stability}), and from the continuity of \nthe discrete forms $\\bs_{\\hh}(\\cdot,\\cdot)$ and $\\bsht(\\cdot,\\cdot)$.\n\n\\medskip\n\\medskip\n\\noindent \n\nFollowing the same arguments as in~\\cite{Ayuso-Lipnikov-Manzini:2016} but with $\\Piz{k}f$ for $k\\geq1$ instead of $\\Piz{k-2}f$, we\nderive the error estimates in the $\\HONE$-norm and $\\LTWO$-norm that\nwe summarize in the following theorem for next reference in the paper.\n\nWe can state the following optimal error estimates between the solution of the continuous and discrete source problems \\eqref{eq:discreteSource} and \\eqref{eq:discreteSource2}.\n\n\\begin{theorem}\\label{thm:aprioriestimate}\n Under the assumptions \\ASSUM{A0},\n let $\\uss\\in\\HS{1+r}(\\Omega)$ with regularity index $r\\geq 1$ be the\n solution to \\eqref{eq:sourcePbm} with $f\\in\\LTWO(\\Omega)$. \\\\\n \n Let $\\us_{\\hh}^{s}$ and $\\widetilde{\\us}_{\\hh}^{s}\\in\\Vhk$ be respectively the solutions to the\n nonconforming virtual element method~\\eqref{eq:discreteSource} and\n ~\\eqref{eq:discreteSource2}.\n \n \n \n \n \n \n \n \n Let $\\vs_{\\hh} \\in \\{\\us_{\\hh}^{s}, \\, \\widetilde{\\us}_{\\hh}^{s}\\}$, then we have the following error estimates\\medskip\t \n \n \n \n \\begin{itemize}\n \\item $\\HONE$-error estimate:\n \\begin{align*}\n \n \\snorm{ \\uss-\\vs_{\\hh} }{1,h} &\\leq C \\left( \n h^{t}\\snorm{\\uss}{1+r} \n + h\\norm{(I-\\Piz{k})f}{0}\n \\right) \n \\end{align*}\n\n \\medskip\n \\item $\\LTWO$-error estimate (for a convex $\\Omega$):\n \\begin{align*}\n \\norm{ \\uss-\\vs_{\\hh} }{0} &\\leq \n C \\left( h^{t+1}\n \\snorm{\\uss}{1+r}\n + h^2\\norm{(I-\\Piz{k})f}{0}\n \\right) \n \\end{align*}\n \\end{itemize}\n with $t=min(k,r)$.\n\\end{theorem}\n\\medskip The proofs of these estimates are omitted as they are almost\nidentical to those\nof~\\cite[Theorem~4.3,Theorem~4.5]{Ayuso-Lipnikov-Manzini:2016}. \nThe only difference is that here the forcing term $f$ is\napproximated in each element by the orthogonal projection onto\npolynomials of degree $k$ instead of $\\overline{k}=max(k-2,0)$ (see\nestimates \\eqref{eq:bs-bsh} and \\eqref{eq:bs-bsht}).\nNote, indeed, that in the $\\LTWO$-estimate\nof~\\cite[Theorem~4.5]{Ayuso-Lipnikov-Manzini:2016}, the term that\ndepends on the approximation of $f$ is given by\n\\begin{align*}\n (h^2+h^{\\overline{k}+1})\\norm{(I-\\Piz{k})f}{0}\n =\n \\begin{cases}\n (h^2 + h)\\norm{(I-\\Piz{k})f}{0} & \\textrm{for~}k=1,2,\\\\[0.5em]\n h^2\\norm{(I-\\Piz{k})f}{0} & \\textrm{for~}k\\geq3.\n \\end{cases}\n\\end{align*}\nThe difference in the coefficients is a consequence of the\northogonality of $(I-\\Piz{k})f$ to the polynomials of degree $k$ instead\nof $\\overline{k}$.\n\n\\medskip\n\n\n\\begin{remark}\n\\label{rm:estimate-regular-load}\n We observe that if the load term $f$ is an eigenfunction of problem~\\eqref{eq:eigPbm}, \n then $f$ solves the continuous source problem~\\eqref{eq:sourcePbm} with\n datum $\\lambda f$ and thus, thanks to the regularity\n result~\\eqref{eq:regularity}, it belongs to $\\HS{1+r}(\\Omega)$ and\n $\\snorm{f}{1+r} \\le C\\norm{f}{0}$. Then, the a priori\n error estimates in Theorem ~\\ref{thm:aprioriestimate} reduce to\\medskip\n \\begin{itemize}\n \\item $\\HONE$-error estimate:\n \n \\begin{align*}\n \\snorm{ \\uss-\\vs_{\\hh} }{1,h} &\\leq C \\left( \n h^{t}\\snorm{\\uss}{1+r} \n + h\\norm{(I-\\Piz{k})f}{0}\n \\right) \n \\le h^{t}\\snorm{\\uss}{1+r} \\leq C h^{t}\\norm{f}{0} \\leq C h^{t},\n \\end{align*}\n \n \\item $\\LTWO$-error estimate:\n \\begin{align*}\n \\norm{ \\uss - \\vs_{\\hh} }{0} &\\leq \n C \\left( h^{t+1}\n \\snorm{\\uss}{1+r}\n + h^2\\norm{(I-\\Piz{k})f}{0} \\right) \n \\leq C h^{t+1} \\snorm{u^s}{1+r}\n \\leq C h^{t+1}\\norm{f}{0} \\leq C h^{t+1},\n \\end{align*}\n \\end{itemize}\n \n since\n \\begin{equation*}\n \\norm{(I-\\Piz{k})f}{0}\\leq C h^{\\min\\{k+1,1+r\\}}\\snorm{f}{1+r} \\leq C h^{t+1}\\norm{f}{0}.\n \\end{equation*}\n \n\\end{remark}\n\n\n\\begin{remark}\n \n In~\\cite{Ayuso-Lipnikov-Manzini:2016} it is proved that the discrete\n problem \\eqref{eq:discreteSource} is well-posed by taking, \n in the definition of the discrete bilinear form $\\bs_{\\hh}(\\cdot,\\cdot)$, \n instead of $\\Piz{k}f$,\n $\\Piz{k-2}f$ for $k\\geq 2$ and a first-order\n approximation of $\\Piz{0}f$ for $k=1$.\n However, in the definition of the discrete bilinear forms $\\bs_{\\hh}(\\cdot,\\cdot)$\n and $\\bsht(\\cdot,\\cdot)$, we project onto the space $\\PS{k}(\\P)$. \n This choice does not provide a better convergence rate, due\n to the $k$-consistency property, but it has been numerically observed that it gives \n more accurate results.\n \n\n\\end{remark}\n\n\n\\subsection{The virtual element discretization of the eigenvalue\n problem.}\nIn the light of the nonconforming virtual element\nmethod~\\eqref{eq:discreteSource} and ~\\eqref{eq:discreteSource2}\nintroduced in the previous section,\nfollowing~\\cite{Gardini-Vacca:2017}, we consider two different\ndiscretizations of the eigenvalue problem~\\eqref{eq:eigPbm}.\nThe first formulation in inspired to the source problem \\eqref{eq:discreteSource},\nand uses definition~\\eqref{eq:bscal:globdef}. We formulate the virtual\nelement approximation of~\\eqref{eq:eigPbm} as:\n\\emph{Find $(\\lambda_h,\\us_{\\hh})\\in\\mathbbm{R}\\times\\Vhk$, $\\NORM{\\us_{\\hh}}{0}=1$, such that}\n\\begin{align}\n \\label{eq:discreteEigPbm}\n \\as_{\\hh}(\\us_{\\hh},\\vs_{\\hh}) = \\lambda_{h}\\bs_{\\hh}(\\us_{\\hh},\\vs_{\\hh})\\quad\\forall\\vs_{\\hh}\\in\\Vhk.\n\\end{align}\n\n\\medskip\n\\noindent\n\nThe second formulation is inspired to the virtual source problem \\eqref{eq:discreteSource2} and\nuses definition~\\eqref{eq:btilde:globdef}. We formulate the second\napproximation of problem~\\eqref{eq:eigPbm} as:\n\\emph{Find $(\\widetilde{\\lambda}_{h},\\widetilde{\\us}_{\\hh})\\in\\mathbbm{R}\\times\\Vhk$,\n $\\NORM{\\widetilde{\\us}_{\\hh}}{0} = 1$, such that}\n\\begin{align}\\label{eq:discreteEigPbm2}\n \\as_{\\hh}(\\widetilde{\\us}_{\\hh},\\vs_{\\hh}) = \\widetilde{\\lambda}_{h}\\bsht(\\widetilde{\\us}_{\\hh},\\vs_{\\hh})\\quad\\forall\\vs_{\\hh}\\in\\Vhk.\n\\end{align}\n\n\n\n\n\n\\section{Spectral approximation for compact operators}\n\\label{sc:compact}\n\nIn this section, we briefly recall some spectral approximation results\nthat can be deduced from~\\cite{Babuska-Osborn:1991,Boffi:2010,Kato:1976}. \nFor more general results, we refer to the original papers.\nBefore stating the spectral approximation results, we introduce a\nnatural compact operator associated with problem~\\eqref{eq:eigPbm} and\nits discrete counterpart, and we recall their connection with the\neigenmode convergence.\n\n\\medskip\nWe associate problem~\\eqref{eq:eigPbm} with its\nsolution operator $T\\in\\mathcal{L}(\\LTWO(\\Omega))$, which is the \nbounded linear operator $T:\\LTWO(\\Omega)\\rightarrow\\LTWO(\\Omega)$\nmapping the forcing term $f$ to $\\uss=:Tf$:\n\\begin{align*}\n \\label{eq:T}\n \\left\\{\n \\begin{array}{l}\n Tf\\in\\VS{}\\textrm{~such~that}\\\\\n a(Tf,v) = b(f,v)\\quad\\forallv\\in\\VS{}.\n \\end{array}\n \\right.\n\\end{align*}\nOperator $T$ is self-adjoint and positive definite with respect to the\ninner products $a(\\cdot, \\cdot)$ and $b(\\cdot, \\cdot)$ on $\\VS{}$, and compact due to the\ncompact embedding of $\\HONE(\\Omega)$ in $\\LTWO(\\Omega)$.\n\n\\medskip\nSimilarly, we associate problem~\\eqref{eq:discreteSource} with its\nsolution operator $T_{h}\\in\\mathcal{L}(\\LTWO(\\Omega))$ and\nproblem~\\eqref{eq:discreteSource2} with its solution operator\n$\\widetilde{T}_{h}\\in\\mathcal{L}(\\LTWO(\\Omega))$.\nThe former is the bounded linear operator mapping the forcing term\n$f$ to $\\us_{\\hh}^{s}=:T_{h}f$ and satisfies:\n\\begin{equation*}\n \\left\\{\n \\begin{array}{l}\n \\label{eq:Th}\n T_{h}f\\in\\Vhk\\textrm{~such~that}\\\\\n \\as_{\\hh}(T_{h}f,\\vs_{\\hh}) = \\bs_{\\hh}(f,\\vs_{\\hh})\\quad\\forall\\vs_{\\hh}\\in\\Vhk.\n \\end{array}\n \\right.\n\\end{equation*}\nThe latter is the bounded linear operator mapping the forcing term\n$f$ to $\\widetilde{\\us}_{\\hh}^{s}=:\\widetilde{T}_{h}f$ and satisfies:\n\\begin{align*}\n \\left\\{\n \\begin{array}{l}\n \\label{eq:Thtilde}\n \\widetilde{T}_{h}f\\in\\Vhk\\textrm{~such~that}\\\\\n \\as_{\\hh}(\\widetilde{T}_{h}f,\\vs_{\\hh}) = \\bsht(f,\\vs_{\\hh})\\quad\\forall\\vs_{\\hh}\\in\\Vhk.\n \\end{array}\n \\right.\n\\end{align*}\n\nBoth operators $T_{h}$ and $\\widetilde{T}_{h}$ are self-adjoint\nand positive definite with respect to the inner products\n$\\as_{\\hh}(\\cdot, \\cdot)$, $\\bs_{\\hh}(\\cdot, \\cdot)$ and $\\as_{\\hh}(\\cdot, \\cdot)$, $\\bsht(\\cdot, \\cdot)$.\nThey are also compact since their ranges are finite dimensional.\n\n\\medskip\n\\noindent\nThe eigensolutions of the continuous problem~\\eqref{eq:eigPbm} and the\ndiscrete problems ~\\eqref{eq:discreteEigPbm}\nand~\\eqref{eq:discreteEigPbm2} are respectively related to the\neigenmodes of the operators $T$, $T_{h}$, and $\\widetilde{T}_{h}$.\nIn particular, $(\\lambda,u)$ is an eigenpair of\nproblem~\\eqref{eq:eigPbm} if and only if $Tu=(1\\slash{\\lambda})u$,\ni.e. $(\\frac{1}{\\lambda}, u)$ is an eigenpair for the operator $T$,\nand analogously for problems ~\\eqref{eq:discreteEigPbm}\nand~\\eqref{eq:discreteEigPbm2} and operators $T_{h}$ and\n$\\widetilde{T}_{h}$.\nBy virtue of this correspondence, the convergence analysis can be\nderived from the spectral approximation theory for compact operators.\nIn the rest of this section we refer only to operators $T$ and\n$\\widetilde{T}_{h}$.\nIdentical considerations hold for operators $T$ and $T_{h}$\nand we omit them for brevity.\n\n\\medskip\nA sufficient condition for the correct spectral approximation of a\ncompact operator $T$ is the uniform convergence to $T$ of the family\nof discrete operators $\\{\\widetilde{T}_{h}\\}_{h}$ (see~\\cite[Proposition\n7.4]{Boffi:2010}, cf. also~\\cite{Babuska-Osborn:1991}):\n\\begin{equation}\n \\label{eq:unifconv}\n \\norm{T-\\widetilde{T}_{h}}{\\mathcal{L}(\\LTWO(\\Omega))}\\to 0, \\quad\\textrm{as~}h\\to 0,\n\\end{equation}\nor, equivalently, \n\\begin{equation}\n \\label{eq:unifconv2}\n \\norm{(T-\\widetilde{T}_{h})f}{0}\\leq C\\rho(h)\\norm{f}{0}\n \\quad\\forallf\\in\\LTWO(\\Omega),\n\\end{equation}\nwith $\\rho(h)$ tending to zero as $h$ goes to zero. \nCondition\\eqref{eq:unifconv2} usually follows by a-priori estimates\nwith no additional regularity assumption on $f$. \nBesides the convergence of the eigenmodes,\ncondition~\\eqref{eq:unifconv}, or the equivalent\ncondition~\\eqref{eq:unifconv2}, implies that no spurious eigenvalues\nmay pollute the spectrum.\nIn fact, \n\\renewcommand{\\theenumi}{\\roman{enumi}}\n\\renewcommand{\\labelenumi}{(\\theenumi)}\n\\begin{enumerate}\n\\item each continuous eigenvalue is approximated by a number of\n discrete eigenvalues (counted with their multiplicity) that\n corresponds exactly to its multiplicity;\n\\item each discrete eigenvalue approximates a continuous eigenvalue.\n\\end{enumerate}\n\nCondition~\\eqref{eq:unifconv} does not provide any indication on the\napproximation rate. \nIt is common to split the convergence analysis for\neigenvalue problems into two steps: first, the convergence and the\nabsence of spurious modes is studied; then, suitable convergence\nrates are proved.\nWe now report the main results about the spectral approximation for compact operators.\n(cf.~\\cite[Theorems 7.1--7.4]{Babuska-Osborn:1991}; see\nalso~\\cite[Theorem 9.3--9.7]{Boffi:2010}),\nwhich deal with the order of\nconvergence of eigenvalues and eigenfunctions.\n\\begin{theorem}\n \\label{thm:BOeigfun}\n Let the uniform convergence~\\eqref{eq:unifconv} holds true. \n \n Let $\\mu$ be an eigenvalue of $T$, with multiplicity $m$, and denote \n the corresponding eigenspace by $E_{\\mu}$. \n Then, exactly $m$ discrete eigenvalues $\\widetilde{\\mu}_{1, h}, \\dots, \\widetilde{\\mu}_{m, h}$\n (repeated according to their respective multiplicities) converges to $\\mu$. \n \n Moreover, let $\\widetilde{E}_{\\mu,h}$ be the direct sum of the eigenspaces\n corresponding to the discrete eigenvalues\n $\\widetilde{\\mu}_{1,h},\\cdots,\\widetilde{\\mu}_{m,h}$ converging to $\\mu$.\n \n Then\n \\begin{equation}\n \\delta(E_{\\mu}, \\widetilde{E}_{\\mu,h})\\leq\n C \\norm{(T-\\widetilde{T}_{h})_{|E_{\\mu}}}{\\mathcal{L}(\\LTWO(\\Omega))},\n \\label{eq:eigenmodeRate}\n \\end{equation}\n with \n \\begin{align*}\n \\delta(E_{\\mu},\\widetilde{E}_{\\mu,h}) = \\max( \\hat\\delta(E_{\\mu},\\widetilde{E}_{\\mu,h}) , \\hat\\delta(\\widetilde{E}_{\\mu,h},E_{\\mu}) ) \n \\end{align*}\n where, in general,\n \\[\n \\hat\\delta(U, \\, W) = \\sup_{u\\in U, \\NORM{u}{0} = 1}\\inf_{w\\in W}\\norm{u-w}{0}\n \\]\n denotes the gap between $U$, $W \\subseteq L^2(\\Omega)$.\n\\end{theorem}\n\nConcerning the eigenvalue approximation error, we recall the following result.\n\n\\begin{theorem}\n \\label{th:BOeig}\n Let the uniform convergence~\\eqref{eq:unifconv} holds true. \n Let $\\phi_1,\\ldots,\\phi_m$ be a basis of the eigenspace $E_{\\mu}$ of\n $T$ corresponding to the eigenvalue $\\mu$.\n \n Then, for $i=1,\\ldots,m$\n \\begin{equation}\n \\label{eq:eigRate}\n \\abs{\\mu - \\widetilde{\\mu}_{i,h}} \\le C \n \\displaystyle\n \\Big(\\,\\sum_{j,k=1}^m \\abs{ b((T-\\widetilde{T}_{h})\\phi_k,\\phi_j) } + \n \\norm{ (T-\\widetilde{T}_{h})_{|E_{\\mu}} }{\\mathcal{L}(\\LTWO(\\Omega))}^2\\,\\Big),\n \\end{equation}\n where $\\widetilde{\\mu}_{1,h},\\ldots,\\widetilde{\\mu}_{m,h}$ are the $m$ discrete\n eigenvalues converging to $\\mu$ repeated according to their\n multiplicities.\n\\end{theorem}\n\n\\section{Convergence analysis and error estimates}\n\\label{sc:analysis}\n\nIn this section we study the convergence of the discrete eigenmodes\nprovided by the VEM approximation to the continuous ones.\nWe will consider the stabilized discrete\nformulation~\\eqref{eq:discreteEigPbm2}. The analysis can be easily applied \nto the non--stabilized one~\\eqref{eq:discreteEigPbm}.\n\n\\subsection{Convergence analysis for the stabilized formulation}\n\\label{subsec:convnonstab}\nIn the case of the first VEM approximation of\nproblem~\\eqref{eq:eigPbm}, which uses the stabilized form\n$\\bsht(\\cdot,\\cdot)$, the uniform convergence of the sequence of\noperators ${\\widetilde{T}_{h}}$ to $T$ directly stems from the $\\LTWO$-\\emph{a\n priori} error estimate of Theorem~\\ref{thm:aprioriestimate}.\n\n\\begin{theorem}\n \\label{thm:thm1}\n The family of operators $\\widetilde{T}_{h}$ associated with\n problem~\\eqref{eq:discreteSource2} converges uniformly to the operator\n $T$ associated with problem~\\eqref{eq:sourcePbm}, that is,\n \\begin{equation}\n \\label{eq:uniformconv}\n \\norm{T-\\widetilde{T}_{h}}{\\mathcal{L}(\\LTWO(\\Omega))}\\to 0\n \\quad\\textrm{for}\\quadh\\to 0.\n \\end{equation}\n\\end{theorem}\n\\begin{proof}\n Let $\\uss$ and $\\widetilde{\\us}_{\\hh}^{s}$ be the solutions to the continuous and the\n discrete source problems~\\eqref{eq:sourcePbm}\n and~\\eqref{eq:discreteSource2} respectively.\n \n The $\\LTWO$-estimate of Theorem~\\ref{thm:aprioriestimate} with\n $f\\in\\LTWO(\\Omega)$ and the stability\n condition~\\eqref{eq:regularity} imply that\n \\begin{align*} \n \\norm{ \\uss-\\widetilde{\\us}_{\\hh}^{s} }{0} \\leq Ch^{\\min(t+1,2)}\\norm{f}{0}\n \\end{align*}\n with $t=\\min(k,r)$, $k\\geq 1$ being the order of the method and $r$\n at least in $(1\\slash{2},1]$ being the regularity index of the\n solution $\\uss\\in\\HS{1+r}(\\Omega)$ to the continuous source problem\n in equation~\\eqref{eq:regularity}.\n \n From this inequality it follows that\n \\begin{align*}\n \\norm{T-\\widetilde{T}_{h}}{\\mathcal{L}(\\LTWO(\\Omega))} \n = \\sup_{f\\in\\LTWO(\\Omega)}\\dfrac{\\norm{Tf-\\widetilde{T}_{h}f}{0}}{\\norm{f}{0}} \n = \\sup_{f\\in\\LTWO(\\Omega)}\\dfrac{\\norm{\\uss-\\widetilde{\\us}_{\\hh}^{s}}{0}}{\\norm{f}{0}}\n \\leq Ch^{\\min(t+1,2)}.\n \\end{align*}\n\\end{proof}\n\\begin{remark}\n \\label{rm:regularity}\n We observe that if $f\\in\\mathcal{E}_{\\mu}$ then, thanks to the $L^2$ a priori error estimate \n in Remark~\\eqref{rm:estimate-regular-load}, it holds \n \\begin{align*}\n \\norm{(T-\\widetilde{T}_{h})_{|E_{\\mu}}}{\\mathcal{L}(\\LTWO(\\Omega))}\n = \\sup_{f\\in\\mathcal{E}_{\\mu}}\\dfrac{\\norm{Tf-\\widetilde{T}_{h}f}{0}}{\\norm{f}{0}} \n = \\sup_{f\\in\\mathcal{E}_{\\mu}}\\dfrac{\\norm{\\uss-\\widetilde{\\us}_{\\hh}^{s}}{0}}{\\norm{f}{0}}\n \\leq Ch^{t+1}.\n \\end{align*}\n \n\\end{remark}\nPutting together Theorem~\\ref{thm:BOeigfun}~ ,Theorem~\\ref{thm:thm1}, and Remark~\\ref{rm:regularity}, we can state the following result. \n \\begin{theorem}\n\n Let $\\mu$ be an eigenvalue of $T$, with multiplicity $m$, and\n denote the corresponding eigenspace by $E_{\\mu}$. \n \n Then, exactly $m$ discrete eigenvalues $\\widetilde{\\mu}_{1, h},\n \\dots, \\widetilde{\\mu}_{m, h}$ (repeated according to their\n respective multiplicities) converges to $\\mu$.\n \n Moreover, let $\\widetilde{E}_{\\mu,h}$ be the direct sum of the eigenspaces\n corresponding to the discrete eigenvalues\n $\\widetilde{\\mu}_{1,h},\\cdots,\\widetilde{\\mu}_{m,h}$ converging to $\\mu$.\n \n Then\n \\begin{equation}\n \\delta(E_{\\mu}, \\widetilde{E}_{\\mu,h})\\leq\n C h^{t+1}.\n \\end{equation}\n \\end{theorem}\n \nA direct consequence of the previous result\n(cf.~\\cite{Babuska-Osborn:1991,Boffi:2010}) is the following one.\n\\begin{theorem}\n \\label{th:convl2}\n Let $u$ be a unit eigenfunction associated with the eigenvalue\n $\\lambda$ of multiplicity $m$ and let $\\widetilde{\\ws}_{\\hh}^{(1)},\\ldots,\\widetilde{\\ws}_{\\hh}^{(m)}$\n denote linearly independent eigenfunctions associated with the $m$\n discrete eigenvalues converging to $\\lambda$. \n \n Then there exists $\\widetilde{\\us}_{\\hh}\\in\\SPAN{\\widetilde{\\ws}_{\\hh}^{(1)},\\ldots,\\widetilde{\\ws}_{\\hh}^{(m)}}$\n such that\n \\begin{align*}\n \\norm{u-\\widetilde{\\us}_{\\hh}}{0}\\leq Ch^{t+1},\n \\end{align*}\n where $t=\\min\\{k,r\\}$, being $k$ the order of the method and $r$ the\n regularity index of $u$.\n\\end{theorem}\n\n\\medskip\nUsing Theorem~\\ref{thm:aprioriestimate} we can obtain an estimate of\nthe conformity error~\\eqref{eq:def:Nh} better than the one in\nLemma~\\eqref{le:Nh} when its argument are the solution to the\ncontinuous problem~\\eqref{eq:sourcePbm} and the discrete\nproblem~\\eqref{eq:discreteSource2}.\nIt is worth noting that the solution to the discrete problem does not\nneed to be an approximation of the solution to the continuous problem, cf.~\\cite{Boffi:2010}. \n\\begin{lemma}\\label{lemma:Nh:def}\n \n Consider $u\\in\\HS{1+r}(\\Omega)$, $v\\in\\LTWO(\\Omega)$ and let\n $Tu,Tv\\in\\HS{1+r}(\\Omega)$ with $r >1/2$ be, \n respectively, the solutions to problem~\\eqref{eq:sourcePbm} \n with load term $u$ and $v$.\n \n Assume that \\ASSUM{A0} is satisfied and let\n $\\widetilde{T}_{h}u\\in\\Vhk\\subsetH^{1,nc}(\\Omega_{\\hh};k)$ for some integer $k\\geq1$\n be the virtual element approximation of $Tu$ that solves\n problem~\\eqref{eq:discreteSource2}.\n \n Then, there exists a constant $C>0$, independent of $h$, \n \n such that\n \\begin{align}\n \\abs{\\mathcal{N}_{\\hh}(Tv, \\widetilde{T}_{h}u)} \n \\leq Ch^{2t}\\,\\snorm{Tv}{1+r}\\,\\snorm{Tu}{1+r}\n \\end{align}\n where $t=\\min\\{k,r\\}$ and $\\mathcal{N}_{\\hh}(u,v)$ is the conformity error\n defined in \\eqref{eq:def:Nh}.\n\\end{lemma}\n\\begin{proof}\n We start the following chain of developments from the definition of\n the conformity error given in~\\eqref{eq:def:Nh}, note that\n $\\jump{Tu}=0$ on every mesh side and that the moments \n up to order $k-1$ of $\\jump{\\widetilde{T}_hu}$ across all\n mesh interfaces are zero, and apply the Cauchy-Schwarz\n inequality in the last two steps:\n \\begin{align*}\n \\mathcal{N}_{\\hh}(Tv,\\widetilde{T}_{h}u) \n & = \\sum_{\\S\\in\\Sset_{h}}\\int_{\\S}\\nabla Tv\\cdot\\jump{\\widetilde{T}_{h}u}\\, {\\rm d}\\S \\\\\n & = \\sum_{\\S\\in\\Sset_{h}}\\int_{\\S}\\big(I-\\PiSz{k-1}\\big)\\nabla Tv\\cdot\\big(I-\\PiSz{0}\\big)\\jump{\\widetilde{T}_{h}u}\\, {\\rm d}\\S \\\\\n & = \\sum_{\\S\\in\\Sset_{h}}\\int_{\\S}\\big(I-\\PiSz{k-1}\\big)\\nabla Tv\\cdot\\Big(\n \\big(I-\\PiSz{0}\\big)\\jump{\\widetilde{T}_{h}u})-\\big(I-\\PiSz{0}\\big)\\jump{Tu}\\Big)\\, {\\rm d}\\S \\\\\n & = \\sum_{\\S\\in\\Sset_{h}}\\int_{\\S}\\big(I-\\PiSz{k-1}\\big)\\nabla Tv\\cdot\\big(I-\\PiSz{0}\\big)\\jump{(\\widetilde{T}_{h}-T)u}\\, {\\rm d}\\S\\\\[0.35em]\n & \\leq\\sum_{\\S\\in\\Sset_{h}}\\norm{\\big(I-\\PiSz{k-1}\\big)\\nabla Tv\\cdot\\norS}{0,\\S}\\,\\norm{(I-\\PiSz{0})\\jump{(\\widetilde{T}_{h}-T)u}\\cdot\\norS}{0,\\S}\\\\\n & \\leq\n \\left[\\,\\sum_{\\S\\in\\Sset_{h}}\\norm{\\big(I-\\PiSz{k-1}\\big)\\nabla Tv\\cdot\\norS}{0,\\S}^2\\right]^{\\frac{1}{2}}\\times\n \\left[\\,\\sum_{\\S\\in\\Sset_{h}}\\norm{(I-\\PiSz{0})\\jump{(\\widetilde{T}_{h}-T)u}\\cdot\\norS}{0,\\S}^2\\right]^{\\frac{1}{2}}\\\\[0.5em]\n &= \\mathcal{N}_1\\times\\mathcal{N}_2.\n \\end{align*}\n \n Trace inequality~\\eqref{eq:approx:pitrace} yields \n\n \\begin{align*}\n \\norm{(I-\\Pi_{k-1}^{0,\\S})\\nabla Tv\\cdot\\norS}{0,\\S}\n \\leq C\\hh_{\\S}^{t-\\frac{1}{2}}\\snorm{Tv}{1+r,\\Omega_{\\S}},\n \\end{align*}\n and summing over all the mesh sides, noting that $\\hh_{\\S}\\leqh$, the\n number of sides per element is uniformly bounded due to \\ASSUM{A0}\n and using definition~\\eqref{eq:norm-broken} yield\n \\begin{align*}\n \\abs{\\mathcal{N}}_1^2\n = \\sum_{\\S\\in\\Sset_{h}}\\norm{\\big(I-\\PiSz{k-1}\\big)\\nabla Tv\\cdot\\norS}{0,\\S}^2\n \\leq C\\big(h^{t-\\frac{1}{2}}\\big)^2\\sum_{\\S\\in\\Sset_{h}}\\snorm{Tv}{1+r,\\Omega_{\\S}}^2\n \\leq C\\big(h^{t-\\frac{1}{2}}\\big)^2\\sum_{\\P\\in\\Omega_{\\hh}}\\snorm{Tv}{1+r,\\P}^2,\n \\end{align*}\n and finally\n \\begin{align*}\n \\abs{\\mathcal{N}_1} \\leq Ch^{t-\\frac{1}{2}}\\snorm{Tv}{1+r}.\n \\end{align*}\n \n Similarly, trace inequality~\\eqref{eq:approx:pitrace} and the jump definition yield \n \n \\begin{align*}\n \\norm{(I-\\PiSz{0})\\jump{(\\widetilde{T}_{h}-T)u}\\cdot\\norS}{0,\\S}\n \\leq C\\hh_{\\S}^{\\frac{1}{2}}\\snorm{(\\widetilde{T}_{h}-T)u}{1,\\Omega_{\\S}}\n \\end{align*}\n and using the same arguments as above we have that\n \\begin{align*}\n \\abs{\\mathcal{N}_2}^2\n &= \\sum_{\\S\\in\\Sset_{h}}\\norm{(I-\\Pi^{0,\\S}_{0})\\jump{(\\widetilde{T}_{h}-T)u}\\cdot\\norS}{0,\\S}^2 \n \\leq Ch\\sum_{\\S\\in\\Sset_{h}}\\snorm{(\\widetilde{T}_{h}-T)u}{1,\\Omega_{\\S}}^2 \n \\leq Ch\\sum_{\\P\\in\\Omega_{\\hh}}\\snorm{(\\widetilde{T}_{h}-T)u}{1,\\P}^2\\\\[0.5em]\n &= Ch\\snorm{(\\widetilde{T}_{h}-T)u}{1,h}^2.\n \\end{align*}\n Using this relation and the \\emph{a priori} error estimate in the\n energy norm of Remark~\\ref{rm:estimate-regular-load} finally yield:\n \\begin{align*}\n \\abs{\\mathcal{N}}_2\\leqh^{t+\\frac{1}{2}}\\snorm{Tu}{1+r}.\n \\end{align*}\n \n The assertion of the lemma follows by collecting the above estimates\n together.\n\\end{proof}\n\nWe now prove the usual double order convergence of the eigenvalues.\n\\begin{theorem}\n \\label{theorem:double:convergence:rate}\n Let $\\lambda$ be an eigenvalue of problem~\\eqref{eq:eigPbm} with multiplicity $m$, \n and denote by $\\widetilde{\\lambda}_{1,h},\\cdots,\\widetilde{\\lambda}_{m,h}$ the $m$ discrete eigenvalues \n converging towards $\\lambda$. \n \n Then the following optimal double order convergence holds:\n \\begin{align}\n \\label{eq:double:convergence:rate}\n \\abs{\\lambda - \\widetilde{\\lambda}_{i,h}}\n \\leq C h^{2t}\\quad\\forall i=1,\\ldots,m, \n \\end{align}\n with $t=\\min\\{k,r\\}$, being $k$ the order of the method and $r$ the\n regularity index of the eigenfunction corresponding to $\\lambda$.\n\\end{theorem}\n\\begin{proof}\n We use the result stated in Theorem~\\ref{th:BOeig}. \n \n It is clear that the second term in the estimate of\n Theorem~\\ref{th:BOeig} is of double order compared to the\n $\\HONE$-rate of convergence. \n \n Hence, we analyse in detail the term\n $b \\left((T-\\widetilde{T}_{h})\\phi_j,\\phi_k\\right)$.\n \n Let $u$ and $v$ be two eigenfunctions associated with the\n eigenvalue $\\lambda$. \n \n Then, we note that $(T-\\widetilde{T}_{h})u\\inH^{1,nc}(\\Omega_{\\hh};k)$ and begin the\n chain of developments that follows from the definition of the\n conformity error in~\\eqref{eq:def:Nh}:\n \\begin{align}\n \\label{eq:eigproof}\n \\begin{array}{rll}\n &b \\left( (T-\\widetilde{T}_{h})u,v \\right) \n \n = a(Tv,(T-\\widetilde{T}_{h})u) - \\mathcal{N}_{h}(Tv,(T-\\widetilde{T}_{h})u) & \\mbox{\\big[\\textrm{note~that~}$\\mathcal{N}_{h}(Tv,Tu)=0$\\big]} \\nonumber\\\\[0.5em]\n &\\qquad = a(Tv,(T-\\widetilde{T}_{h})u) + \\mathcal{N}_{h}(Tv,\\widetilde{T}_{h}u) & \\mbox{\\big[\\textrm{add and subtract~}$\\widetilde{T}_{h}v$ in the first term\\big]}\\nonumber\\\\[0.5em]\n &\\qquad = a((T-\\widetilde{T}_{h})v,(T-\\widetilde{T}_{h})u) + a(\\widetilde{T}_{h}v,(T-\\widetilde{T}_{h})u) + \\mathcal{N}_{h}(Tv,\\widetilde{T}_{h}u) & \\mbox{\\big[split the middle term and use~\\eqref{eq:def:Nh}\\big]} \\nonumber\\\\[0.5em]\n &\\qquad = a((T-\\widetilde{T}_{h})v,(T-\\widetilde{T}_{h})u) + b(\\widetilde{T}_{h}v,u) + \\mathcal{N}_{h}(Tu,\\widetilde{T}_{h}v) & \\nonumber\\\\\n &\\qquad\\phantom{=}- a(\\widetilde{T}_{h}v,\\widetilde{T}_{h}u) + \\mathcal{N}_{h}(Tv,\\widetilde{T}_{h}u) & \\mbox{\\big[add both sides of~\\eqref{eq:discreteSource2} \\big]} \\nonumber\\\\[0.5em]\n &\\qquad = a((T-\\widetilde{T}_{h})v,(T-\\widetilde{T}_{h})u) + \\big[b(u, \\widetilde{T}_{h}v) - \\bsht(u, \\widetilde{T}_{h}v)\\big] + & \\nonumber\\\\\n &\\qquad\\phantom{=}\n \\big[ \\as_{\\hh}(\\widetilde{T}_{h}v,\\widetilde{T}_{h}u) - a(\\widetilde{T}_{h}v,\\widetilde{T}_{h}u) \\big] +\n \\big[ \\mathcal{N}_{h}(Tu,\\widetilde{T}_{h}v) + \\mathcal{N}_{h}(Tv,\\widetilde{T}_{h}u) \\big]\n \\\\[0.5em] &\\qquad \n = \\sum_{i=1}^{4}\\mathsf{R}_i.\n \\end{array}\n \\end{align}\n \n Term $\\mathsf{R}_1$ is clearly of order $h^{2t}$, being $u$ and $v$ eigenfunctions (see Remark~\\eqref{rm:estimate-regular-load}) :\n \\begin{align*}\n \\abs{\\mathsf{R}_1} \n = \\abs{ a((T-\\widetilde{T}_{h})v,(T-\\widetilde{T}_{h})u)}\n \\leq \\snorm{(T-\\widetilde{T}_{h})v}{1,h}\\,\\snorm{(T-\\widetilde{T}_{h})u}{1,h}\n \\leq Ch^{2t}\n \\end{align*}\n To bound term $\\mathsf{R}_2$ using \\eqref{eq:discretebstab} and \\eqref{eq:ctbot}, the definition of $L^2$-orthogonal projection and triangular inequality, we get\n \\begin{equation}\n\\label{eq:R2}\n\\begin{split}\n \\abs{\\mathsf{R}_2} & \\leq \\abs{b(u,\\widetilde{T}_{h}v) - \\bsht(u, \\widetilde{T}_{h}v)} \n \\leq\\sum_{\\P\\in\\Omega_{\\hh}} \\abs{\\bs^{\\P}(u,\\widetilde{T}_{h}v) - \\widetilde{\\bs}^{\\P}_{\\hh}(u, \\widetilde{T}_{h}v)} \\\\\t\n & \\leq\\sum_{\\P\\in\\Omega_{\\hh}} \\left( \\left|\\bs^{\\P}(u-\\Piz{k}u,\\widetilde{T}_{h}v)\\right| + \n \\left| \\widetilde{S}^{\\P}\\left( (I-\\Piz{k})u, (I-\\Piz{k})\\widetilde{T}_{h}v \\right)\\right| \\right) \\\\\n & \\leq\\sum_{\\P\\in\\Omega_{\\hh}} \\left( \\big|\\bs^{\\P}((I-\\Piz{k})u,(I-\\Piz{k})\\widetilde{T}_{h}v)\\big| +\n \\widetilde{c}^* \\norm{(I-\\Piz{k})u}{0} \\norm{(I-\\Piz{k})\\widetilde{T}_{h}v}{0} \n \\right)\\\\\n & \\leq\n \\sum_{\\P\\in\\Omega_{\\hh}}(1 + \\widetilde{c}^*)\\norm{(I-\\Piz{k})u}{0,P}\\norm{(I-\\Piz{k})\\widetilde{T}_{h}v}{0,P}\n \\\\\n &\\leq C \\sum_{\\P\\in\\Omega_{\\hh}}\n \\left(\n \\norm{(I-\\Piz{k})u}{0,P} \\left( \n \\norm{(I-\\Piz{k})(T -\\widetilde{T}_{h})v}{0,P} + \\norm{(I-\\Piz{k})Tv}{0,P} \n \\right) \n \\right)\n \\end{split}\n\\end{equation}\n \n Now, note that\n \\begin{align}\n \\label{eq:r21}\n \\norm{u-\\Piz{k}u}{0,\\P}\n \\leq Ch^{\\min(k+1,r+1)}\\snorm{u}{r+1,\\P}\n \\leq Ch^{t+1}\\snorm{u}{r+1,\\P}.\n \\end{align}\n By the continuity of $L^2$-projection with respect the $L^2$-norm we get\n \n \\begin{align}\n \\label{eq:r22}\n \\norm{(I-\\Piz{k})(T -\\widetilde{T}_{h})v}{0,P}\n \\leq \\norm{(T -\\widetilde{T}_{h})v}{0,P}.\n \n \\end{align}\n \n Moreover polynomial approximation estimate \\eqref{eq:approx:pi} yields\n \n \\begin{align}\n \\label{eq:r23}\n \\norm{(I-\\Piz{k})Tv}{0,P}\n \\leq Ch^{t+1}\\snorm{Tv}{1+r,\\P}.\n \\end{align}\n \n Hence collecting \\eqref{eq:r21}, \\eqref{eq:r22}, \\eqref{eq:r23} in \\eqref{eq:R2}, \n and using the $L^2$ \\emph{a priori} error estimate in Remark~\\eqref{rm:estimate-regular-load} \n and the stability estimate~\\eqref{eq:regularity}, \n we obtain\n \n \\begin{align*}\n \\abs{b(u,\\widetilde{T}_{h}v) - \\bsht(u, \\widetilde{T}_{h}v)} \n \\leq Ch^{2t+2}\\snorm{u}{r+1} \\, \\snorm{Tv}{r+1}\n \\leq Ch^{2t+2} \\norm{u}{0} \\, \\norm{v}{0}=C h^{2t+2}.\n \\end{align*}\n \n \n \n \n \n \n \n \n \n \n To estimate term $\\mathsf{R}_3$, we first consider the developments:\n \n \\begin{align*}\n \\as^{\\P}(\\widetilde{T}_{h}u,\\widetilde{T}_{h}v) - \n \\as^P_{\\hh}(\\widetilde{T}_{h}u,\\widetilde{T}_{h}v) \n & = \\as^{\\P}(\\widetilde{T}_{h}u - (Tu)_{\\pi},\\widetilde{T}_{h}v) + \n \\as^P_{\\hh}\\big( (Tu)_{\\pi},\\widetilde{T}_{h}v \\big) - \\as^P_{\\hh}(\\widetilde{T}_{h}u,\\widetilde{T}_{h}v) \\\\\n & = \\as^{\\P}(\\widetilde{T}_{h}u - (Tu)_{\\pi},\\widetilde{T}_{h}v - (Tv)_{\\pi}) + \\as^{\\P}_h(\\widetilde{T}_{h}u - (Tu)_{\\pi}, (Tv)_{\\pi}) \\\\ \n & \\phantom{=} + \\as^P_{\\hh}\\big( (Tu)_{\\pi} - \\widetilde{T}_{h}u, \\widetilde{T}_{h}v \\big) \\\\\n & = \\as^{\\P}(\\widetilde{T}_{h}u - (Tu)_{\\pi}, \\widetilde{T}_{h}v - (Tv)_{\\pi} ) + \\as^P_{\\hh}( \\widetilde{T}_{h}u - (Tu)_{\\pi}, (Tv)_{\\pi} - \\widetilde{T}_{h}v ), \n \n \n \\end{align*}\n \n where we make use of the consistency\n condition~\\eqref{eq:k-consistency} to introduce $(Tu)_{\\pi}$ and\n $(Tv)_{\\pi}$, the elemental polynomial approximations of $Tu$\n and $Tv$ that exist in accordance with~\\eqref{eq:approx:pi}.\n \n The terms on the right-hand side of the previous equation are\n similar and we can estimate both as follows:\n \\begin{equation*}\n \\abs{ \\as^{\\P}( \\widetilde{T}_{h}u - (Tu)_{\\pi}, \\widetilde{T}_{h}v - (Tv)_{\\pi} ) } \n \\le \\Big( \\snorm{ (\\widetilde{T}_{h}-T)u }{1,\\P} + \\snorm{ Tu - (Tu)_{\\pi} }{1,\\P} \\Big) \n \\Big( \\snorm{ (\\widetilde{T}_{h} - T)v }{1,\\P} + \\snorm{ Tv - (Tv)_{\\pi} }{1,\\P} \\Big),\n \\end{equation*}\n and, using the \\emph{a priori} error estimate in the broken\n $\\HONE$-norm in Remark~\\eqref{rm:estimate-regular-load}, \n \n the local approximation\n properties of the VEM space by polynomials~\\eqref{eq:approx:pi}, \n and the stability estimate~\\eqref{eq:regularity}, it holds\n \n \\begin{align*}\n \\abs{\\mathsf{R}_3} \n &\\leq\\sum_{\\P\\in\\Th}\\Big| \\as^{\\P}(\\widetilde{T}_{h}u, \\widetilde{T}_{h}v) - \\as^P_{\\hh}( \\widetilde{T}_{h}u,\\widetilde{T}_{h}v ) \\Big| \\\\\n &\\leq C\n \\Big( \\snorm{(\\widetilde{T}_{h}-T)u}{1,h} + \\snorm{Tu-(Tu)_{\\pi}}{1,h} \\Big)\n \\Big( \\snorm{(\\widetilde{T}_{h}-T)v}{1,h} + \\snorm{Tv-(Tv)_{\\pi}}{1,h} \\Big) \\\\\n & \\leq Ch^{2t} \\snorm{\\Tu}{1+r} \\snorm{\\Tv}{1+r} \\le C h^{2t}.\n \\end{align*} \n \n Finally, for term $\\mathsf{R}_4$ we apply the triangle inequality, \n Lemma~\\ref{lemma:Nh:def}, and the stability estimate~\\eqref{eq:regularity} to obtain:\n \\begin{align*}\n \\abs{\\mathsf{R}_4} \n \\leq \\abs{\\mathcal{N}_{\\hh}(Tu,\\widetilde{T}_{h}v)} + \\abs{\\mathcal{N}_{\\hh}(Tv,\\widetilde{T}_{h}u)} \n \\leq Ch^{2t}\\snorm{Tv}{1+r}\\snorm{Tu}{1+r} \\leq C h^{2t}.\n \\end{align*}\n \n The assertion of the theorem follows from the above estimates.\n\\end{proof}\n\nThe proof of the optimal error estimate for the eigenfunctions in the \ndiscrete energy norm follows along the same line as the one for the \nnonconforming finite element method. \nWe briefly report it here for the sake of completeness.\n\\begin{theorem}\n With the same notation as in Theorem~\\ref{th:convl2}, we have\n \n \\begin{align*}\n \\snorm{u-\\widetilde{\\us}_{\\hh}}{1,h}\\le Ch^{t},\n \\end{align*}\n where $t=\\min(k,r)$, $k$ being the order of the method and $r$ the\n regularity index of $u$.\n\\end{theorem}\n\\begin{proof}\n \n \\begin{equation*}\n u - \\widetilde{\\us}_{\\hh} = \\lambda Tu - \\widetilde{\\lambda}_{h} \\widetilde{T}_{h}\\widetilde{\\us}_{\\hh} \n = (\\lambda - \\widetilde{\\lambda}_{h})Tu + \\widetilde{\\lambda}_{h}(T - \\widetilde{T}_{h})u + \\widetilde{\\lambda}_{h} \\widetilde{T}_{h}(u - \\widetilde{\\us}_{\\hh}),\n \\end{equation*}\n then \n \n \\begin{align*}\n \\snorm{u -\\widetilde{\\us}_{\\hh}}{1,h}\n \\leq \\abs{\\lambda - \\lambda_{h}} \\, \\snorm{Tu}{1,h} + \\lambda_{h}\\snorm{(T-\\widetilde{T}_{h})u}{1,h} \n + \\widetilde{\\lambda}_{h} \\snorm{\\widetilde{T}_{h}(u - \\widetilde{\\us}_{\\hh})}{1,h}.\n \\end{align*}\n The first term at the right-hand side of the previous equation is of\n order $h^{2t}$, while the second one is of order $h^{t}$.\n \n Finally, for the last term, using \\eqref{eq:stability}, the continuity of the operator $\\widetilde{T}_{h}$, and Theorem \\ref{th:convl2}, we obtain\n \\begin{align*}\n \\snorm{ \\widetilde{T}_{h}(u-\\widetilde{\\us}_{\\hh})}{1,h}^2 \n &\\leq \\dfrac{1}{\\alpha_*}\\as_{\\hh}(\\widetilde{T}_{h}(u - \\widetilde{\\us}_{\\hh}),\\widetilde{T}_{h}(u - \\widetilde{\\us}_{\\hh})) \\\\\n &= \\dfrac{1}{\\alpha_*}\\bsht(u-\\widetilde{\\us}_{\\hh},\\widetilde{T}_{h}(u - \\widetilde{\\us}_{\\hh})) \n \\le C\\norm{u-\\widetilde{\\us}_{\\hh}}{0}^2 \n \\le C h^{2t+2}.\n \\end{align*}\n\\end{proof}\n", "meta": {"timestamp": "2018-02-09T02:08:55", "yymm": "1802", "arxiv_id": "1802.02942", "language": "en", "url": "https://arxiv.org/abs/1802.02942", "abstract": "We analyse the nonconforming Virtual Element Method (VEM) for the approximation of elliptic eigenvalue problems. The nonconforming VEM allow to treat in the same formulation the two- and three-dimensional case.We present two possible formulations of the discrete problem, derived respectively by the nonstabilized and stabilized approximation of the L^2-inner product, and we study the convergence properties of the corresponding discrete eigenvalue problem. The proposed schemes provide a correct approximation of the spectrum, in particular we prove optimal-order error estimates for the eigenfunctions and the usual double order of convergence of the eigenvalues. Finally we show a large set of numerical tests supporting the theoretical results, including a comparison with the conforming Virtual Element choice.", "subjects": "Numerical Analysis (math.NA)", "title": "The nonconforming virtual element method for eigenvalue problems", "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.987758725428898, "lm_q2_score": 0.865224072151174, "lm_q1q2_score": 0.8546326267184445}} {"url": "https://arxiv.org/abs/2002.07555", "title": "Convergence analysis of multi-level spectral deferred corrections", "abstract": "The spectral deferred correction (SDC) method is class of iterative solvers for ordinary differential equations (ODEs). It can be interpreted as a preconditioned Picard iteration for the collocation problem. The convergence of this method is well-known, for suitable problems it gains one order per iteration up to the order of the quadrature method of the collocation problem provided. This appealing feature enables an easy creation of flexible, high-order accurate methods for ODEs. A variation of SDC are multi-level spectral deferred corrections (MLSDC). Here, iterations are performed on a hierarchy of levels and an FAS correction term, as in nonlinear multigrid methods, couples solutions on different levels. While there are several numerical examples which show its capabilities and efficiency, a theoretical convergence proof is still missing. This paper addresses this issue. A proof of the convergence of MLSDC, including the determination of the convergence rate in the time-step size, will be given and the results of the theoretical analysis will be numerically demonstrated. It turns out that there are restrictions for the advantages of this method over SDC regarding the convergence rate.", "text": "\\section{Numerical Results}\n\\label{sec:examples}\n\n\nIn this section, the convergence behavior of MLSDC, theoretically analyzed in the previous section, is verified by numerical examples. The method is applied to three different initial value problems and the results are compared to those from classical, single-level SDC.\nThe key question here is whether the conditions derived in the previous sections (smoothness, high spatial/temporal resolution and high interpolation order) are actually sharp, i.e.\\ whether MLSDC does indeed show only low order convergence if any of these conditions are violated. \nThe corresponding programs were written in Python using the \\texttt{pySDC} code \\cite{pysdc,10.1145/3310410}.\n\n\n\n\n\\subsection{Heat equation}\n\nThe first numerical example is the one-dimensional heat equation defined by the following initial value problem:\n\\begin{equation}\\begin{gathered}\n \\frac{\\partial}{\\partial t} u(x,t) = \\nu \\frac{\\partial^2}{\\partial x^2} u(x,t), \\quad \\forall t \\in [0,t_{end}], x \\in [0,1]\\\\\n u(0,t) = 0, \\quad u(1,t) = 0,\\\\\n u(x,0) = u_0(x),\n\\end{gathered}\\end{equation}\nwhere $u(x,t)$ represents the temperature at the location $x$ and time $t$ and $\\nu > 0$ defines the thermal diffusivity of the medium.\nThis partial differential equation is discretized in space using standard second-order finite differences with $N$ degrees-of-freedom.\n\n\nAs initial value a sine wave with frequency $\\kappa$ is selected, i.e.\\ $u_0(x) = \\sin(\\kappa \\pi x)$. Under these conditions, the analytical solution of the spatially discretized initial value problem is given by\n\\begin{align*}\n\t\\vec{u}(t) = \\sin(\\kappa \\pi \\vec{x})e^{-t \\nu \\rho}\\text{ with } \\rho = \\frac{1}{\\Delta x^2} (2 - 2\\cos(\\pi \\nu \\vec{x})\n\\end{align*}\nwith $\\vec{x} \\coloneqq (x_n)_{1 \\leq n \\leq N}$ and an element-wise application of the trigonometric functions.\nFor the tests, we choose $\\kappa=4$, $\\nu = 0.1$ and $M=5$ Gauß-Radau collocation nodes.\n\nNote that we are using this linear ODE for our tests even though the linearity of the right-hand side $f$ is not a necessary condition in theorems \\ref{theo:sdc_conv}, \\ref{theo:mlsdc_conv_gen} or \\ref{theo:mlsdc_conv_prec}. In fact, $f$ is just assumed to be Lipschitz continuous. However, we will consider the heat equation here since it is well studied and has a convenient exact solution needed to compute the errors of SDC and MLSDC. \n\nThe following tests are structured in a particular way: In the first one, we will adjust the method parameters according to the results of \\fref{theo:mlsdc_conv_prec} to observe an improved convergence of MLSDC over SDC. More specifically, we will use a small spatial step size $\\Delta x$, a high interpolation order $p$ and try to generate smooth errors using smooth initial guesses for the iteration. In a second step, we will then subsequently change these parameters leading to a lower convergence order of MLSDC as described in \\fref{theo:mlsdc_conv_gen}. Thereby, we will reveal the dependence of MLSDC's convergence behavior on those parameters and simultaneously verify the general minimal achievable convergence order of the method. Altogether, this will confirm the theoretical results of the previous section.\n\n\n\nFor the first test, the number of degrees-of-freedom was set to $N_h = 255$ on the fine and $N_H = 127$ on the coarse level of MLSDC. This parameter particularly determines the spatial grid size $\\Delta x = \\frac{1}{N+1}$.\nAs discussed before, we use injection as restriction and a piecewise $p$-th order Lagrange interpolation as interpolation, for now with $p=8$.\nMoreover, the smooth initial value $u_0(x) = \\sin(4\\pi x)$ was spread across the different nodes $\\tau_m$ to form the initial guess.\n\n\n\\begin{figure}\n \\includegraphics[width=\\textwidth]{errors_heat1d}\n \\caption{Convergence behavior of SDC and MLSDC applied to the discretized one-dimensional heat equation with coarsening in space (for MLSDC) and different parameters}\n \\label{fig:heat}\n\\end{figure}\n\nAn illustration of the corresponding numerical results is shown in \\fref{fig:heat}c, with the reference SDC result in \\fref{fig:heat}a. MLSDC was applied with different step sizes $\\Delta t$ and numbers of iterations $k$ to the considered problem and the resulting errors were plotted as points in the respective graphs. The drawn lines, on the other hand, represent the expected behavior, i.e.\\ the predicted convergence orders of the method according to \\fref{theo:mlsdc_conv_prec}. In particular, we assume that the terms $\\Delta x^p$ and $C(E)$ are sufficiently small such that $\\Delta t^{k_0+2k}$ is the leading order in the corresponding error estimation. As a result, MLSDC is expected to gain two orders per iteration. In the figure, it can be seen that all of the computed points nearly lie on the expected lines which always start at the error resulting for the largest step size. Therefore, the numerical results match the theoretical predictions. \n\nIf, by contrast, the spatial grid size $\\Delta x$ is chosen to be significantly larger, in particular as large as $\\frac{1}{16}$ on the fine and $\\frac{1}{8}$ on the coarse level, the leading order in \\fref{theo:mlsdc_conv_prec}, presenting an error estimation for MLSDC, changes to $\\Delta t^{k_0+k}$. Hence, in this example, we expect MLSDC to only gain one order in $\\Delta t$ with each iteration, as SDC does and as it was described in the general convergence \\fref{theo:mlsdc_conv_gen}. The corresponding numerical results, presented in \\fref{fig:heat}d, confirm this prediction. For both methods, the error decreases by one order in $\\Delta t$ with each iteration.\n\nAnother possible modification of the first example is a decrease of the interpolation order $p$. \\Fref{fig:heat}e shows the numerical results if this parameter is changed to $p=4$. Apparently, this also leads to an order reduction of MLSDC compared to \\fref{fig:heat}c. According to \\fref{theo:mlsdc_conv_prec}, this is a reasonable behavior. In particular, the leading order in the presented error estimation is again reduced to $\\Delta t^{k_0+k}$ due to the higher magnitude of $\\Delta x^p$. Besides, it should be noted that the considered values of $\\Delta t$ are significantly smaller here. This is caused by the fact that MLSDC does not converge for greater values of this parameter, i.e.\\ the upper bound for $\\Delta t$, implicitly occurring in the assumptions of the respective theorem, seems to be lower here. The smaller step sizes $\\Delta t$ also entail overall smaller errors. As a result, the accuracy of the collocation solution is reached earlier which explains the outliers in the considered plots.\n\nThe third necessary condition for the improved convergence of MLSDC is the magnitude of the remainders $E_m$ or, in other words, the smoothness of the error. In this context, we will now have a look at the changes which result from a higher oscillatory initial guess. In particular, we will assign random values to $U^{(0)}$. The corresponding errors are shown in \\fref{fig:heat}b for SDC and \\fref{fig:heat}f for MLSDC. It can be seen that this change results again in a lower convergence order of MLSDC, in particular it gains one order per iteration as SDC. Since this time, as the crucial term $\\Delta x^p$ is left unchanged, the result can only be assigned to a higher value of $C(E)$ and thus to an insufficient smoothness of the error. This may lead to the assumption that for this problem type a smooth initial guess is a sufficient condition for the smoothness of the error and thus, a low value of $C(E)$.\n\n\\subsection{Allen-Cahn equation}\n\nThe second test case is the non-linear, two-dimensional Allen-Cahn equation\n\\begin{align}\\label{eq:ac}\n\tu_t &= \\Delta u + \\frac{1}{\\epsilon^2} u(1-u^2)\\quad \\mathrm{on}\\quad [-0.5, 0.5]^2\\times[0,T],\\ T>0,\\\\\n u(x,0) &= u_0(x),\\quad x\\in[-0.5,0.5]^2,\\nonumber\n\\end{align}\nwith periodic boundary conditions and scaling parameter $\\epsilon > 0$.\nWe use again second-order finite differences in space and choose a sine wave in 2D as initial condition, i.e.\\ $u_0(x) = \\sin(\\kappa \\pi x)\\sin(\\kappa\\pi y)$.\nThere is no analytical solution, neither for the continuous nor for the spatially discretized equations. \nTherefore, reported errors are computed against a numerically computed high-order reference solution.\nFor the tests, we choose $\\kappa=4$, $\\epsilon = 0.2$ and $M=3$ Gauß-Radau collocation nodes.\n\nThe tests are structured precisely as for the heat equation: we first show second-order convergence factors using appropriate parameters and then test the sharpness of the conditions on smoothness, the resolution and the interpolation order.\nFor the first test, the number of degrees-of-freedom per dimension was set to $N_h = 128$ on the fine and $N_H = 64$ on the coarse level of MLSDC.\nTransfer operators are the same as before.\n\n\\begin{figure\n \\includegraphics[width=\\textwidth]{errors_allencahn_2d.pdf}\n \\caption{Convergence behavior of SDC and MLSDC applied to the discretized two-dimensional Allen-Cahn equation with coarsening in space (for MLSDC) and different parameters}\n \\label{fig:allencahn}\n\\end{figure}\n\nFigure~\\ref{fig:allencahn} shows the results of our tests for the Allen-Cahn equation for both SDC and MLSDC.\nThe main conclusion here is the same as before: using less degrees-of-freedom (here $N=32$ on the fine level instead of $128$), a lower interpolation order (here $p=2$ instead of $8$) or a non-smooth (here random) initial guess leads to a degraded order of the convergence factor.\n\n\\subsection{Auzinger's test case}\n\nThe third test case is the following two-dimensional ODE introduced in \\cite{auzinger2004modified}:\n\n\\begin{equation}\\begin{gathered}\n \\dot{u} = \\begin{pmatrix} \\dot{x} \\\\ \\dot{y} \\end{pmatrix} \n = \\begin{pmatrix} -y - \\lambda x(1-x^2-y^2) \\\\ x - \\lambda \\rho y(1-x^2-y^2) \\end{pmatrix}, \\quad \\forall t \\in [0,t_{end}] \\\\\\\n u(0) = u_0,\n\\end{gathered}\\end{equation}\nwhere $\\lambda<0$ determines the stiffness of the problem and $\\rho>0$ is a positive parameter.\nFor the tests, we choose $\\lambda=-0.75$, $\\rho=3$ and $u_0 = (1,0)^T$.\nThe analytical solution of this initial value problem is known. It is given by\n\\begin{align*}\n u(t) = \\begin{pmatrix} x(t) \\\\ y(t) \\end{pmatrix} = \\begin{pmatrix} \\cos(t) \\\\ \\sin(t) \\end{pmatrix},\\quad t \\in [0,t_{end}].\n\\end{align*}\n\nThe corresponding tests are structured in a similar way as before but this time the ODE version of \\fref{theo:mlsdc_conv_prec}, given by \\fref{equ:err_coarse_time}, is considered. So, first appropriate parameters are used to reach second-order convergence of MLSDC, and then the sharpness of the implied conditions is tested. In particular, the improved convergence behavior of MLSDC is expected to depend on the time step size $\\Delta \\tau = C \\Delta t$, the (now temporal) interpolation order $p$ and the smoothness of the error in time. In our tests, we always used the maximal interpolation order $p=M_H$ corresponding to the number of collocation nodes on the coarse level, since otherwise it was not possible to get a second order convergence at all. The number of nodes on the fine grid was chosen to be $M_h=8$.\n\n\\begin{figure}\n \\includegraphics[width=\\textwidth]{errors_auzinger}\n \\caption{Convergence behavior of SDC and MLSDC applied to the Auzinger problem with coarsening in the collocation nodes (for MLSDC) and different parameters}\n \\label{fig:auzinger}\n\\end{figure}\n\n\nThe numerical results are shown in \\fref{fig:auzinger}. Again, they agree with our theoretical predictions: All of the three conditions implied by \\fref{equ:err_coarse_time} need to be fulfilled to reach second-order convergence of MLSDC. A larger time step size (here $\\Delta t \\in [2^{-1}, 2^{-4}]$ instead of $[2^{-3}, 2^{-6}]$), a lower interpolation order (here $p=M_H=2$ instead of $6$) or a non-smooth (here random) initial guess immediately led to a decrease in the order.\n\nHowever, there are a few oddities in the graphs that we would like to discuss here. First of all, the orders shown in \\fref{fig:auzinger}a, c and d are not $2k-1$ and $k-1$ as we would expect, but rather $2k$ and $k$. This behavior is probably related to the $k_0$-term in the estimates which stems from the initial guess of SDC and MLSDC. This explanation would also agree with the result that this additional order gets lost if a random initial guess is used (see \\fref{fig:auzinger}b, f).\nAside from that, it should be noted that the use of a lower interpolation order (\\fref{fig:auzinger}e) led to a convergence order of $k+1$ instead of $k$ as we would have expected. The reason for this is not clear but could be related to \\fref{equ:mlsdc_conv_prec_lte} which implies that all orders between $k$ and $2k$ can potentially be reached. In any case, the result shows that the second-order convergence is lost if the interpolation order is decreased. Finally, we want to discuss the plot in \\fref{fig:auzinger}d resulting from the use of a larger time step size $\\Delta t$.\nIt can be seen that the data points do not perfectly agree with the predicted lines here. Apparently, the numerical results are often much better than expected. However, they do not reach order $2k$ and hence confirm our theory that the second-order convergence of MLSDC is also dependent of a small time step size. The deviations in the data are, in fact, not too surprising here, considering that the time step size is a very crucial parameter for the convergence of MLSDC in general. As described in \\fref{theo:mlsdc_conv_gen} and \\ref{theo:mlsdc_conv_prec}, $\\Delta t$ has to be small enough in order for MLSDC to converge at all. For that reason, the possible testing scope for the time step size is rather small, making it difficult to find appropriate parameters where MLSDC converges exactly with order $k$.\n\nThese artifacts shed some light on the ``robustness'' of the results, a fact that we would like to share here: during the tests with all three examples, we saw that it is actually very hard to get these more or less consistent results. \nAll model and method parameters had to be chosen carefully in order to support the theory derived above so clearly.\nIn many cases the results were much more inconsistent, showing e.g.\\ convergence orders somewhere between $k$ and $2k$, changing convergence orders or stagnating results close to machine precision or discretization errors.\nNone of the tests we did contradicted our theoretical results, though, but they revealed that the bounds we obtained are indeed rather pessimistic.\n\\section{Introduction}\n\n\n\n\nThe original spectral deferred correction (SDC) method for solving ordinary differential equations (ODEs), a variant of the defect and deferred correction methods developed in the 1960s \\cite{def-corr-1,def-corr-2,def-corr-3,def-corr-4}, was first introduced in \\cite{sdc-orig} and then subsequently improved, e.g. in \\cite{hansen,huang,imex-1,imex-2}. It relies on a discretization of the initial value problem in terms of a collocation problem which is then iteratively solved using a preconditioned fixed-point iteration. The iterative structure of SDC has been proven to provide many opportunities for algorithmic and mathematical improvements. \nThese include the option of using Newton-Krylov schemes such as the Newton-GMRES method to solve the resulting preconditioned nonlinear systems, leading to the so-called Krylov deferred correction methods~\\cite{huang,kdc}. Various semi-implicit and multi-implicit formulations of the method have been explored~\\cite{hagstrom,imex-1,imex-2,BourliouxEtAl2003,LaytonMinion2004}. In the last decade, SDC has been applied e.g.\\ to gas dynamics and incompressible or reactive flows~\\cite{appl-1,imex-2} as well as to fast-wave slow-wave problems~\\cite{sdc-expl-euler} or particle dynamics~\\cite{appl-2}. The generalized integral deferred correction framework includes further variations of SDC, where the used discretization approach is not limited to collocation methods~\\cite{idc-1,idc-2}. Moreover, the SDC approach was used to derive efficient parallel-in-time solvers addressing the needs of modern high-performance computing architectures~\\cite{pfasst,sdc-notation}.\n\nHere, we will focus on the multi-level extension of SDC, namely multi-level spectral deferred corrections (MLSDC), which was introduced in \\cite{mlsdc-1}. It uses a multigrid-like approach to solve the collocation problem with SDC iterations (now called ``sweeps'' in this context) being performed on the individual levels. The solutions on the different levels are then coupled using the Full Approximation Scheme (FAS) coming from nonlinear multigrid methods. This variation was designed to improve the efficiency of the method by shifting some of the work to coarser, less expensive levels.\nWhile there are several numerical examples which show the correctness and efficiency of MLSDC \\cite{mlsdc-2,doi:10.1080/13647830.2019.1566574,HAMON2019435}, a theoretical proof of its convergence is still missing. The convergence of SDC, however, was already extensively examined \\cite{causley,hagstrom,hansen,huang,xia,tang}. It could be shown that, under certain conditions, the method gains one order per iteration up to the accuracy of the solution of the collocation problem. The aim of this work now is to prove statements on the convergence behavior of MLSDC using similar concepts and ideas as they were used in the convergence proof of SDC, in particular the one presented in \\cite{tang}.\n\nFor that, we first review SDC along with one of its existing convergence proofs, forming the basis for the following convergence analysis of MLSDC. Then, MLSDC is described and a first convergence theorem is provided. The theorem specifically states that MLSDC behaves at least as good as SDC does. Since this result contradicts our intuitive expectations in that we would assume the multi-level extension to be more efficient than the original one, we will again examine the convergence proof in greater detail, now for a specific choice of transfer operators between the different levels. As a result, a second theorem on the convergence of MLSDC will be derived, describing an improved behavior of the method if particular conditions are fulfilled. \nMore specifically, we will provide theoretical guidelines for parameter choices in practical applications of MLSDC in order to achieve this improved efficiency. Finally, the theoretical results will be verified by numerical examples.\n\\section*{Declarations}\n\n\n\n\n\n\n\n\n\n\\bibliographystyle{gtart} \n\n\n\\section{Multi-Level Spectral Deferred Corrections}\n\\label{sec:mlsdc}\n\n\n\nMulti-level SDC (MLSDC) is a method that uses a multigrid-like approach to solve the collocation problem \\eqref{equ:coll_prob}. It is an extension of SDC in which the iterations, now called ``sweeps'' in this context, are computed on a hierarchy of levels and the individual solutions are coupled in the same manner as used in the full approximation scheme (FAS) for non-linear multigrid methods.\n\nThe different levels are typically created by using discretizations of various resolutions. In this paper, only the two-level algorithm is considered. For this purpose, let $\\Omega_h$ denote the fine level and $\\Omega_H$ the coarse one. Then, $U_h$ denotes the discretized vector on $\\Omega_h$. Furthermore, $C_h$, $F_h$ and $Q_h$ are the discretizations of the operators and the quadrature matrix. Likewise, $U_H$, $C_H$, $F_H$ and $Q_H$ represent the corresponding values for the discretization parameter $H$.\n\nHere, we will consider two coarsening strategies. The first one is a re-discretization in time at the collocation problem, i.e.\\ a reduction of collocation nodes. The second possibility, only applicable if a partial differential equation has to be solved, is a re-discretization in space, i.e.\\ the use of less variables for the conversion into an ODE.\n\nSince it is necessary to perform computations on different levels, a method to transfer vectors between the individual levels is needed. For this purpose, let $I_H^h$ denote the operator that transfers a vector from the coarse level $\\Omega_H$ to the fine level $\\Omega_h$. This operator is called the interpolation operator. $I_h^H$, on the other hand, shall represent the operator for the reverse direction. It is called the restriction operator. Both operators together are called transfer operators.\n\n\nIn detail, the MLSDC two-level algorithm consists of these four steps:\n\\begin{enumerate}\n \\item Compute the $\\tau$-correction as the difference between coarse and fine level:\n \\begin{align}\\begin{split}\n \\label{equ:tau-korr}\n \\tau &= C_H(I_h^H U_h^{(k)}) - I_h^H C_h(U_h^{(k)})\\\\\n &= I_h^H ( \\Delta t Q_h F_h (U_h^{(k)}) ) - \\Delta t Q_H F_H(I_h^H U_h^{(k)}).\n \\end{split}\\end{align}\n \\item Perform an SDC sweep to approximate the solution of the modified collocation problem on the coarse level\n \\begin{equation}\n \\label{equ:qu:coll_coarse}\n C(U_H) = U_{0,H} + \\tau\n \\end{equation}\n on $\\Omega_H$, beginning with $I_h^H U_h^{(k)}$:\n \\begin{align}\\begin{split}\n \\label{equ:mlsdc-alg1}\n U_H^{(k+\\half)} &= U_{0,H} + \\tau + \\Delta t Q_{\\Delta,H} F_H(U_H^{(k+\\half)})\\\\\n &\\quad+ \\Delta t (Q_H - Q_{\\Delta,H}) F_H(I_h^H U_h^{(k)}).\n \\end{split}\\end{align}\n \\item Compute the coarse level correction:\n \\begin{align}\n \\label{equ:mlsdc-alg2}\n U_h^{(k+\\half)} = U_h^{(k)} + I_H^h (U_H^{(k+\\half)} - I_h^H U_h^{(k)}).\n \\end{align}\n \\item Perform an SDC sweep to approximate the solution of the original collocation problem\n \\begin{equation*}\n C(U_h) = U_{0,h}\n \\end{equation*}\n on $\\Omega_h$, beginning with $U_h^{(k+\\half)}$:\n \\begin{align}\\begin{split}\n \\label{equ:mlsdc-alg3}\n U_h^{(k+1)} &= U_{0,h} + \\Delta t Q_{\\Delta,h} F_h(U_h^{(k+1)})\\\\\n &\\quad+ \\Delta t (Q_h - Q_{\\Delta,h}) F_h(U_h^{(k+\\half)}).\n \\end{split}\\end{align}\n\\end{enumerate}\n\nNote that for better readability, the enlargements of the matrices $Q$ and $Q_\\Delta$ by applying the Kronecker product with the identity matrix are no longer indicated.\n\n\n \n \n \n \n \n\n\\subsection{A first convergence proof}\n\nHere, we will extend the existing convergence proof for SDC, as presented in \\fref{theo:sdc_conv}, to prove the convergence of its multi-level extension MLSDC. The following theorem provides an appropriate convergence statement. In the proof, we use very similar ideas as in the one for the convergence of SDC. \n\n\\begin{theorem}\n\t\\label{theo:mlsdc_conv_gen}\n\tConsider a generic initial value problem like (\\ref{equ:ivp}) with a Lipschitz-continuous function $f$ on the right-hand side. \n \n If the step size $\\Delta t$ is sufficiently small, MLSDC converges linearly to the solution of the collocation problem with a convergence rate in $\\mathcal{O}(\\Delta t)$, i.e. the following estimate for the error of the $k$-th iterated $U_h^{(k)}$ of MLSDC compared to the solution of the collocation problem $U_h$ is valid:\n\t\\begin{align}\n\t \\label{equ:mlsdc_conv_gen_iter}\n \t\\infnorm{U_h - U_h^{(k)}} \\leq C_9 \\Delta t \\infnorm{U_h - U_h^{(k-1)}},\n\t\\end{align}\n\twhere the constant $C_9$ is independent of $\\Delta t$.\n \t\n \tIf, additionally, the solution of the initial value problem $u$ is $(M+1)$-times continuously differentiable, the LTE of MLSDC compared to the solution of the ODE can be bounded by\n \t\\begin{align}\n\t \\label{equ:mlsdc_conv_gen_lte}\n\t\t\\|\\bar{U} - U_h^{(k)}\\|_\\infty &\\leq C_{10} \\Delta t^{k_0+k} \\|u\\|_{k_0+1} \n\t\t+ C_{11} \\Delta t^{M+1} \\|u\\|_{M+1}\\\\\n &= \\mathcal{O}(\\Delta t^{\\min(k_0+k, M+1)}),\n\t\\end{align}\n\twhere the constants $C_{10}$ and $C_{11}$ are independent of $\\Delta t$, $k_0$ denotes the approximation order of the initial guess $U^{(0)}$ and $\\|u\\|_p$ is defined by $\\infnorm{u^{(p)}}$.\n\\end{theorem}\n\\begin{proof}\n For better readability, the maximum norm $\\infnorm{\\cdot}$ is denoted with the simple norm $\\norm{\\cdot}$ within this proof.\n \\renewcommand{\\infnorm}[1]{\\lVert#1\\rVert}\n Besides, we consider $U_h^{(k+1)}$ instead of $U_h^{(k)}$ here, in order to enable consistent references to the definition of the MLSDC algorithm above.\n \n As the last step of an MLSDC iteration, in particular \\fref{equ:mlsdc-alg3}, corresponds to an SDC iteration, we can use \\fref{theo:sdc_conv} to get an initial error estimation. Keeping in mind that the SDC iteration is based on $U_h^{(k+\\half)}$ as initial guess here, the application of the mentioned theorem yields the estimation\n \\begin{align}\n \\label{equ:u-1_rek_u-half}\n \\infnorm{U_h - U_h^{(k+1)}} \\leq C_4 \\Delta t \\infnorm{U_h - U_h^{(k+\\half)}},\n \\end{align}\n if the step size $\\Delta t$ is sufficiently small.\n \n Now, the expression on the right-hand side of the above equation will be further examined. In this context, the definition of an MLSDC iteration, in particular \\fref{equ:mlsdc-alg2}, yields\n \\begin{align*}\n \\infnorm{U_h - U_h^{(k+\\half)}} &= \\infnorm{U_h - U_h^{(k)} - I_H^h \\left( U_H^{(k+\\half)} - I_h^H U_h^{(k)} \\right)}\\\\\n &= \\infnorm{U_h - U_h^{(k)} - I_H^h \\left( U_H^{(k+\\half)} + U_H - U_H - I_h^H U_h^{(k)} \\right)}\\\\\n &= \\infnorm{(I - I_H^h I_h^H) (U_h - U_h^{(k)}) + I_H^h ( U_H - U_H^{(k+\\half)} )},\n \\end{align*}\n where in the last step we used the identity $U_H = I_h^H U_h$ which applies in consequence of the $\\tau$-correction stemming from the usage of FAS. We get\n \\begin{align}\n \\label{equ:u-half_Ihh}\n \\infnorm{U_h - U_h^{(k+\\half)}} &\\leq \\infnorm{(I - I_H^h I_h^H) (U_h - U_h^{(k)})} + \\infnorm{I_H^h ( U_H - U_H^{(k+\\half)} )}\\\\\n &\\leq \\tilde{C}_1 \\infnorm{U_h - U_h^{(k)}} + \\tilde{C}_2 \\infnorm{U_H - U_H^{(k+\\half)}},\n \\label{equ:u-half}\n \\end{align}\n where here and in the following, temporary arising constants will again be denoted by symbols like $\\tilde{C}_i$.\n \n Now, it follows a further investigation of the newly emerged term, in particular the second summand of \\fref{equ:u-half}. An insertion of the corresponding definitions, namely \\fref{equ:qu:coll_coarse} and (\\ref{equ:mlsdc-alg1}), together with the application of the triangle inequality and \\fref{lem:estimation} yields\n \\begin{align*}\n \\begin{split}\n \\infnorm{U_H - U_H^{(k+\\half)}} \n &= \\infnorm{\\Delta t Q_H (F_H(U_H) - F_H(I_h^H U_h^{(k)}))\\\\\n &\\quad\\quad + \\Delta t Q_{\\Delta,H} (F_H(I_h^H U_h^{(k)}) - F_H(U_H^{(k+\\half)}))}\n \\end{split}\\\\\n &\\leq C_{2,H} \\Delta t \\infnorm{U_H - I_h^H U_h^{(k)}} + C_{3,H} \\Delta t \\infnorm{I_h^H U_h^{(k)} - U_H^{(k+\\half)}} \\nonumber\\\\\n &\\leq \\tilde{C}_3 \\Delta t \\infnorm{U_H - I_h^H U_h^{(k)}} + C_{3,H} \\Delta t \\infnorm{U_H - U_H^{(k+\\half)}} \\nonumber.\n \\end{align*}\n Subtracting $C_{3,H} \\Delta t \\infnorm{U_H - U_H^{(k+\\half)}}$ from both sides and dividing by $1-C_{3,H} \\Delta t$, results in\n \\begin{align*}\n \\infnorm{U_H - U_H^{(k+\\half)}} &\\leq \\frac{\\tilde{C}_3}{1-C_{3,H} \\Delta t} \\Delta t \\infnorm{U_H - I_h^H U_h^{(k)}}.\n \\end{align*}\n With the same argumentation as above, given a sufficiently small step size, it follows\n \\begin{align}\n \\infnorm{U_H - U_H^{(k+\\half)}} &\\leq \\tilde{C}_4 \\Delta t \\infnorm{U_H - I_h^H U_h^{(k)}} \\nonumber\\\\\n &\\leq \\tilde{C}_5 \\Delta t \\infnorm{U_h - U_h^{(k)}},\n \\label{equ:u-half_H}\n \\end{align}\n where in the last step we used the identity $U_H = I_h^H U_h$.\n \n By inserting \\fref{equ:u-half} and this result subsequently into \\fref{equ:u-1_rek_u-half}, we obtain\n \\begin{align*}\n \\infnorm{U_h - U_h^{(k+1)}} &\\leq C_4 \\Delta t \\infnorm{U_h - U_h^{(k+\\half)}}\\\\\n &\\leq C_4 \\Delta t (\\tilde{C}_1 \\infnorm{U_h - U_h^{(k)}} + \\tilde{C}_2 \\infnorm{U_H - U_H^{(k+\\half)}})\\\\\n &\\leq \\tilde{C}_6 \\Delta t \\infnorm{U_h - U_h^{(k)}} + \\tilde{C}_7 \\Delta t ( \\tilde{C}_5 \\Delta t \\infnorm{U_h - U_h^{(k)}})\\\\\n &= (\\tilde{C}_6 + \\tilde{C}_8 \\Delta t) \\Delta t \\infnorm{U_h - U_h^{(k)}}.\n \\end{align*}\n Since the step size $\\Delta t$ is assumed to be sufficiently small, i.e.\\ bounded above, the following estimate is valid\n \\begin{align*}\n \\tilde{C}_6 + \\tilde{C}_8 \\Delta t \\leq C_9,\n \\end{align*}\n which concludes the proof for \\fref{equ:mlsdc_conv_gen_iter}.\n \n\tThe proof of \\fref{equ:mlsdc_conv_gen_lte} is similar to the one of \\fref{equ:sdc_conv_lte} in \\fref{theo:sdc_conv}, using the previous result.\n\\end{proof}\n\n\\begin{remark}\n As before, if the Lipschitz constant of $f$ depends on the spatial resolution, then constants $\\tilde{C}_3$ and $C_{3,H}$ do as well.\n In~\\eqref{equ:u-half_H}, constant $\\tilde{C}_5$ then only depends on the original condition $1 - C_{3,H} < 1$, which is just the condition we know from SDC, but with a scaled constant if spatial coarsening is applied.\n As in remark~\\ref{rem:dx_sdc} we can write $\\tilde{C}_5 = \\tilde{C}_5(\\delta^{-1})$. \n This only affects $C_{10}$ in \\fref{equ:mlsdc_conv_gen_lte}, where now $C_{10} = C_{10}(\\delta^{-(k+1)})$.\n\\end{remark}\n\nAs \\fref{theo:sdc_conv}, this theorem can also be read as a convergence statement. It shows that MLSDC, interpreted as an iterative method solving the collocation problem, converges linearly with a convergence rate in $\\mathcal{O}(\\Delta t)$ if $C_9 \\Delta t < 1$. Moreover, the second part of the theorem shows that MLSDC, in the sense of a discretization method for ODEs, converges with order $\\min(k_0+k, M+1)$.\n\n\\begin{remark}\n \\label{rem:mlsdc_analogous}\n The results regarding consistency and stability of SDC, namely \\fref{cor:conv_last} and \\fref{theo:sdc_stab}, can be easily adapted for MLSDC. Analogous to SDC, it can be proven that the error at the last collocation node can be bounded by $\\mathcal{O}(\\Delta t^{\\min(k_0+k, 2M)})$ and the increment function of MLSDC is Lipschitz continuous for $\\Delta t$ small enough.\n\\end{remark}\n\nAlthough \\fref{theo:mlsdc_conv_gen} is the first general convergence theorem for MLSDC, its statement is rather disappointing: It merely establishes that MLSDC converges as least as fast as SDC, although more work is done per iteration.\n\n\nA deeper look into the proof of the theorem gives an idea on the cause for the rather unexpected low convergence order. In particular, it is the estimate leading to \\fref{equ:u-half} which is responsible for this issue. This equation implies that \n\\begin{align*}\n \\norm{U_h - U_h^{(k+\\half)}} \\leq C \\norm{U_h - U_h^{(k)}},\n\\end{align*} \nwhich essentially means that the additional iteration on the coarse level does not gain any additional order in $\\Delta t$ compared to the previously computed iterated on the fine level $U_h^{(k)}$. More specifically, it is the estimation \n\\begin{align*}\n \\infnorm{(I-I_H^h I_h^H)(U_h - U_h^{(k)})} \\leq C \\infnorm{U_h - U_h^{(k)}},\n\\end{align*}\nin equation \\eqref{equ:u-half} which leads to this result.\n\n\n\t\t\nThus, a possibly superior behavior of MLSDC seems to depend on the magnitude of $\\infnorm{(I - I_H^h I_h^H)e_h}$ with $e_h = U_h - U_h^{(k)} \\in \\mathbb{C}^{MN}$ which describes the difference between an original vector and the one which results from restricting and interpolating it. Consequently, the term can be interpreted as the quality of the approximation on the coarse level or the accuracy loss it causes, respectively. In the following, this term will be examined in detail, resulting in a new theorem for the convergence of MLSDC with a higher convergence order but additional assumptions which have to met. \n\n\\subsection{An improved convergence result}\\label{ssec:improved_mlsdc}\n\nFor this purpose, we will mainly focus on a specific coarsening strategy here, in particular coarsening in space. The differences occurring from coarsening in time will be discussed at the end of this section. Moreover, we will focus on particular methods used for the transfer operators. For $I_H^h$ we consider a piece-wise Lagrange interpolation of order $p$. This means that instead of using all $N_H$ available values to approximate the value at a particular point $x_i \\in \\Omega_h$, only its $p$ neighbors are taken into account for this purpose. Hence, $I_H^h$ corresponds to the application of a $p$-th order Lagrange interpolation for each point. For the restriction operator $I_h^H$, on the other hand, we consider simple injection. Thereby, we can mainly focus on the interpolation order and disregard the restriction order.\n\nThe following lemma now provides an appropriate estimation for the considered term $\\infnorm{(I - I_H^h I_h^H)e_h}$.\n\n\\begin{lemma}\n\t\\label{lem:ihh_gen}\n Let $E \\coloneqq (E_m)_{1 \\leq m \\leq M}$ denote the remainder of the truncated inverse discrete Fourier transformation of $U_h - U_h^{(k)}$, i.e.\n \\begin{align*}\n E_m \\coloneqq \\sum_{\\ell=N_0}^{N-1} \\abs{c_{m,\\ell}},\\quad m = 1, \\dots, M.\n \\end{align*}\n for some cutoff index $N_0 \\le N$ and $c_{m,\\ell}\\in\\mathbb{C}$ being the Fourier coefficients.\n \n Then, the following estimate for this error is valid:\n \\begin{align*}\n \\infnorm{(I - I_H^h I_h^H)(U_h - U_h^{(k)})} \\leq (C_{11} \\Delta x^p + C_{12}(E)) \\infnorm{U_h - U_h^{(k)}},\n \\end{align*}\n where $I \\equiv I_{MN}$ denotes the identity matrix of size $MN$, $I_H^h$ is the piece-wise spatial Lagrange interpolation of order $p$ and $I_h^H$ denotes the injection operator. Furthermore, $\\Delta x \\equiv \\Delta x_H$ is defined as the resolution in space on the coarse level $\\Omega_H$ of MLSDC.\n\\end{lemma}\n\\begin{proof}\n\t\\renewcommand{\\infnorm}[1]{\\lVert#1\\rVert_\\infty}\n\tFirst of all, we need to introduce some definitions. \n\tFor ease of notation, the considered error vector $U_h - U_h^{(k)}$ will be denoted by $e_h$ within the proof. \n\tIn detail, the following definition is used:\n \\begin{align*}\n e_{m,n} = e_{h,n}(\\tau_m) = u_{h,n}(\\tau_m)- u_{h,n}^{(k)}(\\tau_m), \\quad \\forall m=1,\\dots,M,\\; n=1,\\dots,N,\n \\end{align*}\n where $m$ denotes the temporal index identifying the particular collocation node $\\tau_m$ and $n$ represents the spatial index referring to some discretized points $(x_n)_{1 \\leq n \\leq N}$ within the considered interval in space $[0,S]$. Additionally, we assume the spatial steps to be equidistant here.\n \n Another definition needed for the proof is $g_{m,p}(x)$. It denotes the Lagrangian interpolation polynomial of order $p$ for the restricted vector $I_h^H e_h$ for each point in time $\\tau_m$, $m=1, \\dots, M$. As the two levels $\\Omega_h$ and $\\Omega_H$ differ in their spatial resolution, the transfer operators are applied at the spatial axis and thus can be considered separately for each component $e_h(\\tau_m)$. More specifically, the restriction operator, corresponding to simple injection according to the assumptions, omits several values of $e_h(\\tau_m) \\in \\mathbb{C}^N$ resulting in $(I_h^H e_h)(\\tau_m) \\in \\mathbb{C}^{N_H}$, where $N_H$ denotes the number of degrees of freedom at the coarse level. The subsequent application of the interpolation operator at this vector then leads to the $M$ interpolating polynomials $(g_{m,p}(x))_{1 \\leq m \\leq M}$ with $p$ referring to their order of accuracy.\n \n Having introduced these notations, the considered term can be written as\n \\begin{align*}\n (I - I_H^h I_h^H) (U_h - U_h^{(k)}) = (I - I_H^h I_h^H) e_h = \\left(e_{m,n} - g_{m,p}(x_n)\\right)_{\\substack{1 \\leq m \\leq M,\\\\1 \\leq n \\leq N}},\n \\end{align*}\n so that we can focus on $\\abs{e_{m,n} - g_{m,p}(x_n)}$.\n \n Since $g_{m,p}(x)$ partially interpolates the points $(e_{m,n})_{1 \\leq n \\leq N}$, it seems very reasonable to use the general error estimation of Lagrangian interpolation to determine an estimate for the considered term. However, there is a crucial issue: The corresponding error bound, generally defined in \\cite{bartels,interp-err} by\n \\begin{align}\n \\label{equ:interpol_error}\n \\max_{x \\in [a,b]} \\abs{f(x) - g_p(x)} \\leq \\frac{\\Delta x^p}{4p} \\abs{f^{(p)} (\\xi)}, \\quad \\xi \\in [a,b]\n \\end{align}\n apparently depends on the function $f$ from which the interpolation points are obtained. In our case, namely $\\abs{e_{m,n} - g_{m,p}(x_n)}$, we do not have a specific function directly available to which the error components $e_{m,n}$ correspond. \n \n To fill this gap, we will now derive an appropriate function for this purpose, using a continuous extension of the inverse Discrete Fourier Transformation (iDFT). \n The iDFT at the spatial axes for the points $e_{m,n}$ is given by\n \\begin{align*}\n e_{m,n} = \\frac{1}{\\sqrt{N}} \\sum_{\\ell=0}^{N-1} c_{m,\\ell} \\exp{i \\frac{2\\pi}{N} (n-1) \\ell}, \\quad m=1, \\dots, M, \\quad n=1, \\dots, N\n \\end{align*}\n with $c_{m,\\ell}$ as Fourier coefficients and $i$ symbolizing the imaginary unit \\cite{fourier}. \n \n \n A continuous extension $\\tilde{e}_m(x), x \\in [0,S]$, on the whole spatial interval can then be derived by enforcing $\\tilde{e}_m(x_n) \\overset{!}{=} e_{m,n}$ for all $m$ and $n$.\n With the transformation $x_n = \\frac{S}{N} (n-1)$, i.e.\\ $n = \\frac{N}{S} x_n +1$, implied by an equidistant spatial discretization, it follows\n \\begin{align*}\n \\tilde{e}_m(x_n) &= \\frac{1}{\\sqrt{N}} \\sum_{\\ell=0}^{N-1} c_{m,\\ell} \\exp{i \\frac{2\\pi}{S}\\ell x_n}\n \\end{align*}\n and hence\n \\begin{align*}\n \\tilde{e}_m(x) \\coloneqq \\frac{1}{\\sqrt{N}} \\sum_{\\ell=0}^{N-1} c_{m,\\ell} \\exp{i \\frac{2\\pi}{S} \\ell x}, \\quad x \\in [0, S], \\quad m=1, \\dots, M.\n \\end{align*}\n \n Consequently, we have found a function describing the points $e_{m,n}$. Thus, an interpretation of $g_{m,p}(x)$ as interpolation polynomial of $p$ points stemming from $\\tilde{e}_m(x)$ is possible and the error estimation for Lagrange interpolation, presented in \\fref{equ:interpol_error}, can now be applied. As a result, we get\n \\begin{align}\n \\abs{e_{m,n} - g_{m,p}(x_n)} = \\abs{\\tilde{e}_m(x_n) - g_{m,p}(x_n)} &\\leq \\frac{\\Delta x^p}{4p} \\abs{\\tilde{e}_m^{(p)} (\\xi)}\n \\end{align} \n with $\\xi \\in [0,S]$.\n \n With the insertion of the definition of $\\tilde{e}_m(x)$ and its $p$th derivative, it follows\n \\begin{align*}\n \t\\abs{\\tilde{e}_m(x_n) - g_{m,p}(x_n)} &\\leq \\frac{\\Delta x^p}{4p} \\abs*{\\frac{1}{\\sqrt{N}} \\sum_{\\ell=0}^{N-1} c_{m,\\ell} (i \\frac{2\\pi}{S}\\ell)^p \\exp{i \\frac{2\\pi}{S}\\ell \\xi}}\\\\\n \t&= \\frac{1}{\\sqrt{N}} \\frac{\\Delta x^p}{4p} \\left(\\frac{2\\pi}{S}\\right)^p \\abs*{\\sum_{\\ell=0}^{N-1} c_{m,\\ell} \\ell^p \\exp{i \\frac{2\\pi}{S}\\ell \\xi}},\n \\end{align*}\n so that with $\\tilde{C}(p) \\coloneqq \\frac{1}{4p} \\left(\\frac{2\\pi}{S}\\right)^p$ we have\n \\begin{align*}\n \t\\abs{\\tilde{e}_m(x_n) - g_{m,p}(x_n)} &\\leq \\frac{1}{\\sqrt{N}} \\tilde{C}(p) \\Delta x^p \\sum_{\\ell=0}^{N-1} \\abs*{c_{m,\\ell} \\ell^p \\exp{i \\frac{2\\pi}{S}\\ell \\xi}}\\\\\n \t&\\leq \\frac{1}{\\sqrt{N}} \\tilde{C}(p) \\Delta x^p \\sum_{\\ell=0}^{N-1} \\abs{c_{m,\\ell}} \\ell^p.\n \\end{align*}\n We now choose $N_0\\le N$ such that \n \\begin{align*}\n \\sum_{\\ell=0}^{N_0-1} \\abs{c_{m,\\ell}} \\geq \\epsilon_m > 0 \\quad \\forall m\n \\end{align*}\n for given $\\epsilon_m$. Note that this is not possible if there exists an $m$ for which the coefficients $(c_{m,\\ell})_{\\ell = 0, \\dots, N-1}$ are all $0$. This would imply that for this particular $m$ the error $e_{m,n}$ equals to $0$ for all $n = 1, \\dots, N$. However, the respective error $e_h(\\tau_m)$ would then be $0$ and could just be disregarded in our particular context as it does not have an impact on the considered maximum norm. Therefore, the assumption does not lead to a loss of generality. Finally, we define the remainders of the sums as\n \\begin{align*}\n E_m \\coloneqq \\sum_{\\ell=N_0}^{N-1} \\abs{c_{m,\\ell}},\\quad m = 1, \\dots, M.\n \\end{align*}\n \n Now, the sum in the previous estimation will be split at $N_0$ resulting in the following estimate:\n \\begin{align*}\n \\abs{\\tilde{e}_m(x_n) - g_{m,p}(x_n)} &\\leq \\frac{1}{\\sqrt{N}} \\tilde{C}(p) \\Delta x^p \\left( \\sum_{\\ell=0}^{N_0-1} \\abs{c_{m,\\ell}} \\ell^p + \\sum_{\\ell=N_0}^{N-1} \\abs{c_{m,\\ell}} \\ell^p \\right).\n \\end{align*}\n With the simple estimation $\\ell \\leq N_0$ for the first sum and $\\ell \\leq N$ for the second one, the following formulation using the definition of the remainder $E_m$ is obtained:\n \\begin{align*}\n \t\\abs{\\tilde{e}_m(x_n) - g_{m,p}(x_n)} &\\leq \\frac{1}{\\sqrt{N}} \\tilde{C}(p) \\Delta x^p \\left( N_0^p \\sum_{\\ell=0}^{N_0-1} \\abs{c_{m,\\ell}} + N^p \\sum_{\\ell=N_0}^{N-1} \\abs{c_{m,\\ell}} \\right)\\\\\n \t&= \\frac{1}{\\sqrt{N}} \\tilde{C}(p) \\Delta x^p \\left( N_0^p \\sum_{\\ell=0}^{N_0-1} \\abs{c_{m,\\ell}} + N^p E_m \\right).\n \\end{align*}\n \n Now, we will have a look at the norm of the whole vector $(I - I_H^h I_h^H)e_h$, but instead of the maximum norm, we first consider the squared 2-norm given by\n \\begin{align*}\n \t\\norm{(I - I_H^h I_h^H) e_h}_2^2 &= \\sum_{n=1}^N \\sum_{m=1}^M \\abs{\\tilde{e}_m(x_n) - g_{m,p}(x_n)}^2.\n \\end{align*}\n\tBy the insertion of the previous estimation, it follows\n \\begin{align*}\n \t\\norm{(I - I_H^h I_h^H) e_h}_2^2 &\\leq \\sum_{n=1}^N \\sum_{m=1}^M \\left[ \\frac{1}{\\sqrt{N}} \\tilde{C}(p) \\Delta x^p \\left( N_0^p \\sum_{\\ell=0}^{N_0-1} \\abs{c_{m,\\ell}} + N^p E_m \\right) \\right]^2.\n \\end{align*}\n Since the summands are independent of the running index $n$, the equation can be simplified to\n \\begin{align}\n \\label{equ:tmp1}\n \t\\norm{(I - I_H^h I_h^H) e_h}_2^2 &\\leq \\tilde{C}(p)^2 \\Delta x^{2p} \\sum_{m=1}^M \\left( N_0^p \\sum_{\\ell=0}^{N_0-1} \\abs{c_{m,\\ell}} + N^p E_m \\right)^2.\n \\end{align}\n Now, each inner summand can be written as\n \n \n \n \n \n \n \n \\begin{align}\n \\label{equ:tmp2}\n \\left( N_0^p \\sum_{\\ell=0}^{N_0-1} \\abs{c_{m,\\ell}} + N^p E_m \\right)^2 &= \n \tN_0^{2p} \\left( \\sum_{\\ell=0}^{N_0-1} \\abs{c_{m,\\ell}} \\right)^2 + \n \t2 N_0^p N^p E_m \\sum_{\\ell=0}^{N_0-1} \\abs{c_{m,\\ell}} \\nonumber \\\\\n \t&\\quad+ N^{2p} E_m^2\\\\\n \t&=: S_1 + S_2 + S_3.\\nonumber\n \\end{align}\n While $S_1$ is an intended component (it contains the squared sum of $c_{m,\\ell}$ which we will need to get back to the norm of $e_h$), the other two summands $S_2$ and $S_3$ are inconvenient. Therefore, we will now eliminate them by searching a $T(E_m)$ such that\n \\begin{align}\\begin{split}\n \\label{equ:tmp3}\n\t\tS_1 + S_2 + S_3 &\\leq S_1 + T(E_m) \\left( \\sum_{\\ell=0}^{N_0-1} \\abs{c_{m,\\ell}} \\right)^2\\\\\n\t\t&= \\left( N_0^{2p} + T(E_m) \\right) \\left( \\sum_{\\ell=0}^{N_0-1} \\abs{c_{m,\\ell}} \\right)^2.\n \\end{split}\\end{align}\n This is true if \n \\begin{align*}\n\t\tS_2 + S_3 - T(E_m) \\left( \\sum_{\\ell=0}^{N_0-1} \\abs{c_{m,\\ell}} \\right)^2 \\leq 0,\n\t\\end{align*}\n\twhich in turn leads to \n\t\\begin{align*}\t\t\n T(E_m) \\geq \\frac{2 N_0^p N^p}{\\sum_{\\ell=0}^{N_0-1} \\abs{c_{m,\\ell}}} E_m + \\frac{N^{2p}}{\\left( \\sum_{\\ell=0}^{N_0-1} \\abs{c_{m,\\ell}} \\right)^2} E_m^2,\n \\end{align*}\n after using the definitions of $S_2$ and $S_3$.\n Thus, for \n \\begin{align}\n \\label{equ:def_tem}\n \tT(E_m) \\coloneqq \\frac{2 N_0^p N^p}{\\epsilon_m} E_m + \\frac{N^{2p}}{\\epsilon_m^2} E_m^2\n \\end{align}\n\twe can bound \n \\begin{align*}\n \tS_1 + S_2 + S_3 \\le \\left( N_0^{2p} + T(E_m) \\right) \\left( \\sum_{\\ell=0}^{N_0-1} \\abs{c_{m,\\ell}} \\right)^2.\n \\end{align*}\n Using the Cauchy-Schwarz inequality we have\n \\begin{align*}\n \\left( \\sum_{\\ell=0}^{N_0-1} \\abs{c_{m,\\ell}} \\right) ^2 = \\left( \\sum_{\\ell=0}^{N_0-1} \\abs{c_{m,\\ell}} \\cdot 1 \\right) ^2 \\leq \\sum_{\\ell=0}^{N_0-1} \\abs{c_{m,\\ell}}^2 \\cdot \\sum_{\\ell=0}^{N_0-1} 1^2 = N_0 \\sum_{\\ell=0}^{N_0-1} c_{m,\\ell}^2,\n \\end{align*}\n so that with \\eqref{equ:tmp1}, \\eqref{equ:tmp2} and \\eqref{equ:tmp3} we get\n \\begin{align*}\n \t\\norm{(I - I_H^h I_h^H) e_h}_2^2 &\\leq \\tilde{C}(p)^2 \\Delta x^{2p} N_0 \\sum_{m=1}^M \\left( N_0^{2p} + T(E_m) \\right) \\sum_{\\ell=0}^{N_0-1} c_{m,\\ell}^2.\n \\end{align*}\n \n Further, it follows from Parseval's theorem~\\cite{fourier} that\n \\begin{align*}\n \\sum_{\\ell=0}^{N_0-1} c_{m,\\ell}^2 \\leq \\sum_{\\ell=0}^{N-1} c_{m,\\ell}^2 = \\sum_{n=1}^{N} e_{m,n}^2\n \\end{align*}\n and thus\n \\begin{align*}\n \t\\norm{(I - I_H^h I_h^H) e_h}_2^2 &\\leq \\tilde{C}(p)^2 \\Delta x^{2p} N_0 \\sum_{m=1}^M \\left( N_0^{2p} + T(E_m) \\right) \\sum_{n=1}^{N} e_{m,n}^2\\\\\n \t&= \\tilde{C}(p)^2 \\Delta x^{2p} N_0 \\left( N_0^{2p} + \\max_{m=1,\\dots,M} T(E_m) \\right) \\norm{e_h}_2^2.\n \\end{align*}\n Since it is the maximum norm we are interested in and not the Euclidean one, an appropriate transformation is required, given by\n \\begin{align*}\n \\infnorm{x} \\leq \\norm{x}_2 \\leq \\sqrt{n} \\infnorm{x} \\quad \\forall x \\in \\mathbb{C}^n,\n \\end{align*}\n so that\n \\begin{align*}\n \t\\infnorm{(I - I_H^h I_h^H) e_h} &\\leq \\norm{(I - I_H^h I_h^H) e_h}_2\\\\\n \t&\\leq \\tilde{C}(p) \\Delta x^{p} \\sqrt{N_0 \\left( N_0^{2p} + \\max_{m=1,\\dots,M} T(E_m) \\right)} \\norm{e_h}_2\\\\\n \t&\\leq \\tilde{C}(p) \\Delta x^{p} \\sqrt{N_0MN} \\sqrt{N_0^{2p} + \\max_{m=1,\\dots,M} T(E_m)} \\infnorm{e_h}.\n \\end{align*}\n Using the triangle inequality for square roots, the sum can be split as\n \\begin{align*}\n \t\\infnorm{(I - I_H^h I_h^H) e_h} &\\leq \\tilde{C}(p) \\Delta x^{p} \\sqrt{N_0MN} \\left( N_0^{p} + \\sqrt{\\max_{m=1,\\dots,M} T(E_m)} \\right) \\infnorm{e_h}.\n \\end{align*}\n With the insertion of the definition of $T(E_m)$ presented in \\fref{equ:def_tem}, we get\n \\begin{align}\\begin{split}\n \\label{equ:first_sight}\n \t\\infnorm{(I - I_H^h I_h^H) e_h} &\\leq \\tilde{C}(p) \\sqrt{N_0MN} N_0^{p} \\Delta x^{p}\\\\\n &\\quad \\cdot \\left( N_0^p + \\sqrt{\\max_{m=1,\\dots,M} \\frac{2 N_0^p N^p}{\\epsilon_m} E_m + \\frac{N^{2p}}{\\epsilon_m^2} E_m^2} \\right) \\infnorm{e_h}.\n \\end{split}\\end{align}\n\n At the first sight, it looks like $\\Delta x^p$ is the dominating term in this equation. However, a closer look reveals that the root term is in $\\mathcal{O}(N^p)$ which shifts the dominance of this summand towards the remainder $E_m$. To see this, we first transform the root term by extracting $N^{2p}$:\n \\begin{align*}\n \\sqrt{\\max_{m=1,\\dots,M} \\frac{2 N_0^p N^p}{\\epsilon_m} E_m + \\frac{N^{2p}}{\\epsilon_m^2} E_m^2} &\\leq \n \n \\sqrt{\\max_{m=1,\\dots,M} N^{2p} \\left( \\frac{2 N_0^p}{\\epsilon_m} E_m + \\frac{1}{\\epsilon_m^2} E_m^2 \\right)}\\\\\n &= N^p \\sqrt{\\max_{m=1,\\dots,M} \\frac{2 N_0^p}{\\epsilon_m} E_m + \\frac{1}{\\epsilon_m^2} E_m^2}.\n \\end{align*} \n Then, we consider the definition of the step size on the fine level, namely $\\Delta x_h = \\frac{S}{N}$, which allows the representation of $N^p$ as $\\left( \\frac{S}{\\Delta x_h} \\right) ^p$. From this representation, it follows\n \\begin{align*}\n \t\\Delta x^p &\\sqrt{\\max_{m=1,\\dots,M} \\frac{2 N_0^p N^p}{\\epsilon_m} E_m + \\frac{N^{2p}}{\\epsilon_m^2} E_m^2}\\\\&\\quad \\leq S^p \\left( \\frac{\\Delta x_H}{\\Delta x_h} \\right)^{p} \\sqrt{\\max_{m=1,\\dots,M} \\frac{2 N_0^p}{\\epsilon_m} E_m + \\frac{1}{\\epsilon_m^2} E_m^2}.\n \\end{align*}\n The insertion of this estimation into \\fref{equ:first_sight} finally leads to the overall result\n \\begin{align*}\n &\\infnorm{(I - I_H^h I_h^H) e_h} \\leq \\tilde{C}(p) \\sqrt{N_0MN} N_0^{2p} \\Delta x^{p} \\infnorm{e_h}\\\\\n &\\quad\\quad\\quad\\quad\\quad\\quad + \\frac{(2\\pi)^p}{4p} \\left( \\frac{\\Delta x_H}{\\Delta x_h} \\right)^{p} \\left( \\sqrt{\\max_{m=1,\\dots,M} \\frac{2 N_0^p}{\\epsilon_m} E_m + \\frac{1}{\\epsilon_m^2} E_m^2} \\right) \\infnorm{e_h},\n \\end{align*}\n where we used the definition of $\\tilde{C}(p)$ to eliminate $S^p$ in the second summand. \n \n In this estimation, we can finally see that $\\Delta x^p$ does not dominate the second summand anymore. It is replaced by the relation between the step size on the coarse and on the fine level which can be regarded as constant. As a result, the remainder $E_m$ is now the dominating item in the term. Hence, a formulation like\n \\begin{align*}\n \t\\infnorm{(I - I_H^h I_h^H) e_h} &\\leq C_{11} \\Delta x^p \\infnorm{e_h} + C_{12}(E) \\infnorm{e_h}\n \\end{align*}\n with $E \\coloneqq (E_m)_{1 \\leq m \\leq M}$ is reasonable and concludes the proof.\n\\end{proof}\n\n\\begin{remark}\n The splitting of the sum and the consideration of the vector of remainders $E$ is needed since the constant $C_{11}$ would otherwise depend on $N^p$.\n Considering the relation $\\frac{N}{S} = \\frac{1}{\\Delta x}$, this would mean that the approximated boundary would not depend on $\\Delta x$ anymore. As a result, we would not obtain a better estimation than the simple one $\\infnorm{(I-I_H^h I_h^H)(e_h)} \\ \\le\\ C \\infnorm{e_h}$, which was already used in the proof of \\fref{theo:mlsdc_conv_gen}. By the applied split of the series, the term $N^p$ is replaced by $N_0^p$ which yields a more meaningful estimation as it keeps the dependence on the term $\\Delta x^p$ while adding another on the smoothness of the error.\n\\end{remark}\n\n\\begin{remark}\n Note that ideally the error only has a few, low-frequency Fourier coefficients, i.e. $\\tilde{e}_{m}(x)$ can be written as \n \\begin{align*}\n \\tilde{e}_m(x) \\coloneqq \\frac{1}{\\sqrt{N}} \\sum_{\\ell=0}^{N_0-1} c_{m,\\ell} \\exp{i \\frac{2\\pi}{S} \\ell x}, \\quad x \\in [0, S], \\quad m=1, \\dots, M,\n \\end{align*}\n using $N_0$ summands only.\n Then, $E_m = 0$ and the estimate\n \\begin{align*}\n \t\\infnorm{(I - I_H^h I_h^H) e_h} &\\leq C_{11} \\Delta x^p \\infnorm{e_h} + C_{12}(E) \\infnorm{e_h}\n \\end{align*}\n reduces to\n \\begin{align*}\n \t\\infnorm{(I - I_H^h I_h^H) e_h} &\\leq C_{11} \\Delta x^p \\infnorm{e_h}.\n \\end{align*}\n\\end{remark}\n\nThe following theorem uses \\fref{lem:ihh_gen} to extend \\fref{theo:mlsdc_conv_gen}. In particular, the provided estimation for $\\infnorm{(I - I_H^h I_h^H) (U_h - U_h^{(k)})}$ is used in the corresponding proof which results in a new convergence theorem for MLSDC.\n \n\\begin{theorem}\n\t\\label{theo:mlsdc_conv_prec}\n Consider a generic initial value problem like (\\ref{equ:ivp}) with a Lipschitz-continuous function $f$ on the right-hand side. Furthermore, let the conditions of \\fref{lem:ihh_gen} be met.\n \n \n \n Then, if the step size $\\Delta t$ is sufficiently small, MLSDC converges linearly to the solution of the collocation problem with a convergence factor in $\\mathcal{O}((\\Delta x^p + C(E)) \\Delta t + \\Delta t^2)$, i.e. the following estimate for the error is valid:\n \\begin{align}\n \t \\label{equ:mlsdc_conv_prec_iter}\n \\infnorm{U_h - U_h^{(k)}} \\leq ((C_{13} \\Delta x^p + C_{14}(E)) \\Delta t + C_{15} \\Delta t^2) \\infnorm{U_h - U_h^{(k-1)}},\n \\end{align}\n where $\\Delta x \\equiv \\Delta x_H$ is defined as the resolution in space on the coarse level $\\Omega_H$ of MLSDC and the constants $C_{13}$, $C_{14}(E)$ and $C_{15}$ are independent of $\\Delta t$.\n \t\n \tIf, additionally, the solution of the initial value problem $u$ is $(M+1)$-times continuously differentiable, the LTE of MLSDC compared to the solution of the ODE can be bounded by:\n \t\\begin{align}\\begin{split}\n \t \\label{equ:mlsdc_conv_prec_lte}\n\t\t\\infnorm{\\bar{U_h} - U_h^{(k)}} &\\leq C_{17} \\Delta t^{M+1} \\norm{u}_{M+1}\\\\\n\t\t&\\quad+ \\sum_{l=0}^{k} C_{l+18} (\\Delta x^p + C_{14}(E))^{k-l} \\Delta t^{k_0+k+l} \\norm{u}_{k_0+1}\n\t\n\t\n\t\n\t\\end{split}\\end{align}\n\twhere the constants $C_{17}, \\dots, C_{k+18}$ are independent of $\\Delta t$, $k_0$ denotes the approximation order of the initial guess $U_h^{(0)}$ and $\\norm{u}_p$ is defined by $\\infnorm{u^{(p)}}$.\n\\end{theorem}\n\\begin{proof}\n\tThe proof is similar to the one of \\fref{theo:mlsdc_conv_gen} but differs in the used estimation for $\\infnorm{(I - I_H^h I_h^H) (U_h - U_h^{(k)})}$. Here, \\fref{lem:ihh_gen} instead of the simple norm compatibility inequality is used for this purpose. Based on the estimations (\\ref{equ:u-1_rek_u-half}), (\\ref{equ:u-half_Ihh}) and (\\ref{equ:u-half_H}), arising in the proof of the mentioned theorem, it follows\n\t\\begin{align*}\n \t\\infnorm{U_h - U_h^{(k+1)}} \\leq \\tilde{C}_1 \\Delta t \\infnorm{(I - I_H^h I_h^H) (U_h - U_h^{(k)})} + C_{15} \\Delta t^2 \\infnorm{U_h - U_h^{(k)}}.\n\t\\end{align*}\n As already mentioned, we will now apply \\fref{lem:ihh_gen}, namely\n \\begin{align*}\n \\infnorm{(I - I_H^h I_h^H)(U_h - U_h^{(k)})} \\leq (C_{11} \\Delta x^p + C_{12}(E)) \\infnorm{U_h - U_h^{(k)}},\n \\end{align*}\n which yields\n \\begin{align*}\n \t\\infnorm{U_h - U_h^{(k+1)}} &\\leq (C_{13} \\Delta x^p + C_{14}(E)) \\Delta t \\infnorm{U_h - U_h^{(k)}}\\\\\n \t&\\quad + C_{15} \\Delta t^2 \\infnorm{U_h - U_h^{(k)}}\\\\\n \t&= ((C_{13} \\Delta x^p + C_{14}(E)) \\Delta t + C_{15} \\Delta t^2) \\infnorm{U_h - U_h^{k}}.\n \\end{align*}\n This concludes the proof of \\fref{equ:mlsdc_conv_prec_iter}.\n \n\tThe proof of \\fref{equ:mlsdc_conv_prec_lte} is again similar to the one of the second equation in \\fref{theo:sdc_conv}, using the previous result. Additionally, the binomial theorem is applied to simplify the arising term $((C_{13} \\Delta x^p + C_{14}(E)) \\Delta t + C_{15} \\Delta t^2)^k$.\n\\end{proof}\n\n\\begin{remark}\n It is here that a possible dependency of $f$'s Lipschitz constant on $\\Delta x$ plays a key role.\n Similar to the observations before, we find that in this case \\fref{equ:mlsdc_conv_prec_iter} needs to be replaced with\n \\begin{align*}\n \\infnorm{U_h - U_h^{(k)}} \\leq ((C_{13} \\Delta x^{p} + C_{14}&(E))C(\\delta^{-1}) \\Delta t \\\\&+ C_{15}(\\delta^{-2}) \\Delta t^2) \\infnorm{U_h - U_h^{(k-1)}},\n \\end{align*}\n where $\\delta$ denotes the difference between spatial and temporal resolution (up to constants) and comes from the initial step size restriction of SDC.\n The term $C_{13} \\Delta x^{p} + C_{14}(E)$ itself does not depend on $\\delta$, since it comes from the remainder of the interpolation estimate in lemma~\\ref{lem:ihh_gen}, where $f$ and therefore its Lipschitz constant as well as the step size does not play a role.\n As before, \\fref{equ:mlsdc_conv_prec_lte} has to be modified, now including the term $\\delta^{-(k-l+1)}$ in the sum. \n The constant $C_{17}$ is still independent of $\\delta$.\n\\end{remark}\n\n\\begin{remark}\n\tSimilar to \\fref{cor:conv_last}, it can be proven that the order limit $M+1$ in \\fref{theo:mlsdc_conv_prec} can be replaced by $2M$ if only the error at the last collocation node is considered.\n\\end{remark}\n\nThe theorem states that, under the named conditions, MLSDC converges linearly with a convergence rate of $\\mathcal{O}((\\Delta x^p + C(E)) \\Delta t + \\Delta t^2)$ to the collocation solution if $(C_{13} \\Delta x^p + C_{14}(E)) \\Delta t + C_{15} \\Delta t^2 < 1$. This means that if $\\Delta x^p$ and the vector of remainders $E$ are sufficiently small, the error of MLSDC decreases by two orders of $\\Delta t$ with each iteration, which indeed represents an improved convergence behavior compared to the one described in \\fref{theo:mlsdc_conv_gen}. Otherwise, i.e.\\ if $\\Delta x^p$ and $E$ are not that small, it only decreases by one order in $\\Delta t$ which is equivalent to the result of the previous theorem. \n\nIn the second equation of the theorem, it can be seen that again, $\\Delta x^p$ and $E$ are the crucial factors here. If they are small enough such that $\\Delta t^{k_0 + 2k}$ is the leading order, MLSDC converges with order $\\min(k_0+2k-1, 2M-1)$ and thus gains two orders per iteration. Otherwise, the convergence order is only $\\min(k_0+k-1, 2M-1)$, i.e.\\ the error decreases by one order in $\\Delta t$ in each iteration. \n\nNote that, as a result, it is advisable to use a high interpolation order $p$ and a small spatial step size $\\Delta x$ on the coarse level in practical applications of MLSDC. This theoretical result matches the numerical observations described in \\cite{mlsdc-2}. In section 2.2.5 of this paper, it is mentioned that the convergence properties of MLSDC seem to be highly dependent on the used interpolation order and resolution in space. Moreover, it was said that in the considered numerical examples a high resolution in space, i.e.\\ a small $\\Delta x$, led to a lower sensitivity on the interpolation order $p$. Our theoretical investigation provides an explanation for this behavior.\n\nIt seems reasonable to use a similar approach to determine the conditions for a higher convergence order of MLSDC if coarsening in time instead of space is used. Analogous to \\fref{equ:interpol_error} in the proof of \\fref{lem:ihh_gen}, the Lagrangian error estimation could be used for this purpose, resulting in the following estimation\n\\begin{align}\n \\label{equ:err_coarse_time}\n \\infnorm{(I- I_H^h I_h^H)(U_h - U_H^{(k)})} \\leq \\frac{\\Delta \\tau^p}{4p} \\infnorm{e^{(p)}(t)},\n\\end{align}\nwhere $I_H^h$ and $I_h^H$ denote temporal transfer operators now and $e(t)$ is defined as the continuous error of MLSDC compared to the collocation solution. In this case, however, the function $e(t)$ is implicitly known and thus does not have to be approximated by an iDFT. In particular, it is a polynomial of degree $M \\equiv M_h$ as both the collocation solution $U$ and each iterate $U_h^{(k)}$ of MLSDC are polynomials of that degree, respectively. This can be seen by considering that $\\Delta t Q F(U)$ as well as $\\Delta t Q_\\Delta F(U)$ essentially represent a sum of integrals of Lagrange polynomials which apparently results in a polynomial. Consequently, the $p$-th derivative of $e(t)$ is a polynomial of degree $M_h-p$. The maximal interpolation order $p$ is the number of collocation nodes $M_H$ on the coarse level. Here, we will assume that $p = M_H$, i.e. the maximal interpolation order is used. Note that $e^{(p)}(t) = 0$ for $p > M_h$ and hence $\\infnorm{(I- I_H^h I_h^H)(U_h - U_H^{(k)})} = 0$ for $M_H = M_h$ which is consistent with the expected behavior as it means that no coarsening is used at all.\nAs a conclusion, it can be said that, according to \\fref{equ:err_coarse_time} the improved convergence behavior of MLSDC using coarsening in time is dependent on the used time step size $\\Delta \\tau = C \\Delta t$ and the number of collocation nodes on the coarse level $M_H$. Note that it is also dependent on the specific coefficients of $e^{(p)}(t)$. However, these are highly dependent on the right-hand side $f$ of the IVP and thus cannot be controlled by any method parameters.\n\nIn summary, two convergence theorems for MLSDC were established in this section. While the first one, \\fref{theo:mlsdc_conv_gen}, represents a general statement on the convergence of the method, the second one, \\fref{theo:mlsdc_conv_prec}, provides theoretically established guidelines for the parameter choice in practical applications of MLSDC in order to achieve an improved convergence behavior of the method. In the next section, we will examine numerical examples of MLSDC to check if the resulting errors match those theoretical predictions. \n\\section{Conclusions and Outlook}\n\n\nIn this paper, we established two convergence theorems for multi-level spectral deferred correction (MLSDC) methods, using similar concepts and ideas as those presented in \\cite{tang} for the proof of the convergence of SDC. In the first theorem, namely \\fref{theo:mlsdc_conv_gen}, it was shown that with each iteration of MLSDC the error compared to the solution of the initial value problem decreases by at least one order of the chosen step size $\\Delta t$, limited by the accuracy of the underlying collocation solution. The corresponding theorem only requires the operator on the right-hand side of the considered initial value problem to be Lipschitz-continuous, not necessarily linear, and the chosen time step size $\\Delta t$ to be sufficiently small.\nConsequently, we found a first theoretical convergence result for MLSDC proving that it converges as good as SDC does. However, we would expect and numerical results already indicated that the additional computations on the coarse level, more specifically the SDC iterations performed there, would lead to an improved convergence behavior of the method. \n\nFor that reason, we analyzed the errors in greater detail, leading to a second theorem on the convergence of MLSDC, namely \\fref{theo:mlsdc_conv_prec}. Here, we focused on a specific coarsening strategy and transfer operators. In particular, we considered MLSDC using coarsening in space with Lagrangian interpolation. Given these assumptions, we could prove that, if particular conditions are met, the method can even gain two orders of $\\Delta t$ in each iteration until the accuracy of the collocation problem is reached. This consequently led us to theoretically established guidelines for the parameter choice in practical applications of MLSDC in order to achieve the described improved convergence behavior of the method. More specifically, the corresponding theorem says that for this purpose the spatial grid size on the coarse level has to be small, the interpolation order has to be high and the errors have to be smooth.\nWe presented numerical examples which confirm these theoretical results. In particular, it could be observed that the change of one of those crucial parameters immediately led to a decrease in the order of accuracy. Essentially, it resulted in a convergence behavior as it was described in the first presented theorem.\n\nBesides, there are several open questions related to the presented work which have not yet been investigated. Three of them are briefly discussed here.\n\n\\textbf{More information, better results.}\nThe results presented here are quite generic.\nAs a consequence, since we only assume Lipschitz continuity of the right-hand side of the ODE and do not pose conditions on the SDC preconditioner, both constants and step size restrictions are rather pessimistic.\nUsing more knowledge of the right-hand side or the matrix $Q_\\Delta$ will yield better results, as it already did for SDC.\nSince the goal of this paper is to establish a baseline for convergence of MLSDC, exploiting this direction, especially with respect to the treatment of convergence in the stiff limit as done in~\\cite{sdc-lu} for SDC, is left for future work.\n\n\\textbf{Smoothness of the error.}\nThe second theorem, describing conditions for an improved convergence behavior of MLSDC, has a drawback regarding its practical significance. \nThe way \\fref{theo:mlsdc_conv_prec} is currently proven requires a smooth error after its periodic extension.\nThis occurring condition of a smooth error does not always apply and is in particular not easy to control. \nEssentially, something like a smoothing property would be needed to ensure that the error always becomes smooth after enough iterations. Numerical results indicate that this property apparently does not hold for SDC, though~\\cite{conv-pfasst}. In this context, however, it would be sufficient if we could at least control this condition, i.e.\\ derive particular criteria for the parameters of the method ensuring the errors to be smooth. The numerical examples presented in \\fref{sec:examples} particularly lead to the assumption that the selection of a smooth initial guess $U^{(0)}$ would result in smooth errors for $U^{(k)}$, $k \\geq 1$, at least for a particular set of problems. \n\n\\textbf{Other extensions of SDC.}\nFurthermore, it could be tried to adapt the presented convergence proofs of MLSDC to other extensions and variations of SDC, as for example the parallel-in-time method PFASST (Parallel Full Approximation Scheme in Space and Time)~\\cite{pfasst} or general semi-implicit and multi-implicit formulations of SDC (SISDC/MISDC)~\\cite{imex-1,imex-2}. Whereas an adaptation to SISDC and MISDC methods seems to be rather straightforward \\cite{causley}, we found that the application of similar concepts and ideas to prove the convergence of PFASST may involve some difficulties. In particular, the coupling of the different time steps, i.e.\\ the use of the approximation at the endpoint of the last subinterval for the start point of the next one, could cause a problem in this context since the corresponding operator is independent of $\\Delta t$ and would thus add a constant term to our estimations. \n\n\\section{Spectral Deferred Corrections}\n\\label{sec:sdc}\n\nIn the following SDC is presented as preconditioned Picard iterations for the collocation problem. The used approach and notations are substantially based on \\cite{mlsdc-1,conv-pfasst} and references therein. First, the collocation problem for a generic initial value problem is explained. Then, SDC is described as a solver for this problem and compact notations are introduced. Finally, an existing theorem on the convergence of SDC, including its proof, is presented. \n\n\\subsection{SDC and the collocation problem}\n\nConsider the following autonomous initial value problem (IVP)\n\\begin{align}\\begin{split}\n \\label{equ:ivp}\n u'(t) &= f(u(t)), \\quad t\\in[t_0, t_1],\\\\\n u(t_0) &= u_0\n\\end{split}\\end{align}\nwith $u(t), u_0 \\in \\mathbb{C}^N$ and $f:\\mathbb{C}^N\\to\\mathbb{C}^N$, $N\\in\\mathbb{N}$. To guarantee the existence and uniqueness of the solution, $f$ is required to be Lipschitz continuous. Since a high-order method shall be used, $f$ is additionally assumed to be sufficiently smooth.\n\nThe IVP can be written as\n\\begin{align*}\n u(t) = u_0 + \\int_{t_0}^t f(u(s))ds, \\quad t\\in[t_0, t_1]\n\\end{align*}\nand choosing $M$ quadrature nodes $\\tau_1, ..., \\tau_M$ within the time interval such that $t_0 \\le \\tau_1 < \\tau_2 < ... < \\tau_M = t_1$, the integral is now approximated using a spectral quadrature rule like Gauß-Radau. \nThis approach results in the discretized system of equations\n\\begin{align}\n \\label{equ:intgl1}\n u_m = u_0 + \\Delta t \\sum_{j=1}^M q_{m,j} f(u_j), \\quad m=1,...,M,\n\\end{align}\nwhere $u_m \\approx u(\\tau_m)$, $\\Delta t = t_1 - t_0$ denotes the time step size and $q_{m,j}$ represent the quadrature weights for the unit interval with\n\\begin{align*}\n q_{m,j} = \\frac{1}{\\Delta t} \\int_{t_0}^{\\tau_m} l_j(s) ds.\n\\end{align*}\nHere, $l_j$ represents the $j$-th Lagrange polynomial corresponding to the set of nodes $(\\tau_m)_{1 \\leq m \\leq M}$.\nWe can combine these $M$ equations into the following system of linear or non-linear equations, defining the collocation problem:\n\\begin{gather}\n\tC(U) \\coloneqq (I_{MN} - \\Delta t (Q \\otimes I_N) F)(U) = U_0,\n\t\\label{equ:coll_prob}\n\\end{gather}\nwhere $U \\coloneqq (u_1, u_2, \\dots, u_M)^T \\in \\mathbb{C}^{MN}$, $U_0 \\coloneqq (u_0, u_0, \\dots, u_0)^T \\in \\mathbb{C}^{MN}$, $Q \\coloneqq (q_{m,j})_{1 \\leq m,j \\leq M}$ is the matrix gathering the quadrature weights, the vector function $F$ is given by $F(U) \\coloneqq (f(u_1), f(u_2), \\dots, f(u_M))^T$ and $I_{MN}, I_N$ are the identity matrices of dimensions $MN$ and $N$.\n\nAs described above, the solution $U$ of the collocation problem approximates the solution of the initial value problem (\\ref{equ:ivp}). With this in mind, the following theorem, referring to \\cite[Thm.~7.10]{hairer-wanner-1}, provides a statement on its order of accuracy.\n\n\\begin{theorem}\n\t\\label{theo:coll_prob_err}\n\tThe solution $U = (u_1, u_2, \\dots, u_M)^T \\in \\mathbb{C}^{MN}$ of the collocation problem defined by \\fref{equ:coll_prob} approximates the solution $u$ of the IVP \\eqref{equ:ivp} at the collocation nodes. In particular, for $\\bar{U} \\coloneqq (u(\\tau_1), \\dots, u(\\tau_M))^T$ the following error estimation applies:\n\t\\begin{align*}\n \n \t\\| \\bar{U} - U \\|_\\infty \\leq C_1 \\Delta t^{M+1} \\|u\\|_{M+1},\n\t\\end{align*}\n\twhere $C_1$ is independent of $\\Delta t$, $M$ denotes the number of nodes and $\\|u\\|_{M+1}$ represents the maximum norm of $u^{(M+1)}$, the $(M+1)$th derivative of $u$.\n\\end{theorem}\n\nInterpreting the collocation problem as a discretization method with discretization parameter $n \\coloneqq \\Delta t^{-1}$, the theorem shows that the discrete approximation $U$ defined by the collocation problem converges with order $M+1$ to the solution $\\bar{U}$ of the corresponding IVP.\n\nSince the system of equations \\eqref{equ:coll_prob} defining the collocation problem is naturally dense as the matrix $Q$ gathering the quadrature weights is fully populated, a direct solution is not advisable, in particular if the right-hand side of the ODE is non-linear. An iterative method to solve the problem is SDC. \n\nThe standard Picard iteration for the collocation problem \\eqref{equ:coll_prob} is given by\n\\begin{align}\\begin{split}\n \\label{equ:richardson}\n U^{(k+1)} &= U^{(k)} + (U_0 - C(U^{(k)}))\\\\\n &= U_0 + \\Delta t (Q \\otimes I_N)F(U^{(k)}).\n\\end{split}\\end{align}\n\nAs this method only converges for very small step sizes $\\Delta t$, using a preconditioner to increase range and speed of convergence is reasonable. The SDC-type preconditioners are defined by\n\\begin{align*}\n \n P(U) = (I_{MN} - \\Delta t (Q_\\Delta \\otimes I_N) F) (U),\n\\end{align*}\nwhere the matrix $Q_\\Delta = (q_{\\Delta_{m,j}})_{1 \\leq m,j \\leq M} \\approx Q$ is formed by the use of a simpler quadrature rule. In particular, $Q_\\Delta$ is typically a lower triangular matrix, such that solving the system can be easily done by forward substitution.\n\nCommon choices for $Q_\\Delta$ include the matrix\n\\begin{align*}\n \n Q_\\Delta = \\frac{1}{\\Delta t}\n \\begin{pmatrix}\n \\Delta \\tau_1 \\\\\n \\Delta \\tau_1 & \\Delta \\tau_2 \\\\\n \\vdots & \\vdots & \\ddots \\\\\n \\Delta \\tau_1 & \\Delta \\tau_2 & \\hdots & \\Delta \\tau_M\n \\end{pmatrix}\n\\end{align*}\nwith $\\Delta \\tau_m = \\tau_m - \\tau_{m-1}$ for $m=2,...,M$ and $\\Delta \\tau_1 = \\tau_1 - t_0$ representing the right-sided rectangle rule. Similarly, the left-sided rectangle rule \\cite{sdc-expl-euler} or a part of the $LU$ decomposition of the matrix $Q$ \\cite{sdc-lu} are chosen. The theoretical considerations in the next chapters do not rely on a specific matrix $Q_\\Delta$. However, in the numerical examples the right-sided rectangle rule as given above is used.\n\nBy the use of such an operator to precondition the Picard iteration (\\ref{equ:richardson}), the following iterative method for solving the collocation problem is obtained\n\\begin{align}\\begin{split}\n \\label{equ:sdc}\n (I_{MN} - \\Delta t (Q_\\Delta \\otimes I_N)F) (U^{(k+1)})\n = U_0 + \\Delta t ((Q-Q_\\Delta) \\otimes I_N)F(U^{(k)}),\n\\end{split}\\end{align}\nwhich constitutes the SDC iteration \\cite{sdc-orig,huang}. \nWritten down line-by-line, this formulation recovers the original SDC notation given in~\\cite{sdc-orig}.\nA more implicit formulation is given by\n\\begin{align}\\begin{split}\n \\label{equ:sdc_implicit}\n U^{(k+1)} = U_0 + \\Delta t (Q_\\Delta \\otimes I_N)F(U^{(k+1)}) + \\Delta t ((Q-Q_\\Delta) \\otimes I_N)F(U^{(k)})\n\\end{split}\\end{align}\nand this will be used for the following convergence considerations.\n\n\\subsection{Convergence of SDC}\n\nThere already exist several approaches proving the convergence of SDC, particularly those presented in \\cite{causley,hagstrom,hansen,huang,xia,tang}. Here, we will focus on the idea of the proof from \\cite{tang} as it uses the previously introduced matrix formulation of SDC, needed for an appropriate adaptation for a convergence proof of MLSDC, and simultaneously, provides a general result for linear and non-linear initial value problems. \nWe will review the idea of this proof in some detail to introduce the notation and the key ideas.\nThis is followed by a concise discussion on stability and convergence of SDC in the sense of one-step ODE solvers.\n \n\nThe approach in \\cite{tang} relies on a split of the local truncation error (LTE). The key concept used in the proof is a property of the operators $QF(U)$ and $Q_\\Delta F(U)$, respectively, which can be interpreted as a kind of extended Lipschitz continuity. It is presented in the following lemma using the previously introduced notations. For reasons of readability, the sizes of the identity matrices are no longer denoted here.\n \n\\begin{lemma}\n \\label{lem:estimation}\n If $f:\\mathbb{C}^N \\to \\mathbb{C}^N$ is Lipschitz continuous, the following estimates apply\n \\begin{align*}\\begin{split}\n \n \\infnorm{\\Delta t (Q \\otimes I) (F(U_1) - F(U_2))}\n &\\leq C_2 \\Delta t \\infnorm{U_1 - U_2},\\\\\n \\infnorm{\\Delta t (Q_\\Delta \\otimes I) (F(U_1) - F(U_2))}\n &\\leq C_3 \\Delta t \\infnorm{U_1 - U_2},\n \\end{split}\\end{align*}\n where the constants $C_2$ and $C_3$ are dependent on the Lipschitz constant $L$, but independent of $\\Delta t$ and $U_1, U_2 \\in \\mathbb{C}^{NM}$.\n\\end{lemma}\n\\begin{proof}\n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n Can be shown directly using the definition of the maximum norm, the Lipschitz continuity of $f$ and the compatibility between maximum absolute row sum norm for matrices and maximum norm for vectors.\n \n\\end{proof}\n\n\\begin{remark}\n For a system of ODEs stemming from a discretized PDE, the constants $C_2$ and $C_3$ may depend on the spatial resolution given by some grid spacing $\\Delta x$, because the Lipschitz constant of $f$ may depend on it.\n In this case we have $C_2 = C_2(\\Delta x^{-d})$, $C_3 = C_3(\\Delta x^{-d})$ for $d\\in\\mathbb{N}$.\n For example, using second-order finite differences in space for the heat equation results in the ODE system $u' = Au$ with matrix $A \\in \\mathcal{O}(\\Delta x^{-2})$, i.e. $d=2$ in this case.\n This has to be kept in mind for most of the upcoming results and we will address this point separately in remarks where appropriate.\n This will be particularly relevant for the convergence results in section~\\ref{ssec:improved_mlsdc}, where the spatial discretization plays a key role.\n Note, however, that this is a rather pessimistic estimate. \n When focusing on spatial operators with more restrictive properties (e.g.~linearity) or when using a specific matrix $Q_\\Delta$, the convergence results can be improved substantially, both in terms of constants and time-step size restrictions.\n For SDC, this has been already done, see e.g.~\\cite{sdc-lu}.\n\\end{remark}\n\nThe following theorem provides a convergence statement for SDC using the presented lemma in the proof.\n\n\\begin{theorem}\n \t\\label{theo:sdc_conv}\n Consider a generic initial value problem like (\\ref{equ:ivp}) with a Lipschitz-continuous function $f$ on the right-hand side. \n \n If the step size $\\Delta t$ is sufficiently small, SDC converges linearly to the solution $U$ of the collocation problem with a convergence rate in $\\mathcal{O}(\\Delta t)$, i.e. the following estimate for the error of the $k$-th iterated $U^{(k)}$ of SDC compared to the solution of the collocation problem is valid:\n \t\\begin{align}\n \t \\label{equ:sdc_conv_iter}\n \t\t\\infnorm{U-U^{(k)}} &\\leq C_4 \\Delta t \\infnorm{U - U^{(k-1)}}\n \t\\end{align}\n \twhere the constant $C_4$ is independent of $\\Delta t$.\n \t\n \tIf, additionally, the solution of the initial value problem $u$ is $(M+1)$-times continuously differentiable, the LTE of SDC compared to the solution $\\bar{U}$ of the ODE can be bounded by\n \t\\begin{align}\\begin{split}\n \t \\label{equ:sdc_conv_lte}\n \\infnorm{\\bar{U} - U^{(k)}} &\\leq C_5 \\Delta t^{k_0+k} \\|u\\|_{k_0+1} \n + C_6 \\Delta t^{M+1} \\|u\\|_{M+1}\\\\\n &= \\mathcal{O}(\\Delta t^{\\min(k_0+k, M+1)}),\n \\end{split}\\end{align}\n where the constants $C_5$ and $C_6$ are independent of $\\Delta t$, $k_0$ denotes the approximation order of the initial guess $U^{(0)}$ and $\\|u\\|_p$ is defined by $\\infnorm{u^{(p)}}$.\n\\end{theorem}\n\\begin{proof}\n \tWe again closely follow \\cite{tang} here. According to the definition of the collocation problem (\\ref{equ:coll_prob}) and an SDC iteration (\\ref{equ:sdc_implicit}), it follows\n \t\\begin{align*}\n \t\t\\infnorm{U-U^{(k)}} &= \\infnorm{\\Delta t (Q \\otimes I) (F(U) - F(U^{(k-1)}))\\\\\n \t\t\t&\\quad\\quad+ \\Delta t (Q_\\Delta \\otimes I) (F(U^{(k-1)}) - F(U^{(k)}))}.\n \t\\end{align*}\n \tTogether with the triangle inequality and \\fref{lem:estimation}, we obtain\n \t\\begin{align*}\n \t\t\\infnorm{U-U^{(k)}} &\\leq C_2 \\Delta t \\infnorm{U - U^{(k-1)}} + C_3 \\Delta t \\infnorm{U^{(k-1)} - U^{(k)}}.\n \t\\end{align*}\n \tApplying the triangle inequality again yields\n \t\\begin{align*}\n \t\t\\infnorm{U-U^{(k)}}\n \t\t&\\leq \\tilde{C}_1 \\Delta t \\infnorm{U - U^{(k-1)}} + C_3 \\Delta t \\infnorm{U - U^{(k)}},\n \t\\end{align*}\n \twhere here and in the following, we use variables in the form of $\\tilde{C}_i$ to denote temporary arising constants. We continue by subtracting $C_3 \\Delta t \\infnorm{U - U^{(k)}}$ from both sides and dividing by $1 - C_3 \\Delta t$ which results in\n \\begin{align*}\n \\infnorm{U-U^{(k)}} &\\leq \\frac{\\tilde{C}_1}{1-C_3 \\Delta t} \\Delta t \\infnorm{U - U^{(k-1)}}.\n \\end{align*}\n \tIf the step size is sufficiently small, in particular \n \t\\begin{align}\\label{eq:conv_cond}\n \t C_3 \\Delta t < 1,\n \t\\end{align} the following estimate is valid\n \t\\begin{align}\\label{eq:conv_cond_II}\n \t\t\\frac{\\tilde{C}_1}{1-C_3 \\Delta t} \\leq C_4,\n \t\\end{align}\n which concludes the proof for \\fref{equ:sdc_conv_iter}.\n \n Continuing with recursive insertion, we get\n \\begin{align*}\n \\infnorm{U-U^{(k)}} \\leq \\tilde{C}_2 \\Delta t^k \\infnorm{U - U^{(0)}}.\n \\end{align*}\n Since $U^{(0)}$ is assumed to be an approximation of $k_0$-th order, we further know that\n \\begin{align*}\n \\norm{\\bar{U} - U^{(0)}} \\leq \\tilde{C}_3 \\Delta t^{k_0} \\norm{u}_{k_0+1}.\n \\end{align*}\n This estimation together with the triangle inequality and the error estimation for the solution of the collocation problem stated in \\fref{theo:coll_prob_err} yields\n \\begin{align}\n \\infnorm{U-U^{(k)}} &\\leq \\tilde{C}_2 \\Delta t^k (\\infnorm{\\bar{U} - U} + \\infnorm{\\bar{U} - U^{(0)}}) \\nonumber\\\\\n \\label{equ:proof_sdc_err_coll}\n &\\leq \\tilde{C}_4 \\Delta t^{M+k+1} \\|u\\|_{M+1} + C_6 \\Delta t^{k_0+k} \\|u\\|_{k_0+1}.\n \\end{align}\n Altogether, it follows\n \\begin{align*}\n \\infnorm{\\bar{U} - U^{(k)}} &\\leq \\infnorm{\\bar{U} - U} + \\infnorm{U - U^{(k)}}\\\\\n &\\leq C_1 \\Delta t^{M+1} \\|u\\|_{M+1} + \\tilde{C}_4 \\Delta t^{M+k+1} \\|u\\|_{M+1} + C_6 \\Delta t^{k_0+k} \\|u\\|_{k_0+1}\\\\\n &= (C_1 + \\tilde{C}_4 \\Delta t^k) \\Delta t^{M+1} \\|u\\|_{M+1} + C_6 \\Delta t^{k_0+k} \\|u\\|_{k_0+1}.\n \\end{align*}\n Since the step size $\\Delta t$ is assumed to be sufficiently small, i.e.\\ bounded above, the following estimate is valid\n \\begin{align*}\n \tC_1 + \\tilde{C}_4 \\Delta t^k \\leq C_5,\n \\end{align*}\n which finally concludes the proof for \\fref{equ:sdc_conv_lte}.\n\\end{proof}\n\n\n\\begin{remark}\\label{rem:dx_sdc}\n Note that if the right-hand side of the ODE comes from a discretized PDE with a given spatial resolution, no additional restriction is posed.\n In this case, $C_3 = C_3(\\Delta x^{-d}) = \\tilde{C}_5\\Delta x^{-d}$ for some constant $\\tilde{C}_5$, so that the condition~\\eqref{eq:conv_cond} becomes $\\tilde{C}_5\\Delta t < \\Delta x^{d}$.\n Since we did not specify $Q_\\Delta$ here, the SDC iterations can be explicit or implicit and it is natural to obtain such a restriction on the time-step size.\n Similar restrictions can be found in other convergence results for SDC, see e.g.~\\cite{causley,hagstrom,huang}.\n Condition~\\eqref{eq:conv_cond_II} then translates to\n \t\\begin{align*}\n \t\tC_4 \\geq \\frac{\\tilde{C}_1}{1-C_3 \\Delta t} = \\frac{\\tilde{C}_6\\Delta x^{-d}}{1-\\tilde{C}_5\\Delta x^{-d} \\Delta t} = \\frac{\\tilde{C}_6}{\\Delta x^d-\\tilde{C}_5 \\Delta t}\n \t\\end{align*}\n \tfor some constant $\\tilde{C}_6$.\n \tAssuming a fixed distance between $\\Delta x^p$ and $\\tilde{C}_5 \\Delta t$ with $\\Delta x^d-\\tilde{C}_5 \\Delta t = \\delta$ we can write $C_4 = C_4(\\delta^{-1})$ to indicate the dependence of $C_4$ on this distance and not $\\Delta x^p$ or $\\Delta t$ alone.\n \tThus, $C_4$ does not pose an additional restriction to the convergence of SDC.\n \tNote that for $C_5$ we have $C_5 = C_5(\\delta^{-(k+1)})$, so that the choice of $\\delta$ can increase the constant in front of the $\\Delta t^{k_0+k}$-term quite substantially, but it does not affect the $\\Delta t^{M+1}$-term coming from the collocation problem itself. \n\\end{remark}\n\n\nThe theorem can be read as a convergence statement for SDC. In particular, the first estimation (\\ref{equ:sdc_conv_iter}) shows that SDC, interpreted as an iterative method to solve the collocation problem, converges linearly to the solution of the collocation problem with a convergence rate of $\\mathcal{O}(\\Delta t)$ if $C_4\\Delta t < 1$. The second part of the theorem, \\fref{equ:sdc_conv_lte} shows that SDC, in the sense of a discretization method, converges with order $\\min(k_0+k, M+1)$ to the solution of the initial value problem. In other words, the method gains one order per iteration, limited by the selected number of nodes used for discretization. \n\nWe can now immediately extend this result by looking at the right endpoint of the single time interval (which, in our case, is equal to the last collocation node). There, the convergence rate is limited not by the number of collocation nodes, but by the order of the quadrature.\n\n\\begin{corollary}\n\t\\label{cor:conv_last}\n\tConsider a generic initial value problem like (\\ref{equ:ivp}) with a Lipschitz-continuous function $f$ on the right-hand side. Furthermore, let the solution of the initial value problem $u$ be $(2M)$-times continuously differentiable. \n\t\n\tThen, if the step size $\\Delta t$ is sufficiently small, the error of the $k$-th iterated of SDC, defined by \\fref{equ:sdc_implicit}, at the last collocation node $u_M^{(k)}$, compared to the exact value at this point $u(\\tau_M)$, can be bounded by\n\t\\begin{align}\n\t\t\\label{equ:conv_last_point} \n \\begin{split}\n \\infnorm{u(\\tau_m) - u_M^{(k)}} &\\leq C_7 \\Delta t^{2M} \\max(\\|u\\|_{2M}, \\|u\\|_{M+1})\\\\\n &\\quad+ C_8 \\Delta t^{k_0+k} \\max(\\|u\\|_{k_0+1}, \\|u\\|_{M+1})\n \\end{split}\\\\\n &= \\mathcal{O}(\\Delta t^{\\min(k_0+k, 2M)}),\\nonumber\n\t\\end{align}\n\twhere the constants $C_7$ and $C_8$ are independent of $\\Delta t$, $k_0$ denotes the approximation order of the initial guess $U^{(0)}$ and $\\|u\\|_p$ is defined by $\\infnorm{u^{(p)}}$.\n\\end{corollary}\n\n\\begin{proof}\n The proof mainly relies on the interpretation of the solution of the collocation problem evaluated at the last node $\\tau_M$ as the result of a Radau method with $M$ stages. With this in mind, the well-known convergence, or in this case rather consistency, order of Radau methods yields the estimate \\cite{hairer-wanner-2}\n \\begin{align*}\n \n \t\\infnorm{u(\\tau_M) - u_M} \\leq \\tilde{C}_1 \\Delta t^{2M} \\|u\\|_{2M},\n \\end{align*}\n where $\\tilde{C}_1$ is independent of $\\Delta t$. Here and in the following, temporary arising constants will again be denoted by symbols like $\\tilde{C}_i$. However, they are separately defined and thus, do not correspond to the ones used in previous proofs.\n \n To use this estimation, we first have to apply the triangle inequality to the left-hand side of \\fref{equ:conv_last_point}, in particular\n \\begin{align}\n \t\\label{equ:proof_last_point_triangle}\n \t\\infnorm{u(\\tau_m) - u_M^{(k)}} &\\leq \\infnorm{u(\\tau_m) - u_M} + \\infnorm{u_M - u_M^{(k)}}.\n \\end{align}\n Then, with the definition of the vector $R \\coloneqq (0,\\dots,0,1) \\in \\mathbb{R}^{1 \\times M}$ which, multiplied with another vector, only captures its last value, the second term on the right-hand side of the above equation can be transferred to\n \\begin{align*}\n \\infnorm{u_M - u_M^{(k)}} &= \\infnorm{R U - R U^{(k)}} \\leq \\infnorm{R} \\infnorm{U - U^{(k)}} = \\infnorm{U - U^{(k)}}\\\\\n &\\le \\tilde{C}_2 \\Delta t^{M+k+1} \\norm{u}_{M+1} + C_6 \\Delta t^{k_0+k} \\norm{u}_{k_0+1},\n \\end{align*}\n where the last estimate comes from \\fref{equ:proof_sdc_err_coll} in the proof of \\fref{theo:sdc_conv}.\n Finally, by inserting all these results in \\fref{equ:proof_last_point_triangle}, we obtain\n \\begin{align*}\n \\infnorm{u(\\tau_m) - u_M^{(k)}} &\\leq \\infnorm{u(\\tau_m) - u_M} + \\infnorm{U - U^{(k)}}\\\\\n &\\leq \\tilde{C}_1 \\Delta t^{2M} \\|u\\|_{2M} + \\tilde{C}_2 \\Delta t^{M+k+1} \\|u\\|_{M+1} + C_5 \\Delta t^{k_0+k} \\|u\\|_{k_0+1}.\n \\end{align*}\n Note that the leading order of this term is essentially independent of the summand corresponding to $\\Delta t^{M+k+1}$. For $\\Delta t$ small enough, this result can be seen by a case analysis for $k$. For $k+1 \\geq M$, the considered summand is dominated by $\\Delta t^{2M} \\geq \\Delta t^{M+k+1}$ and thus can be disregarded in terms of leading order analysis. In the other case, i.e.\\ for $k+1< M$, the considered summand will, however, be greater than the one of order $2M$. Therefore, we will instead compare it to $\\Delta t^{k_0+k}$ in this case. Since the number of collocation nodes $M$ is usually chosen to be greater than the approximation order $k_0$ of the initial guess, the relation $\\Delta t^{k_0+k} \\geq \\Delta t^{M+k+1}$ applies and hence, ${k_0+k}$ will be the leading order for $k < M$. Thus, the considered summand $\\Delta t^{M+k+1}$ is again dominated by another term and can be disregarded concerning the overall asymptotic behavior. These considerations consequently lead to the following estimation\n \\begin{align*}\n \\infnorm{u(\\tau_m) - u_M^{(k)}} \\leq \\tilde{C}_3 \\Delta t^{2M} \\max(\\|u\\|_{2M}, \\|u\\|_{M+1}) + C_5 \\Delta t^{k_0+k} \\|u\\|_{k_0+1}\n \\end{align*}\n for $ k \\geq M$ and\n \\begin{align*}\n \\infnorm{u(\\tau_m) - u_M^{(k)}} \\leq \\tilde{C}_1 \\Delta t^{2M} \\|u\\|_{2M} + \\tilde{C}_4 \\Delta t^{k_0+k} \\max(\\|u\\|_{k_0+1}, \\|u\\|_{M+1})\n \\end{align*}\n for $k < M$, which can be combined to\n \\begin{align*}\n \\infnorm{u(\\tau_m) - u_M^{(k)}} &\\leq C_7 \\Delta t^{2M} \\max(\\|u\\|_{2M}, \\|u\\|_{M+1})\\\\\n &\\quad+ C_8 \\Delta t^{k_0+k} \\max(\\|u\\|_{k_0+1}, \\|u\\|_{M+1}),\n\t\\end{align*}\n concluding the proof.\n\\end{proof}\n\n\nWith this corollary, it can be concluded that SDC, in the sense of a single-step method to solve ODEs, is consistent of order $\\min(k_0+k-1, 2M-1)$. To extend this result towards a statement on the convergence order of the method an additional proof of its stability is needed.\nThe following theorem provides an appropriate result for SDC.\n\n\\begin{theorem}\n \\label{theo:sdc_stab}\n Consider a generic initial value problem like (\\ref{equ:ivp}) with a Lipschitz-continuous function $f$ on the right-hand side. \n If the step size $\\Delta t$ is sufficiently small and an appropriate initial guess is used, the SDC method, defined by \\fref{equ:sdc}, is stable.\n\\end{theorem}\n\\begin{proof}\n As usual for single-step methods, we will prove the Lipschitz continuity of the increment function of SDC in order to prove the stability of the method.\n \n A general single-step method is defined by the formula \n \\begin{align*}\n u_{n+1} = u_n + \\Delta t \\phi(u_n), \n \\end{align*}\n where $u_n$ denotes the approximation at the time step $t_n$ and $\\phi(u_n)$ is the increment function. Our aim now is to identify the specific increment function $\\phi$ corresponding to SDC and to, subsequently, show its Lipschitz continuity, i.e. prove the validity of $\\abs{\\phi(u_n) - \\phi(v_n)} \\leq L_\\phi \\abs{u_n - v_n}$.\n \n First, note that the $k$-th iterated of SDC $u_m^{(k)}$ at an arbitrary collocation node $\\tau_m$ ($1 \\leq m \\leq M$) can be written as\n \\begin{align*}\n u_m^{(k)} &= u_n + r_m^{(k)} \\mbox{ with}\\\\\n r_m^{(k)} &\\coloneqq \\Delta t (Q_\\Delta F(U_n+r^{(k)}))_m + \\Delta t ((Q - Q_{\\Delta}) F(U_n + r^{(k-1)}))_m,\\\\\n r^{(k)} &\\coloneqq (r_1^{(k)}, \\dots, r_M^{(k)})^T \\mbox{ and } U_n = (u_n, \\dots, u_n)^T\n \\end{align*}\n for $k\\ge1$ according to a line-wise consideration of \\fref{equ:sdc}, where the subscript $m$ denotes the $m$th line of the vectors. Consequently, the corresponding approximation at the time step $t_{n+1} = \\tau_M$ can be written as \n \\begin{gather*}\n u_{n+1} = u_M^{(k)} = u_n + r_M^{(k)} \\eqqcolon u_n + \\Delta t \\phi^{(k)}(u_n)\\\\\n \\mbox{with } \\phi^{(k)}(u_n) = \\frac{1}{\\Delta t} r_M^{(k)}.\n \\end{gather*}\n Hence, we have found an appropriate, albeit implicit definition for the increment function $\\phi^{(k)}$ of SDC.\n \n As a second step, it follows an investigation on the Lipschitz continuity of this function. For that, we start by noting that\n \\begin{align}\n \\label{equ:rel_phi_r}\n \\abs{\\phi^{(k)}(u_n) - \\phi^{(k)}(v_n)} &= \\frac{1}{\\Delta t} \\abs{r_M^{(k)} - s_M^{(k)}} \\leq \\frac{1}{\\Delta t} \\infnorm{r^{(k)} - s^{(k)}},\n \\end{align}\n where the $r$-terms belong to $u_n$ and the $s$-terms to $v_n$. Now, we will further analyze the term $\\norm{r^{(k)} - s^{(k)}}$. With the insertion of the corresponding definitions and an application of the triangle inequality, it follows\n \\begin{align*}\n \\norm{r^{(k)} - s^{(k)}} &\\leq \\norm{\\Delta t Q_\\Delta (F(U_n + r^{(k)}) - F(V_n + s^{(k)}))}\\\\\n &\\quad+ \\norm{\\Delta t (Q - Q_\\Delta) (F(U_n + r^{(k-1)}) - F(V_n + s^{(k-1)}))}.\n \\end{align*}\n The use of \\fref{lem:estimation} and a reapplication of the triangle inequality further yield\n \\begin{align*}\n \\norm{r^{(k)} - s^{(k)}} &\\leq \\tilde{C}_1 \\Delta t(\\abs{u_n - v_n} + \\norm{r^{(k)} - s^{(k)}} + \\norm{r^{(k-1)} - s^{(k-1)}}).\n \\end{align*}\n Continuing with the same trick as in the proof of \\fref{theo:sdc_conv}, namely a subtraction of $\\tilde{C}_1 \\Delta t \\norm{r^{(k)} - s^{(k)}}$ and a subsequent division by $1- \\tilde{C}_1 \\Delta t$, we get\n \\begin{align*}\n \\norm{r^{(k)} - s^{(k)}} &\\leq \\frac{\\tilde{C}_1}{1- \\tilde{C}_1 \\Delta t} \\Delta t (\\abs{u_n - v_n} + \\norm{r^{(k-1)} - s^{(k-1)}}).\n \\end{align*}\n If the step size $\\Delta t$ is sufficiently small, the following estimation applies\n \\begin{align*}\n \\frac{\\tilde{C}_1}{1- \\tilde{C}_1 \\Delta t} &\\leq \\tilde{C}_2\n \\end{align*}\n and a subsequent iterative insertion further yields\n \\begin{align*}\n \\norm{r^{(k)} - s^{(k)}} \\leq \\tilde{C}_3 \\sum_{l = 1}^{k} \\Delta t^l \\abs{u_n-v_n} + \\tilde{C}_4 \\Delta t^{k} \\norm{r^{(0)} - s^{(0)}}.\n \\end{align*}\n With the insertion of this result in \\fref{equ:rel_phi_r} above, it finally follows\n \\begin{align*}\n \\abs{\\phi^{(k)}(u_n) - \\phi^{(k)}(v_n)} \\leq C \\sum_{l = 0}^{k-1} \\Delta t^l \\abs{u_n-v_n} + C \\Delta t^{k-1} \\norm{r^{(0)} - s^{(0)}}.\n \\end{align*}\n The value of $\\norm{r^{(0)} - s^{(0)}}$ depends on the initial guess for the SDC iterations. If, for example, the value at the last time step is used as the initial guess for all collocation nodes, i.e.\\ $U^{(0)} = U_n$ and $V^{(0)} = V_n$, we get $\\norm{r^{(0)} - s^{(0)}} = 0 - 0 = 0$. If, by contrast, the initial guess is chosen to be zero, it follows $\\norm{r^{(0)} - s^{(0)}} = \\abs{u_n - v_n}$. Both variants, however, guarantee that $\\abs{\\phi^k(u_n) - \\phi^k(v_n)} \\leq C \\abs{u_n - v_n}$ which was to be shown. \n\\end{proof}\n\n\n\\begin{remark}\n Note that the assumed upper bound for the step size $\\Delta t$ in the previous theorem is the same as the one in \\fref{cor:conv_last}, describing the consistency of SDC. Hence, there is no additional restriction for the convergence of the method.\n\\end{remark}\n\nTogether with the last theorem, \\fref{cor:conv_last} can be extended towards a convergence theorem for SDC regarded in the context of single-step methods to solve ODEs. Specifically, the proven stability of the method allows a direct transfer of the order of consistency to the order of convergence. Consequently, it follows that SDC, in the sense of a single-step method, converges with order $\\min(k_0+k-1, 2M-1)$.\n\nIn the next chapter, MLSDC, a multi-level extension of SDC, is described. It is motivated by the assumption that additional iterations on a coarser level may increase the order of accuracy while keeping the costs rather low. In the next chapter we will investigate whether this assumption holds true, i.e. if the convergence order is indeed increased by the additional execution of relatively low-cost iterations on the coarse level.", "meta": {"timestamp": "2020-08-17T02:00:52", "yymm": "2002", "arxiv_id": "2002.07555", "language": "en", "url": "https://arxiv.org/abs/2002.07555", "abstract": "The spectral deferred correction (SDC) method is class of iterative solvers for ordinary differential equations (ODEs). It can be interpreted as a preconditioned Picard iteration for the collocation problem. The convergence of this method is well-known, for suitable problems it gains one order per iteration up to the order of the quadrature method of the collocation problem provided. This appealing feature enables an easy creation of flexible, high-order accurate methods for ODEs. A variation of SDC are multi-level spectral deferred corrections (MLSDC). Here, iterations are performed on a hierarchy of levels and an FAS correction term, as in nonlinear multigrid methods, couples solutions on different levels. While there are several numerical examples which show its capabilities and efficiency, a theoretical convergence proof is still missing. This paper addresses this issue. A proof of the convergence of MLSDC, including the determination of the convergence rate in the time-step size, will be given and the results of the theoretical analysis will be numerically demonstrated. It turns out that there are restrictions for the advantages of this method over SDC regarding the convergence rate.", "subjects": "Numerical Analysis (math.NA)", "title": "Convergence analysis of multi-level spectral deferred corrections", "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.9802808747730899, "lm_q2_score": 0.8705972566572503, "lm_q1q2_score": 0.8534298403310215}} {"url": "https://arxiv.org/abs/1805.00343", "title": "Teaching Differentiation: A Rare Case for the Problem of the Slope of the Tangent Line", "abstract": "In this article we discuss an important students' misconception about derivatives, that the expression of the derivative of the function contains the information as to whether the function is differentiable or not where the expression is undefined. As a working example we consider a typical Calculus problem of finding the horizontal tangent lines of a function. Following the standard procedure, we derive the expression for the derivative using Product Rule. The search for the values of the independent variable, that make the derivative equal zero, leads to missing the unique solution of the problem. We show that in this case, even though the expression of the derivative is undefined, the function indeed possesses the derivative at the point. We also provide the methodological treatment of such functions, which can be effectively used in the classroom.", "text": "\\section{Introduction}\n\nDerivate is one of the most important topics not only in mathematics, but also in physics, chemistry, economics and engineering. Every standard Calculus course provides a variety of exercises for the students to learn how to apply the concept of derivative. The types of problems range from finding an equation of the tangent line to the application of differentials and advanced curve sketching. Usually, these exercises heavily rely on such differentiation techniques as Product, Quotient and Chain Rules, Implicit and Logarithmic Differentiation \\cite{Stewart2012}. The definition of the derivative is hardly ever applied after the first few classes and its use is not much motivated.\n\nLike many other topics in undergraduate mathematics, derivative gave rise to many misconceptions \\cite{Muzangwa2012}, \\cite{Gur2007}, \\cite{Li2006}. Just when the students seem to learn how to use the differentiation rules for most essential functions, the application of the derivative brings new issues. A common students' error of determining the domain of the derivative from its formula is discussed in \\cite{Rivera2013} and some interesting examples of the derivatives, defined at the points where the functions themselves are undefined, are provided. However, the hunt for misconceptions takes another twist for the derivatives undefined at the points where the functions are in fact defined.\n\nThe expression of the derivative of the function obtained using differentiation techniques does not necessarily contain the information about the existence or the value of the derivative at the points, where the expression for the derivative is undefined. In this article we discuss a type of continuous functions that have the expression for the derivative undefined at a certain point, while the derivative itself at that point exists. We show, how relying on the formula for the derivative for finding the horizontal tangent line of a function, leads to a false conclusion and consequently to missing a solution. We also provide a simple methodological treatment of similar functions suitable for the classroom.\n\n\\section{Calculating the Derivative}\n\nIn order to illustrate how deceitful the expression of the derivative can be to a students' eye, let us consider the following problem.\n\n\\vspace{12pt}\n\n\\fbox{\\begin{minipage}{5.25in}\n\n\\begin{center}\n\n\\begin{minipage}{5.0in}\n\n\\vspace{10pt}\n\n\\emph{Problem}\n\n\\vspace{10pt}\n\nDifferentiate the function $f\\left(x\\right)=\\sqrt[3]{x}\\sin{\\left(x^2\\right)}$. For which values of $x$ from the interval $\\left[-1,1\\right]$ does the graph of $f\\left(x\\right)$ have a horizontal tangent?\n\n\\vspace{10pt}\n\n\\end{minipage}\n\n\\end{center}\n\n\\end{minipage}}\n\n\\vspace{12pt}\n\nProblems with similar formulations can be found in many Calculus books \\cite{Stewart2012}, \\cite{Larson2010}, \\cite{Thomas2009}. Following the common procedure, let us find the expression for the derivative of the function $f\\left(x\\right)$ applying the Product Rule:\n\\begin{eqnarray}\nf'\\left(x\\right) &=& \\left(\\sqrt[3]{x}\\right)'\\sin{\\left(x^2\\right)}+\\left(\\sin{\\left(x^2\\right)}\\right)'\\sqrt[3]{x} \\notag \\\\ &=& \\frac{1}{3\\sqrt[3]{x^2}}\\sin{\\left(x^2\\right)}+2x\\cos{\\left(x^2\\right)}\\sqrt[3]{x} \\notag \\\\ &=& \\frac{6x^2\\cos{x^2}+\\sin{x^2}}{3\\sqrt[3]{x^2}} \\label{DerivativeExpression}\n\\end{eqnarray}\n\nSimilar to \\cite{Stewart2012}, we find the values of $x$ where the derivative $f'\\left(x\\right)$ is equal to zero:\n\\begin{equation}\n6x^2\\cos{x^2}+\\sin{x^2} = 0 \n\\label{DerivativeEqualZero}\n\\end{equation}\n\nSince the expression for the derivative (\\ref{DerivativeExpression}) is not defined at $x=0$, it is not hard to see that for all values of $x$ from $\\left[-1,1\\right]$ distinct from zero, the left-hand side of (\\ref{DerivativeEqualZero}) is always positive. Hence, we conclude that the function $f\\left(x\\right)$ does not have horizontal tangent lines on the interval $\\left[-1,1\\right]$.\n\nHowever, a closer look at the graph of the function $f\\left(x\\right)$ seems to point at a different result: there is a horizontal tangent at $x=0$ (see Figure \\ref{fig:FunctionGraph}). \n\nFirst, note that the function $f\\left(x\\right)$ is defined in $x=0$. In order to verify if it has a horizontal tangent at this point, let us find the derivative of the function $f\\left(x\\right)$ using definition:\n\\begin{eqnarray}\nf'\\left(0\\right) &=& \\lim_{h\\rightarrow0}{\\frac{f\\left(0+h\\right)-f\\left(0\\right)}{h}} \\notag \\\\ \n&=& \\lim_{h\\rightarrow0}{\\frac{\\sqrt[3]{h}\\sin{\\left(h^2\\right)}}{h}} \\notag \\\\ \n&=& \\lim_{h\\rightarrow0}{\\left(\\sqrt[3]{h} \\cdot {h} \\cdot \\frac{\\sin{\\left(h^2\\right)}}{h^2}\\right)} \\notag \\\\\n&=& \\lim_{h\\rightarrow0}{\\sqrt[3]{h}} \\cdot \\lim_{h\\rightarrow0}{h} \\cdot \\lim_{h\\rightarrow0}{\\frac{\\sin{\\left(h^2\\right)}}{h^2}} \\notag \\\\\n&=& 0 \\cdot 0 \\cdot 1 = 0 \\notag\n\\end{eqnarray}\nsince each of the limits above exists. We see that, indeed, the function $f\\left(x\\right)$ possesses a horizontal tangent line at the point $x=0$.\n\n\\section{Closer Look at the Expression for the Derivative}\n\nWhat is the problem with the standard procedure proposed by many textbooks and repeated in every Calculus class? The explanation lies in the following premise: the expression of the derivative of the function does not contain the information as to whether the function is differentiable or not at the points where it is undefined. As it is pointed out in \\cite{Rivera2013}, the domain of the derivative is determined \\emph{a priori} and therefore should not be obtained from the formula of the derivative itself.\n\nIn the example above the Product Law for derivatives requires the existence of the derivatives of both functions at the point of interest. Since the function $\\sqrt[3]{x}$ is not differentiable in zero, the Product Rule cannot be applied. \n\nIn order to see what exactly happens when we apply the Product Rule, let us find the expression for the derivative using definition of the derivative:\n\\begin{eqnarray}\nf'\\left(x\\right) &=& \\lim_{h\\rightarrow0}{\\frac{f\\left(x+h\\right)-f\\left(x\\right)}{h}} \\notag \\\\ \n&=& \\lim_{h\\rightarrow0}{\\frac{\\sqrt[3]{x+h}\\sin{\\left(x+h\\right)^2}-\\sqrt[3]{x}\\sin{\\left(x^2\\right)}}{h}} \\notag \\\\ \n&=& \\lim_{h\\rightarrow0}{\\frac{\\left(\\sqrt[3]{x+h}-\\sqrt[3]{x}\\right)}{h}\\sin{\\left(x^2\\right)}} + \\notag \\\\\n&& \\lim_{h\\rightarrow0}{\\frac{\\left(\\sin{\\left(x+h\\right)^2}-\\sin{\\left(x^2\\right)}\\right)}{h}\\sqrt[3]{x+h}} \\notag \\\\\n&=& \\lim_{h\\rightarrow0}{\\frac{\\sqrt[3]{x+h}-\\sqrt[3]{x}}{h}} \\cdot \\lim_{h\\rightarrow0}{\\sin{\\left(x^2\\right)}} + \\notag \\\\&& \\lim_{h\\rightarrow0}{\\frac{\\sin{\\left(x+h\\right)^2}-\\sin{\\left(x^2\\right)}}{h}} \\cdot \\lim_{h\\rightarrow0}{\\sqrt[3]{x+h}} \\notag \\\\\n&=& \\frac{1}{3\\sqrt[3]{x^2}} \\cdot \\sin{\\left(x^2\\right)}+2x\\cos{\\left(x^2\\right)} \\cdot \\sqrt[3]{x} \\notag \n\\end{eqnarray}\nwhich seems to be identical to the expression (\\ref{DerivativeExpression}).\n\nStudents are expected to develop a skill of deriving similar results and know how to find the derivative of the function using definition of the derivative only. But how `legal' are the performed operations?\n\n\\begin{figure}[H]\n\\begin{center}\n\t\\includegraphics[width=6.0in]{sin.pdf}\n\t\\vspace{.1in}\n\t\\caption{Graph of the function $g\\left(x\\right)=\\sqrt[3]{x}\\cos{\\left(x^2\\right)}$}\n\t\\label{fig:GFunction}\n\\end{center}\n\\end{figure}\n\nLet us consider each of the following limits: \n\\begin{eqnarray*}\n&& \\lim_{h\\rightarrow0}{\\frac{\\sqrt[3]{x+h}-\\sqrt[3]{x}}{h}} \\notag \\\\\n&& \\lim_{h\\rightarrow0}{\\sin{\\left(x^2\\right)}}\\notag \\\\\n&& \\lim_{h\\rightarrow0}{\\frac{\\sin{\\left(x+h\\right)^2}-\\sin{\\left(x^2\\right)}}{h}}\\notag \\\\\n&& \\lim_{h\\rightarrow0}{\\sqrt[3]{x+h}}.\n\\end{eqnarray*}\nThe last three limits exist for all real values of the variable $x$. However, the first limit does not exist when $x=0$. Indeed\n\\begin{equation*}\n\\lim_{h\\rightarrow0}{\\frac{\\sqrt[3]{0+h}-\\sqrt[3]{0}}{h}} = \\lim_{h\\rightarrow0}{\\frac{1}{\\sqrt[3]{h^2}}} = + \\infty\n\\end{equation*}\n\nThis implies that the Product and Sum Laws for limits cannot be applied and therefore this step is not justifiable in the case of $x=0$. When the derivation is performed, we automatically assume the conditions, under which the Product Law for limits can be applied, i.e. that both limits that are multiplied exist. It is not hard to see that in our case these conditions are actually equivalent to $x\\neq0$. This is precisely why, when we wrote out the expression for the derivative (\\ref{DerivativeExpression}), it already contained the assumption that it is only true for the values of $x$ that are different from zero.\n\nNote, that in the case of $x=0$ the application of the Product and Sum Laws for limits is not necessary, since the term $\\left(\\sqrt[3]{x+h}-\\sqrt[3]{x}\\right)\\sin{\\left(x^2\\right)}$ vanishes.\n\nThe correct expression for the derivative of the function $f\\left(x\\right)$ should be the following:\n\\begin{equation*}\nf'\\left(x\\right) = \n\\begin{cases} \n\\frac{6x^2\\cos{\\left(x^2\\right)}+\\sin{\\left(x^2\\right)}}{3\\sqrt[3]{x^2}}, & \\mbox{if } x \\neq 0 \\\\ \n0, & \\mbox{if } x = 0 \n\\end{cases}\n\\end{equation*}\n\nThe expression for the derivative of the function provides the correct value of the derivative only for those values of the independent variable, for which the expression is defined; it does not tell anything about the existence or the value of the derivative, where the expression for the derivative is undefined. Indeed, let us consider the function\n\\begin{equation*}\ng\\left(x\\right) = {\\sqrt[3]{x}}\\cos{\\left(x^2\\right)}\n\\end{equation*}\nand its derivative $g'\\left(x\\right)$ \n\\begin{equation*}\ng'\\left(x\\right) = \\frac{\\cos{\\left(x^2\\right)}-6x^2\\sin{\\left(x^2\\right)}}{3\\sqrt[3]{x^2}}\n\\end{equation*}\n\nSimilar to the previous example, the expression for the derivative is undefined at $x=0$. Nonetheless, it can be shown that $g\\left(x\\right)$ is not differentiable at $x=0$ (see Figure \\ref{fig:GFunction}). Therefore, we provided two visually similar functions: both have the expressions for their derivatives undefined in zero, however, one of these functions possesses a derivative, but the other one does not.\n\n\\section{Methodological Remarks}\n\nUnfortunately, there exist many functions similar to the ones discussed above and they can arise in a variety of typical Calculus problems: finding the points where the tangent line is horizontal, finding an equation of the tangent and normal lines to the curve at the given point, the use of differentials and graph sketching. Relying only on the expression of the derivative for determining its value at the undefined points may lead to missing a solution (as in the example discussed above) or to some completely false interpretations (as in the case of curve sketching).\n\nAs it was discussed above, the expression for the derivative does not provide any information on the existence or the value of the derivative, where the expression itself is undefined. Here we present a methodology for the analysis of this type of functions.\n\nLet $f\\left(x\\right)$ be the function of interest and $f'\\left(x\\right)$ be the expression of its derivative undefined at some point $x_{0}$. In order to find out if $f\\left(x\\right)$ is differentiable at $x_{0}$, we suggest to follow a list of steps:\n\n\\begin{enumerate}\n \\item Check if the function $f\\left(x\\right)$ itself is defined at the point $x_{0}$. If $f\\left(x\\right)$ is undefined at $x_{0}$, then it is not differentiable at $x_{0}$. If $f\\left(x\\right)$ is defined at $x_{0}$, then proceed to next step. \n \\item Identify the basic functions that are used in the formula of the function $f\\left(x\\right)$, that are themselves defined at the point $x_{0}$, but their derivative is not (such as, for example, the root functions).\n\t\\item Find the derivative of the function $f\\left(x\\right)$ at the point $x_{0}$ using definition.\n\\end{enumerate}\n\nThe importance of the first step comes from the fact that most students tend to pay little attention to the functions domain analysis when asked to investigate its derivative. Formally, the second step can be skipped, however it will give the students the insight into which part of the function presents a problem and teach them to identify similar cases in the future. the difficulty of accomplishing the third step depends on the form of the function and sometimes can be tedious. Nevertheless, it allows the students to apply the previously obtained skills and encourages the review of the material.\n\n\\begin{figure}[H]\n\\begin{center}\n\t\\includegraphics[width=6.0in]{cos.pdf}\n\t\\vspace{.1in}\n\t\\caption{Graph of the function $g\\left(x\\right)=\\sqrt[3]{x}\\cos{\\left(x^2\\right)}$}\n\t\\label{fig:GFunction}\n\\end{center}\n\\end{figure}\n\n\\section{Conclusion}\n\nWe discussed the misconception, that the expression of the derivative of the function contains the information as to whether the function is differentiable or not at the points, where the expression is undefined. We considered a typical Calculus problem of looking for the horizontal tangent line of a function as an example. We showed how the search for the values that make the expression of the derivative equal zero leads to missing a solution: even though the expression of the derivative is undefined, the function still possesses the derivative at the point. We provided an example of the function that similarly has the expression for the derivative undefined, however the function is not differentiable at the point. We also presented the methodological treatment of such functions by applying the definition of the derivative, which can be used in the classroom.\n\n", "meta": {"timestamp": "2018-05-02T02:09:26", "yymm": "1805", "arxiv_id": "1805.00343", "language": "en", "url": "https://arxiv.org/abs/1805.00343", "abstract": "In this article we discuss an important students' misconception about derivatives, that the expression of the derivative of the function contains the information as to whether the function is differentiable or not where the expression is undefined. As a working example we consider a typical Calculus problem of finding the horizontal tangent lines of a function. Following the standard procedure, we derive the expression for the derivative using Product Rule. The search for the values of the independent variable, that make the derivative equal zero, leads to missing the unique solution of the problem. We show that in this case, even though the expression of the derivative is undefined, the function indeed possesses the derivative at the point. We also provide the methodological treatment of such functions, which can be effectively used in the classroom.", "subjects": "History and Overview (math.HO)", "title": "Teaching Differentiation: A Rare Case for the Problem of the Slope of the Tangent Line", "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.9898303416461329, "lm_q2_score": 0.8615382147637196, "lm_q1q2_score": 0.852776665460772}} {"url": "https://arxiv.org/abs/1809.04072", "title": "The 21 Card Trick and its Generalization", "abstract": "The 21 card trick is well known. It was recently shown in an episode of the popular YouTube channel Numberphile. In that trick, the audience is asked to remember a card, and through a series of steps, the magician is able to find the card. In this article, we look into the mathematics behind the trick, and look at a complete generalization of the trick. We show that this trick can be performed with any number of cards.", "text": "\\section{Introduction to the 21 card trick} \nThe \\textbf{21 card trick} (21CT) is a very popular card trick. It was also recently shown in an episode of the popular YouTube channel \\href{https://www.youtube.com/watch?v=d7dg7gVDWyg}{Numberphile}. We first explain how this trick is performed in a series of steps. For the purpose of this demonstration, we will assume that a magician (henceforth, Magi), is showing this trick to his friend and audience (henceforth, Audy).\n\\begin{enumerate}\n\\item Audy randomly chooses 21 cards from a deck of cards. He remembers one card from that set, shuffles the set of 21 cards and hands it back to Magi.\n\\item Magi requests Audy to pay attention as he puts the cards face up one at a time adjacent to each other creating 3 stacks of 7 cards each in the process. Magi asks Audy to tell him which stack contains his card.\n\\item Magi then puts the stack that contained Audy's card between the two other stacks, and then repeats Step (2) two more times. \n\\item Magi now puts the cards in the table one at a time, and stops at the 11th card which turns out to be Audy's card.\n\\end{enumerate} \nNote that this trick can be made more ``magical\" with some extra activities, which we will not discuss as they don't contribute anything to the main problem. Our current goal is to find out why the 11th card on the deck happened to be Audy's card. We will explore the mathematics behind this after we look at a detailed example of the trick. \n\nOnce Audy hands the shuffled deck of cards to Magi, we know that Audy's card could be any one of the 21 cards in that deck. Let's number the cards from top to bottom of the deck by $1, 2, 3, \\ldots, 21$. We shall use the term \\textbf{\\emph{deck id}} to denote the position of Audy's card in the deck. Similarly, we will use the term \\textbf{\\emph{stack id}} to denote the position of Audy's card in a stack from the top.\n\\begin{framed}\n\\begin{definition}\nAn \\textbf{iteration} is considered to be the process of splitting the cards of a deck into stacks.\n\\end{definition}\n\\begin{definition}\nThe \\textbf{deck id} of Audy's card after $k(\\geq 1)$ iterations of splitting into stacks, denoted by $d_k$, is the position of his card from the top, in a deck of $n$ distinct cards. \n\\end{definition}\n\\begin{definition}\nThe \\textbf{stack id} of Audy's card after $k( \\geq 1)$ iterations of splitting into stacks, denoted by $s_k$, is the position of his card from the top, in its individual stack.\n\\end{definition}\n\\end{framed}\n\n\\noindent\nNote that the initial deck id of Audy's card before any iterations happen is denoted by $d_0$. We now show a detailed example of the 21CT in action.\n\n\\begin{enumerate}\n\\item Audy selects a random collection of 21 cards. The cards are:\n\n\n\\smallskip\n\\noindent\n\\psset{inline=boxed}\n\\fourc, \\tenc, \\tred, \\tenh, \\tend, \\eigd, \\sixc, \\tres, \\tens, \\twoc, \\fives, \\sixh, \\twod, \\eigc, \\Ac, \\Kd, \\ninec, \\fours, \\nineh, \\Kh, \\Jd\n\n\\smallskip\n\\noindent\nHe remembers a card from this set. Suppose it's \\fours. He then shuffles this deck again and hands it over to Magi. The shuffled deck looks like this:\n\n\\smallskip\n\\noindent\n\\eigd, \\tend, \\nineh, \\twoc, \\tred, \\fives, \\fourc, \\sixc, \\sixh, \\twod, \\tenc, \\tenh, \\Kd, \\Jd, \\ninec, \\eigc, \\tres, \\Ac, \\tens, \\fours, \\Kh \\, \\, ($d_0 = 20$)\n\n\\enspace\n\\noindent\nNote that an iteration hasn't happened yet. Thus, $s_0$ is not defined, and $d_0 = 20$ (as \\fours \\, is the 20th card in the deck at the moment).\n\n\\smallskip\n\\item Now Magi does the first iteration by putting these cards face up one at a time adjacent to each other creating 3 stacks. The stacks then look like this:\n\n\\smallskip\n\\noindent\n\\begin{table}[H]\n\\centering\n\\begin{tabular}{|c|c|c|}\n\\hline\nStack 1 & Stack 2 & Stack 3 \\\\\n\\hline \n\\eigd & \\tend & \\nineh \\\\ \n\\hline \n\\twoc & \\tred & \\fives \\\\ \n\\hline \n\\fourc & \\sixc & \\sixh \\\\ \n\\hline \n\\twod & \\tenc & \\tenh \\\\ \n\\hline \n\\Kd & \\Jd & \\ninec \\\\ \n\\hline \n\\eigc & \\tres & \\Ac \\\\ \n\\hline \n\\tens & \\fours & \\Kh \\\\ \n\\hline \n\\end{tabular} \n\\caption{Stacks after Iteration 1 ($s_1 = 7$ as \\protect\\psset{inline=boxed} \\protect\\fours \\, is the 7th card in the second stack)}\n\\end{table}\n\n\\smallskip\n\\noindent\nHe then asks Audy which stack contains his card. Audy replies by saying that his card is in Stack 2 (\\fours \\, is in Stack 2). \n\n\\item Magi then puts Stack 2 in between Stack 1 and Stack 3. The set of 21 cards now look as follows from top to bottom: \n\n\\smallskip\n\\noindent\n\\eigd, \\twoc, \\fourc, \\twod, \\Kd, \\eigc, \\tens, \\tend, \\tred, \\sixc, \\tenc, \\Jd, \\tres, \\fours, \\nineh, \\fives, \\sixh, \\tenh, \\ninec, \\Ac, \\Kh \\, \\, ($d_1 = 14$)\n\n\\smallskip\n\\noindent \nMagi now repeats Step (2) two more times. The stacks after iteration 2 looks like this:\n\n\\smallskip\n\\noindent\n\\begin{table}[H]\n\\centering\n\\begin{tabular}{|c|c|c|}\n\\hline\nStack 1 & Stack 2 & Stack 3 \\\\\n\\hline \n\\eigd & \\twoc & \\fourc \\\\ \n\\hline \n\\twod & \\Kd & \\eigc \\\\ \n\\hline \n\\tens & \\tend & \\tred \\\\ \n\\hline \n\\sixc & \\tenc & \\Jd \\\\ \n\\hline \n\\tres & \\fours & \\nineh \\\\ \n\\hline \n\\fives & \\sixh & \\tenh \\\\ \n\\hline \n\\ninec & \\Ac & \\Kh \\\\ \n\\hline \n\\end{tabular} \n\\caption{Stacks after Iteration 2 ($s_2 = 5$)}\n\\end{table}\n\n\\noindent\n\\psset{inline=boxed}\nHe then again asks Audy which stack contains his card. Audy replies by saying that his card is in Stack 2 again (\\fours \\, is in Stack 2). Magi again puts Stack 2 in between Stack 1 and Stack 3. The set of 21 cards now look as follows from top to bottom:\n\n\\smallskip\n\\noindent\n\\eigd, \\twod, \\tens, \\sixc, \\tres, \\fives, \\ninec, \\twoc, \\Kd, \\tend, \\tenc, \\fours, \\sixh, \\Ac, \\fourc, \\eigc, \\tred, \\Jd, \\nineh, \\tenh, \\Kh \\, \\, ($d_2 = 12$)\n\n\\smallskip\n\\noindent\nMagi again repeats the process of splitting the cards into 3 stacks for the third and final time. The stacks after iteration 3 look like this:\n\n\\smallskip\n\\noindent\n\\begin{table}[H]\n\\centering\n\\begin{tabular}{|c|c|c|}\n\\hline\nStack 1 & Stack 2 & Stack 3 \\\\\n\\hline \n\\eigd & \\twod & \\tens \\\\ \n\\hline \n\\sixc & \\tres & \\fives \\\\ \n\\hline \n\\ninec & \\twoc & \\Kd \\\\ \n\\hline \n\\tend & \\tenc & \\fours \\\\ \n\\hline \n\\sixh & \\Ac & \\fourc \\\\ \n\\hline \n\\eigc & \\tred & \\Jd \\\\ \n\\hline \n\\nineh & \\tenh & \\Kh \\\\ \n\\hline \n\\end{tabular} \n\\caption{Stacks after Iteration 3 ($s_3 = 4$)}\n\\end{table}\n\n\\smallskip\n\\noindent\nNow for the final time, Magi asks Audy about the stack that contains his card. Audy replies by saying that his card is in Stack 3 (\\fours \\, is in Stack 3). Magi then puts Stack 3 in between Stack 1 and Stack 2. The set of 21 cards now look as follows from top to bottom:\n\n\\smallskip\n\\noindent\n\\eigd, \\sixc, \\ninec, \\tend, \\sixh, \\eigc, \\nineh, \\tens, \\fives, \\Kd, \\fours, \\fourc, \\Jd, \\Kh, \\twod, \\tres, \\twoc, \\tenc, \\Ac, \\tred, \\tenh \\, \\, ($d_3 = 11$)\n\n\\item Magi now puts each card from the top face down on the table one by one, and flips over the \\textbf{11th card} which turns out to be Audy's card, \\fours.\n\n\\smallskip\n\\noindent\n\\begin{center}\n\\eigd, \\sixc, \\ninec, \\tend, \\sixh, \\eigc, \\nineh, \\tens, \\fives, \\Kd, \n\\fcolorbox{blue}{green}{\\fours}\n\\end{center}\n\n\\end{enumerate}\n\n\n\\smallskip\n\\noindent\nMagi correctly finds out Audy's card, and it leaves Audy startled. He wonders what kind of voodoo did Magi apply in all of this. Little does he know about the power of mathematics behind this trick.\n\n\n\\section{The Mathematics behind the 21 card trick}\nMost card tricks rely on mathematics. The 21CT is no exception. We now look at the mathematics behind this trick. \n\n\\smallskip\n\\noindent\nBefore we look into the steps of the 21CT again, we would like to look at the definition of the \\textbf{ceiling} and \\textbf{floor} functions, and three lemmas that will be used extensively in this article. \n\n\\bigskip\n\\noindent\n\n\\begin{framed}\n\\begin{definition}\nSuppose $x \\in \\mathbb{R}$. The \\textbf{ceiling} of $x$ denoted by $\\ceil{x}$ is the smallest integer greater than or equal to $x$. In general, if $\\ceil{x} = n \\in \\mathbb{Z}$, then $n-1 < x \\leq n$.\n\\end{definition}\n\\begin{definition}\nSuppose $x \\in \\mathbb{R}$. The \\textbf{floor} of $x$ denoted by $\\floor{x}$ is the largest integer less than or equal to $x$. In general, if $\\floor{x} = n \\in \\mathbb{Z}$, then $n \\leq x < n+1$. In this case $x - n = x - \\floor{x}$ is called the \\emph{fractional part} of $x$, and is denoted by $\\{x\\}$.\n\\end{definition}\n\\end{framed}\n\n\\begin{lemma}\\label{lem1}\nFor $x, y \\in \\mathbb{R}$, if $x \\leq y$, then\n$$\\ceil{x} \\leq \\ceil{y}$$\n\\end{lemma}\n\n\\begin{proof}\nWe divide this into two cases.\n\n\\enspace\n\n\\noindent\n\\underline{Case 1 :} ($y \\leq \\ceil{x}$) As $\\ceil{x} - 1 < x \\leq \\ceil{x}$ by definition, and $x \\leq y$, therefore, $\\ceil{x} - 1 < y \\leq \\ceil{x}$. This implies that $\\ceil{y} = \\ceil{x}$, and therefore, $\\ceil{x} \\leq \\ceil{y}$.\n\n\\enspace\n\n\\noindent\n\\underline{Case 2 :} ($y > \\ceil{x}$). By definition, $y \\leq \\ceil{y}$. Therefore, we have $\\ceil{x} < y \\leq \\ceil{y}$. Hence, $\\ceil{x} < \\ceil{y}$.\n\n\\enspace\n\n\\noindent\nCombining the two above case, we have the desired inequality.\n\\end{proof}\n\n\n\\begin{lemma}\\label{lem2}\nFor $n \\in \\mathbb{Z}$ and $x \\in \\mathbb{R}$, \n$$\\ceil{n+x} = n + \\ceil{x}$$\n\\end{lemma}\n\n\\begin{proof}\nSuppose $\\ceil{n+x} = d$. Thus,\n\\begin{equation*}\n\\begin{split}\nd-1 < n + & x \\leq d \\\\\n(d-n)-1 < \\, & x \\leq d-n\n\\end{split}\n\\end{equation*}\nThis implies, $\\ceil{x} = d - n$. Adding $n$ to both sides, we have $n + \\ceil{x} = d$ which shows the desired result.\n\\end{proof}\n\n\\begin{lemma}\\label{lem3}\nFor $n, m \\in \\mathbb{Z}$ with $m > 0$, and $x \\in \\mathbb{R}$, \n$$\\ceil[\\Bigg]{\\frac{n + \\ceil{x}}{m}} = \\ceil[\\Bigg]{\\frac{n+x}{m}}$$ \n\\end{lemma}\n\n\\begin{proof}\nSuppose $\\ceil[\\Bigg]{\\dfrac{n+x}{m}} = d$. Thus,\n\\begin{alignat*}{2}\nd-1 &< &\\frac{n+x}{m} &\\leq d \\\\\nmd - m &< &n + x &\\leq md \\\\\nmd - m - n &< &x \\, \\, \\, \\, \\, &\\leq md - n \n\\end{alignat*}\nAs $m, d, n \\in \\mathbb{Z}$, thus $md-m-n$ and $md-n$ are integers. By definition of the ceiling function, $x \\leq \\ceil{x}$. Hence, $md - m - n < \\ceil{x}$. On the other hand, $x, md-n \\in \\mathbb{R}$. Hence, by Lemma \\ref{lem1}, $\\ceil{x} \\leq \\ceil{md-n} = md-n$ as $md - n \\in \\mathbb{Z}$. Thus, we have \n\n\\begin{alignat*}{2}\nmd-m-n &< &\\ceil{x} \\, \\, \\, \\, \\, \\, \\, &\\leq md-n \\\\\nmd - m &< &n + \\ceil{x} &\\leq md \\\\\n\\frac{md - m}{m} &< &\\frac{n + \\ceil{x}}{m} &\\leq \\frac{md}{m} \\\\\nd-1 &< &\\frac{n + \\ceil{x}}{m} &\\leq d\n\\end{alignat*}\nThis implies, $\\ceil[\\Bigg]{\\dfrac{n + \\ceil{x}}{m}} = d$, which proves the result.\n\n\\end{proof}\n\n\n\\noindent\nNow we look back at the steps of the card trick again. \n\n\\enspace\n\n\\begin{enumerate}[Step 1:]\n\\item Audy hands the shuffled deck to Magi. We denote the initial deck id of Audy's card by $d_0$ as no iterations of splitting into stacks has been done yet. Therefore, \n$$1 \\leq d_0 \\leq 21$$\n\\item Magi now puts the cards from the top, face up, on the table adjacent to each other, creating 3 stacks of 7 cards each. Audy's card is in one of the three stacks. We claim the following.\n\\begin{claim}\\label{claim1}\nIf $s_k$ denotes the stack id, and $d_k$ denote the deck id of Audy's card after $k$ iterations, then,\n$$s_{k} = \\ceil[\\Big]{\\dfrac{d_{k-1}}{3}} \\hspace{0.2in} \\text{for} \\hspace{0.1in} k \\geq 1$$\n\\end{claim}\n\\begin{proof}\nIt is imperative for us to note that a stack id for Audy's card is created after every iteration. Hence, $s_k$ is defined only when $k \\geq 1$. On the other hand, a deck id for Audy's card is created when the 21 cards are all in a single deck, which happens for the first time when Audy hands over the shuffled deck to Magi. Thus, $d_k$ is defined for $k \\geq 0$. Also, note that the $k$th iteration which creates $s_{k}$ is performed using the 21 card deck after the $(k-1)$th iteration which creates $d_{k-1}$. Hence the relationship between $s_{k}$ and $d_{k-1}$. It is clear that the stack id of a card is the ``row number\" of the card in the stack. Thus for $k \\geq 1$, the first row consists of cards with $d_{k-1} = 1, 2, 3$. The second row consists of cards with $d_{k-1} = 4, 5, 6$ and so on. This means that the $n$th row consists of cards with $d_{k-1} = 3n-2, 3n-1, 3n$. As $n =s_{k}$. Thus,\n\\begin{alignat*}{2}\n3s_{k}-3 &< &d_{k-1} &\\leq 3s_{k} \\\\\ns_{k} -1 &< &\\frac{d_{k-1}}{3} &\\leq s_{k}\n\\end{alignat*}\nTherefore, $\\ceil[\\Big]{\\dfrac{d_{k-1}}{3}} = s_{k}$.\n\\end{proof}\n \n\\noindent\nAs $1 \\leq d_0 \\leq 21$, hence, \n\\begin{alignat*}{2}\n\\dfrac{1}{3} &\\leq &\\dfrac{d_0}{3} \\, &\\leq 7 \\\\\n1 &\\leq &\\ceil[\\Big]{\\dfrac{d_0}{3}} &\\leq 7 \\qquad (\\text{By Lemma } \\ref{lem1}) \\\\\n1 &\\leq &s_1 \\, \\, &\\leq 7 \\qquad (\\text{By Claim } \\ref{claim1})\n\\end{alignat*}\n \n\\noindent\nNow Magi asks Audy to tell him the stack that contains his card. Audy responds by saying that it's Stack 2.\n \n\\item Magi then puts Stack 2 in between the other two stacks and creates a full deck of 21 cards. Note that there is a stack of 7 cards on top of Stack 2 at the moment. The position of Audy's card in Stack 2 is currently $s_1$. Hence, the new deck id of Audy's card after the first iteration is, \n$$d_1 = 7+ s_1$$\nSince $1 \\leq s_1 \\leq 7$, therefore,\n\\begin{alignat*}{2}\n8 &\\leq &7+s_1 &\\leq 14 \\\\\n8 &\\leq &d_1 \\, \\, \\, &\\leq 14 \n\\end{alignat*}\n\n\\noindent\nNow the second iteration is performed, and the stack id $s_2$ is created,\n\\begin{alignat*}{2}\n8 &\\leq &d_1 &\\leq 14 \\\\\n\\dfrac{8}{3} &\\leq &\\dfrac{d_1}{3} &\\leq \\dfrac{14}{3} \\\\\n3 &\\leq &\\ceil[\\Big]{\\dfrac{d_1}{3}} &\\leq 5 \\qquad (\\text{By Lemma } \\ref{lem1}) \\\\\n3 &\\leq &s_2 \\, \\, &\\leq 5 \\qquad (\\text{By Claim } \\ref{claim1})\n\\end{alignat*}\n\n\\noindent\nNote that Magi now knows that the stack id of Audy's card is either 3, 4 or 5, but that is not good enough, as he needs to find the exact card. He again asks Audy to tell him the stack where his card belongs. Audy mentions that it's Stack 2, and Magi puts Stack 2 in between the other two stacks to create a deck of 21 cards. The position of Audy's card in Stack 2 is currently $s_2$. Therefore, the new deck id of Audy's card after the second iteration is,\n$$d_2 = 7 + s_2$$\nNow, $3 \\leq s_2 \\leq 5$, therefore,\n\\begin{alignat*}{2}\n10 &\\leq &7+s_2 &\\leq 12 \\\\\n10 &\\leq &d_2 \\, \\, \\, &\\leq 12 \n\\end{alignat*}\n\n\\noindent\nNow the third iteration is performed, and the stack id $s_3$ is created,\n\\begin{alignat*}{2}\n10 &\\leq &d_2 \\, \\, &\\leq 12 \\\\\n\\dfrac{10}{3} &\\leq &\\dfrac{d_2}{3} \\, \\, &\\leq \\dfrac{12}{3} \\\\\n4 &\\leq &\\ceil[\\Big]{\\dfrac{d_2}{3}} &\\leq 4 \\qquad (\\text{By Lemma } \\ref{lem1}) \\\\\n4 &\\leq &s_3 \\, \\, &\\leq 4 \\qquad (\\text{By Claim } \\ref{claim1})\n\\end{alignat*}\n\n\\noindent\nThis means $s_3 = 4$. Magi can now see the finish line. Magi has now found the exact stack id of Audy's card in a stack, but he does not know exactly which stack contains Audy's card. He thus asks Audy one final time about the stack that contains his card. Audy mentions Stack 3, and Magi puts Stack 3 in between the other two stacks. The position of Audy's card in the middle stack is currently $s_3 = 4$. Hence, the new deck id of Audy's card after the third iteration is,\n$$d_3 = 7 + s_3 = 11$$\n\n\\item Magi now knows that Audy's card is the 11th card from the top of the 21 card deck. He puts each of the card from the top on the table, face down, until he flips over the 11th card to Audy's delight.\n\n\\end{enumerate}\n\n\\section{Generalization of the 21 card trick}\\label{sec3}\nNow that we have seen how the 21CT works, and the mathematics behind it, we ask ourselves whether the same trick can be performed by Magi using a random set of, let's say $C$ number of cards. He also wants to find out the number of stacks he needs to split his cards into, where should he put the stack that contains Audy's card, how many iterations should he perform, and lastly, the deck id of Audy's card after the final iteration.\n\n\\begin{comment}\n\\begin{enumerate}\n\\item Can this trick be done with any number of cards?\n\\item Can arbitrary number of stacks be used during an iteration?\n\\item Does the stack that contains the desired card need to always be in the middle?\n\\end{enumerate}\nWe'll answer these questions in this section. Let's say Magi is given a shuffled deck of $C$ distinct cards by his friend Audy again after remembering a card. Magi wonders whether he can perform a similar trick to the 21CT using $C$ cards. \n\\end{comment}\n\n\\noindent\nLet's help out Magi perform his trick. Before we proceed, we introduce some notations.\n\\begin{itemize}\n\\item Number of given cards = $C(>0) \\in \\mathbb{Z}$.\n\\item Number of stacks to split into = $n(>0 \\text{ and} \\leq C) \\in \\mathbb{Z}$.\n\\item Number of stacks to put on top of the stack which contains Audy's card = $j(\\geq 0 \\text{ and} 0) \\in \\mathbb{Z}$.\n\\item Deck id of Audy's card after the final iteration = $l(>0) \\in \\mathbb{Z}$.\n\\end{itemize}\nWe have thus created a 4-tuple ($C, n, j, k$) that provides information about the magic trick. \n\\begin{framed}\n\\begin{definition}\nA magic trick ($C, n, j, k$) is \\textbf{solvable}, if there exists integers $k$ and $l$ such that the magic trick can be performed with the given parameters in $k$ iterations. In this case, we write $$(C, n, j, k) = l$$\nIf such a $k$ or $l$ does not exist then the magic trick cannot be performed, and we say that ($C, n, j, k$) is \\textbf{not solvable}.\n\\end{definition}\n\\end{framed}\n\n\\noindent\nThe 21 card trick is represented by (21, 3, 1, 3) and is solvable and $(21, 3, 1, 3) = 11$. Following along the lines of the 21CT, our main assumption here is that each stack contains the same number of cards. Thus, $n \\mid C$. Thus, $n \\leq C$.\n\n\\noindent\nBefore we proceed any further, we first look at a generalization of Claim \\ref{claim1} in the form of a theorem and an additional result.\n\n\\begin{theorem}\\label{thm4}\nSuppose $C$ cards are split into $n$ stacks during an iteration. If $s_k$ denotes the stack id, and $d_k$ denote the deck id of Audy's card after $k$ iterations, then,\n$$s_{k} = \\ceil[\\Big]{\\dfrac{d_{k-1}}{n}} \\hspace{0.2in} \\text{for} \\hspace{0.1in} k \\geq 1$$\n\\end{theorem}\n\n\\begin{proof}\nThe proof of this theorem is very similar to the earlier proof of Claim \\ref{claim1}. We know that the stack id of a card is the ``row number\" of the card in the stack. Thus for $k \\geq 1$, the first row consists of cards with $d_{k-1} = 1, 2, 3, \\ldots, n$. The second row consists of cards with $d_{k-1} = n+1, n+2, \\ldots, 2n$, and so on. This means that the $q$th row consists of cards with $d_{k-1} = (q-1)n + 1, (q-1)n + 2, \\ldots, (q-1)n + n$. As $q = s_{k}$. Hence, \n\\begin{alignat*}{2}\nn(s_{k}-1) &< &d_{k-1} &\\leq n(s_{k} - 1) + n \\\\\nn(s_{k}-1) &< &d_{k-1} &\\leq ns_{k} \\\\\ns_{k}-1 &< &\\frac{d_{k-1}}{n} &\\leq s_{k}\n\\end{alignat*}\nTherefore, $\\ceil[\\Big]{\\dfrac{d_{k-1}}{n}} = s_{k}$.\n\\end{proof}\n\n\n\\begin{theorem}\\label{thm5}\nSuppose $k \\geq 1$, then\n$$d_k = \\Big(\\frac{C}{n}\\Big)j + s_k$$\n\\end{theorem}\n\n\\begin{proof}\nAs $C$ cards are being split into $n$ stacks, hence each stack contains $\\dfrac{C}{n}$ cards. $s_k$ is the stack id of the desired card after $k$ iterations. $j$ stacks are kept on top of the stack that contains the desired card. There are a total of $\\Big(\\dfrac{C}{n}\\Big)j$ cards in these $j$ stacks. Thus, the new deck id after $k$ iteration is,\n$$d_k = \\Big(\\frac{C}{n}\\Big)j +s_k \\hspace{0.2in} \\text{for} \\hspace{0.1in} k \\geq 1$$\n\\end{proof}\n\n\\noindent\nNote that our goal is to find an exact value for $d_k$. This would tell us the deck id of the desired card which would successfully conclude the trick. Due to Theorem \\ref{thm5}, we now have an expression for $d_k$ that we can investigate further. This results in the next theorem which will be used extensively in the next section.\n\n\\begin{theorem}\\label{thm6}\nSuppose $\\dfrac{C}{n} = m$ and $k \\geq 1$. Then,\n$$d_k = \n\\begin{cases}\nmj + \\ceil[\\Bigg]{\\dfrac{mjn\\Big(\\dfrac{n^{k-1}-1}{n-1}\\Big) + d_0}{n^k}}, & \\text{for} \\hspace{0.1in} n > 1 \\\\\nd_0, & \\text{for} \\hspace{0.1in} n = 1\n\\end{cases}\n$$\n\\end{theorem}\n\n\\begin{proof}\n\\underline{Case 1 :} ($n>1$)\nWe will use mathematical induction on $k$ to prove this result. \n\n\\noindent\nFor $k=1$, L.H.S. = $d_1 = mj + s_1$ from Theorem \\ref{thm5}. Thus, $d_1 = mj + \\ceil[\\Big]{\\dfrac{d_0}{n}}$.\n\n\\noindent\nR.H.S. = $mj + \\ceil[\\Bigg]{\\dfrac{mjn\\Big(\\dfrac{n^{0}-1}{n-1}\\Big) + d_0}{n}} = mj + \\ceil[\\Big]{\\dfrac{d_0}{n}}$. Hence, the result is true for $k=1$. Now assume the result is true for some $k = t \\geq 1$. Thus,\n$$d_t = mj + \\ceil[\\Bigg]{\\frac{mjn\\Big(\\dfrac{n^{t-1}-1}{n-1}\\Big) + d_0}{n^t}}$$\nNow,\n$$d_{t+1} = mj + s_{t+1} = mj + \\ceil[\\Big]{\\dfrac{d_t}{n}} = mj + \\ceil[\\Bigg]{\\dfrac{mj + \\ceil[\\Big]{\\frac{mjn\\Big(\\frac{n^{t-1}-1}{n-1}\\Big) + d_0}{n^t}}}{n}}$$\nAs $mj$ and $n$ are integers, hence by Lemma \\ref{lem3}, \n$$\\ceil[\\Bigg]{\\dfrac{mj + \\ceil[\\Big]{\\frac{mjn\\Big(\\frac{n^{t-1}-1}{n-1}\\Big) + d_0}{n^t}}}{n}} = \\ceil[\\Bigg]{\\dfrac{mj + \\frac{mjn\\Big(\\frac{n^{t-1}-1}{n-1}\\Big) + d_0}{n^t}}{n}} = \\ceil[\\Bigg]{\\dfrac{mjn^t + \\Big(\\frac{mjn^t-mjn}{n-1}\\Big) +d_0}{n^{t+1}}}$$\nNow,\n$$\\ceil[\\Bigg]{\\dfrac{mjn^t + \\Big(\\dfrac{mjn^t-mjn}{n-1}\\Big) +d_0}{n^{t+1}}} = \\ceil[\\Bigg]{\\dfrac{\\Big(\\dfrac{mjn^{t+1}-mjn}{n-1}\\Big) +d_0}{n^{t+1}}} = \\ceil[\\Bigg]{\\dfrac{mjn\\Big(\\dfrac{n^{t}-1}{n-1}\\Big) +d_0}{n^{t+1}}}$$\nTherefore,\n$$d_{t+1} = mj + \\ceil[\\Bigg]{\\dfrac{mjn\\Big(\\dfrac{n^{(t+1)-1}-1}{n-1}\\Big) +d_0}{n^{(t+1)}}}$$\nand thus the result is true for $k = t+1$. Therefore, by induction, the result is true for all $k \\geq 1$.\n\n\\noindent\n\\underline{Case 2 :} ($n=1$) As $n=1$, and $0 \\leq j < n$, hence $j = 0$. By Theorem \\ref{thm5}, \n$$d_k = s_k \\qquad for \\hspace{0.1in} k \\geq 1$$\nNow, according to Theorem \\ref{thm4}, $s_k = \\ceil{d_{k-1}} = d_{k-1}$ for $k \\geq 1$. Therefore,\n$$d_k = d_{k-1} \\qquad for \\hspace{0.1in} k \\geq 1$$\nSolving this recurrence relation, we get $d_k = d_0$.\n\\end{proof}\n\n\n\n\n\n\\section{Main Result}\n\\noindent\nThe 21CT and the results that we have seen so far pushes us to the more general question. \n\\begin{framed}\n\\noindent\n\\begin{center}\\label{gen-ques}\n\\underline{\\textbf{General Question}}\n\\end{center}\nGiven integers $C, n,$ and $j$, does there exist a positive integer $k$, such that the trick $(C, n, j, k)$ is solvable? In that case, what is $(C, n, j, k)$?\n\\end{framed}\n\n\\noindent\nWe answer this question using an algorithm. We will then look at the proof of the results in our algorithm.\n\n\\begin{framed}\\label{algorithm}\n\\begin{enumerate}\n\\item Start with given integers $C, n, j$.\n\\item Find $m = \\dfrac{C}{n}$ and $b = \\dfrac{mj}{n-1}$ (for $n>1$). \n\\item If $C = 1$, then $(1, 1, 0, k)$ is solvable for any $k \\geq 1$. In this case, $(1, 1, 0, k) = 1$. \n\\item If $C>1$ and $n = 1$, then $(C, 1, 0, k)$ is not solvable for any $k \\geq 1$.\n\\item If $C, n > 1$, then $(C, n, 0, k)$ is solvable for any integer $k \\geq \\log_nC$. In this case, $(C, n, 0, k) = 1$.\n\\item If $C, n > 1$, then $(C, n, n-1, k)$ is solvable for any integer $k > \\log_n(C-1)$. In this case, $(C, n, n-1, k) = C$.\n\\item If $C, n > 1$ and $0 \\log_nt$, where $t = \\max\\Big\\{\\dfrac{C-bn}{1-\\{b\\}}, \\dfrac{bn-1}{\\{b\\}}\\Big\\}$ and, $(C, n , j, k) = mj + \\floor{b} + 1$. \n\\end{enumerate}\n\\end{framed}\n\n\\noindent\nThe steps (3) - (7) of the above algorithm gives a complete answer to the general question posed at the beginning of this section. We will prove each step in the form of a theorem, culminating with proof of the all important Step (7).\n\n\\begin{theorem} (Step (3)) \\label{thm:step3}\nIf $C = 1$, then $(1, 1, 0, k)$ is solvable for any $k \\geq 1$ with $(1, 1, 0, k) = 1$.\n\\end{theorem}\n\n\\begin{proof}\nThis is a trivial case as there is only 1 card. As $0 < n \\leq C$, therefore, $n=1$. Also, as $j < n$, hence $j = 0$. By Theorem \\ref{thm6}, $d_k = d_0$. We know that the initial deck id $d_0$ of the desired card is between $1$ and $C$. Thus, $1 \\leq d_0 \\leq 1$. Therefore, $d_k = d_0 = 1$. Thus $(1, 1, 0, k)$ is solvable for any $k \\geq 1$, with $(1, 1, 0, k) = 1$.\n\\end{proof}\n\n\\begin{theorem} (Step (4)) \\label{thm:step4}\nIf $C>1$ and $n = 1$, then $(C, 1, 0, k)$ is not solvable for any $k \\geq 1$.\n\\end{theorem}\n\n\\begin{proof}\nBy Theorem \\ref{thm6}, $d_k = d_0$, as $n=1$. But $1 \\leq d_0 \\leq C$. Thus,\n$$1 \\leq d_k \\leq C \\hspace{0.3in} \\text{ for } \\hspace{0.1in} k \\geq 1$$\nAs $C>1$, there is no specific value of $d_k$ that we can find for any $k$. Thus, the trick $(C, 1, 0, k)$ is not solvable.\n\\end{proof}\n\n\\begin{theorem} (Step (5)) \\label{thm:step5}\nIf $C, n > 1$, then $(C, n, 0, k)$ is solvable for any integer $k \\geq \\log_nC$. In this case, $(C, n, 0, k) = 1$.\n\\end{theorem}\n\n\\begin{proof}\nFrom Theorem \\ref{thm6} with $j = 0$, we have \n$$d_k = \\ceil[\\Big]{\\frac{d_0}{n_k}}$$\nAs $1 \\leq d_0 \\leq C$, therefore, \n\\begin{alignat*}{2}\n\\dfrac{1}{n^k} &\\leq & \\dfrac{d_0}{n^k} \\, \\,&\\leq \\dfrac{C}{n^k} \\\\\n\\ceil[\\Big]{\\dfrac{1}{n^k}} &\\leq & \\, \\ceil[\\Big]{\\dfrac{d_0}{n^k}} &\\leq \\ceil[\\Big]{\\dfrac{C}{n^k}} \n\\end{alignat*}\nAs $n>1$, therefore $0 < \\dfrac{1}{n^k} < 1$. Hence,\n\\begin{alignat*}{2}\n1 &\\leq & \\, d_k & \\leq \\ceil[\\Big]{\\dfrac{C}{n^k}} \\numberthis \\label{step5:ineq1}\n\\end{alignat*}\n\n\\noindent\nNow, $k \\geq \\log_nC$ and $C>1$. This implies, \n\\begin{align*}\nn^k &\\geq C \\\\\n0 < & \\frac{C}{n^k} \\leq 1\n\\end{align*}\nTherefore, $\\ceil[\\Big]{\\dfrac{C}{n^k}} = 1$. Using this result in inequality \\eqref{step5:ineq1}, we have\n$$1 \\leq d_k \\leq 1$$\nThus, $d_k = 1$. This shows that after $k$ iterations, we have a specific value for the the deck id $d_k$, and this value is the position of the desired card which happens to be the first card in the deck. Hence, the trick $(C, n, 0, k)$ is solvable for any integer $k \\geq \\log_nC$, and $(C, n, 0, k) = 1$. \n\\end{proof}\n\n\\begin{theorem} (Step (6)) \\label{thm:step6}\nIf $C, n > 1$, then $(C, n, n-1, k)$ is solvable for any integer $k > \\log_n(C-1)$. In this case, $(C, n, n-1, k) = C$. \n\\end{theorem}\n\n\\begin{proof}\nHere $j = n-1$. Therefore, by Theorem \\ref{thm6},\n\\begin{equation*}\n\\begin{split}\nd_k = m(n-1) + \\ceil[\\Bigg]{\\frac{m(n-1)n\\Big(\\dfrac{n^{k-1}-1}{n-1}\\Big) + d_0}{n^k}} & = m(n-1) + \\ceil[\\Bigg]{\\frac{m(n^k-n)}{n^k} + \\frac{d_0}{n^k}} \\\\\n& = m(n-1) + \\ceil[\\Bigg]{m + \\frac{d_0 - mn}{n^k}}\n\\end{split}\n\\end{equation*}\nAs $m \\in \\mathbb{Z}$, therefore, by Lemma \\ref{lem2}, \n\\begin{equation*}\n\\begin{split}\nd_k = m(n-1) + \\ceil[\\Bigg]{m + \\frac{d_0 - mn}{n^k}} &= m(n-1) + m + \\ceil[\\Bigg]{\\frac{d_0 - mn}{n^k}} \\\\\n&= mn + \\ceil[\\Bigg]{\\frac{d_0 - C}{n^k}} \\hspace{0.2in} \\text{as } C = mn.\n\\end{split}\n\\end{equation*}\n$d_0$ is the initial deck id before any iteration. Hence $1 \\leq d_0 \\leq C$. This implies,\n\\begin{alignat*}{2}\n1-C &\\leq & \\, d_0 - C \\,\\,\\, &\\leq 0 \\\\\n\\ceil[\\Bigg]{\\frac{1-C}{n^k}} &\\leq & \\, \\ceil[\\Bigg]{\\frac{d_0 - C}{n^k}} &\\leq 0 \\numberthis \\label{step6:ineq1}\n\\end{alignat*}\n\n\\noindent\nWe are told that $k > \\log_n(C-1)$, hence $n^k >C-1$. This implies, \n\\begin{equation*}\n\\begin{split}\n\\frac{1-C}{n^k} &> -1 \\\\\n\\ceil[\\Bigg]{\\frac{1-C}{n^k}} & > -1\n\\end{split}\n\\end{equation*}\nThis along with inequality $\\ceil[\\Bigg]{\\dfrac{1-C}{n^k}} \\leq 0$ from \\eqref{step6:ineq1}, implies that $\\ceil[\\Bigg]{\\dfrac{1-C}{n^k}} = 0$. Thus,\n$$0 \\leq \\ceil[\\Bigg]{\\frac{d_0 - C}{n^k}} \\leq 0$$\nwhich means $\\ceil[\\Bigg]{\\dfrac{d_0 - C}{n^k}} = 0$. Therefore,\n$$d_k = mn = C$$\nThis shows that after $k$ iterations, we have a specific value of the deck id $d_k$, and this value is the position of the desired card from the top of the deck. Thus, the trick $(C, n, n-1, k)$ is always solvable for $k > \\log_n(C-1)$, and $(C, n, n-1, k) = C$.\n\n\n\\end{proof}\n\n\\begin{theorem} (Step (7)) \\label{thm:step7}\nSuppose $m = \\dfrac{C}{n}$ and $b = \\dfrac{mj}{n-1}$. If $C, n > 1$ and $0 \\log_nt$, where $t = \\max\\Big\\{\\dfrac{C-bn}{1-\\{b\\}}, \\dfrac{bn-1}{\\{b\\}}\\Big\\}$ and, $(C, n , j, k) = mj + \\floor{b} + 1$. \n\\end{theorem}\n\n\\begin{proof}\nBefore we proceed further, we first verify that each of these logarithms are defined. As $0 < j < n-1$, hence,\n\\begin{align*}\nmj & < m(n-1) \\\\\nb(n-1) & < m(n-1) \\\\\nb &< m \\qquad \\qquad \\text{ as } \\hspace{0.1in} n > 1 \\\\\nbn &< mn \\\\\nC &> bn \\qquad \\qquad \\text{ as } \\hspace{0.1in} C = mn\n\\end{align*}\nAlso, when the trick is solvable then $b \\not\\in \\mathbb{Z}$. So, $0 < \\{b\\} < 1$, hence $1 - \\{b\\} > 0$. Thus $\\dfrac{C-bn}{1-\\{b\\}} > 0$, and hence, $\\log_n\\Big(\\dfrac{C-bn}{1-\\{b\\}} \\Big)$ is defined. For the other logarithm, we know that $C \\geq n$, thus, $C > n-1$, and since $0 < j < n-1$, hence,\n\\begin{align*}\nCj &> n-1 \\\\\n\\frac{Cj}{n-1} &> 1 \\\\\n\\frac{mnj}{n-1} &>1 \\\\\nbn & > 1 \\qquad \\qquad \\text{ as } \\hspace{0.1in} \\frac{mj}{n-1} = b\n\\end{align*}\nAs $\\{b\\} > 0$, hence, $\\dfrac{bn-1}{\\{b\\}} > 0$, and thus, $\\log_n\\Big(\\dfrac{bn-1}{\\{b\\}}\\Big)$ is defined.\n\n\\noindent\nNow we come back to the main proof. \n\\begin{center}\n\\underline{\\textbf{Part 1}}\n\\end{center}\nWe first show that if $(n-1) \\nmid mj$ then $(C, n, j, k)$ is solvable. As $(n-1) \\nmid mj$, thus $b \\not\\in \\mathbb{Z}$. Hence $0<\\{b\\} < 1$. From Theorem \\ref{thm6}, we have,\n\\begin{align*}\nd_k = mj + \\ceil[\\Bigg]{\\frac{b(n-1)n\\Big(\\dfrac{n^{k-1}-1}{n-1}\\Big) + d_0}{n^k}} & = mj + \\ceil[\\Bigg]{\\frac{b(n^k-n)}{n^k} + \\frac{d_0}{n^k}} \\\\\n& = mj + \\ceil[\\Bigg]{b + \\frac{d_0 - bn}{n^k}}\n\\end{align*}\nAs $b = \\floor{b} + \\{b\\}$ and $\\floor{b} \\in \\mathbb{Z}$, hence, by Lemma \\ref{lem2},\n$$d_k = mj + \\floor{b} + \\ceil[\\Bigg]{\\{b\\} + \\frac{d_0 - bn}{n^k}}$$\nAs usual, $1 \\leq d_0 \\leq C$. Thus, \n\\begin{alignat*}{2}\n\\{b\\} + \\frac{1 - bn}{n^k} &\\leq &\\{b\\} + \\frac{d_0 - bn}{n^k} \\, \\, \\, &\\leq \\{b\\} + \\frac{C - bn}{n^k} \\numberthis \\label{step7:ineq1}\\\\\n\\text{Hence, } \\qquad \\ceil[\\Bigg]{\\{b\\} + \\frac{1 - bn}{n^k}} &\\leq &\\ceil[\\Bigg]{\\{b\\} + \\frac{d_0 - bn}{n^k}} &\\leq \\ceil[\\Bigg]{\\{b\\} + \\frac{C - bn}{n^k}} \\numberthis \\label{step7:ineq2}\n\\end{alignat*}\nNow suppose $t = \\dfrac{C-bn}{1-\\{b\\}}$. Thus $k > \\log_n\\Big(\\dfrac{C-bn}{1-\\{b\\}} \\Big) \\geq \\log_n\\Big(\\dfrac{bn-1}{\\{b\\}}\\Big)$. This implies,\n\\begin{align*}\nn^k &> \\frac{C-bn}{1-\\{b\\}} \\qquad \\text{ and } \\qquad n^k > \\frac{bn-1}{\\{b\\}}\\\\\n\\frac{C-bn}{n^k} &<1-\\{b\\} \\qquad \\text{ and } \\qquad \\frac{1-bn}{n^k} > -\\{b\\}\\\\\n\\{b\\} + \\frac{C-bn}{n^k} &<1 \\qquad \\qquad \\text{ and } \\qquad \\{b\\} + \\frac{1-bn}{n^k} > 0\\\\\n\\end{align*}\nWe have already seen that $C>bn$ and $\\{b\\} >0$. Thus,\n$$0 < \\{b\\} + \\frac{C-bn}{n^k} < 1$$\nThis implies,\n\\begin{align*}\n\\ceil[\\Bigg]{\\{b\\} + \\frac{C - bn}{n^k}} = 1 \\numberthis \\label{step7:ineq3}\n\\end{align*}\nFrom inequality \\eqref{step7:ineq1}, we see that, \n$$\\{b\\} + \\frac{1 - bn}{n^k} \\leq \\{b\\} + \\frac{C - bn}{n^k} < 1$$\nThus,\n$$0 < \\{b\\} + \\frac{1-bn}{n^k} < 1$$ \nThis again implies,\n\\begin{align*}\n\\ceil[\\Bigg]{\\{b\\} + \\frac{1 - bn}{n^k}} = 1 \\numberthis \\label{step7:ineq4}\n\\end{align*}\nUsing results from \\eqref{step7:ineq3} and \\eqref{step7:ineq4} in inequality \\eqref{step7:ineq2}, we have\n$$1 \\leq \\ceil[\\Bigg]{\\{b\\} + \\frac{d_0 - bn}{n^k}} \\leq 1$$\nTherefore,\n$$ \\ceil[\\Bigg]{\\{b\\} + \\frac{d_0 - bn}{n^k}} = 1$$\nand $d_k = mj + \\floor{b} + 1$. As there is a specific value of $d_k$ after $k$ iterations, thus $(C, n, j, k)$ is solvable in this case with $(C, n, j, k) = mj + \\floor{b} + 1$. \n\n\\enspace\n\\noindent\nNow suppose $t = \\dfrac{bn-1}{\\{b\\}}$. Thus $k > \\log_n\\Big(\\dfrac{bn-1}{\\{b\\}}\\Big) \\geq \\log_n\\Big(\\dfrac{C-bn}{1-\\{b\\}} \\Big)$. We follow the same argument as before to conclude that \n$$ \\ceil[\\Bigg]{\\{b\\} + \\frac{d_0 - bn}{n^k}} = 1$$\nand $d_k = mj + \\floor{b} + 1$. Again, as there is a specific value of $d_k$ after $k$ iterations, thus $(C, n, j, k)$ is solvable in this case with $(C, n, j, k) = mj + \\floor{b} + 1$. \n\n\n\\noindent \nNow we prove the other direction.\n\\begin{center}\n\\underline{\\textbf{Part 2}}\n\\end{center}\nWe have to prove that for $C, n>1$ and $01$ and $0 bn$. Thus,\n\\begin{align*}\n\\frac{C-bn}{n^k} &> 0 \\\\\n\\ceil[\\Bigg]{\\frac{C-bn}{n^k}} & > 0\n\\end{align*}\nSimilarly, we have also seen that $bn > 1$. Hence,\n\\begin{align*}\n\\frac{1-bn}{n^k} &< 0 \\\\\n\\ceil[\\Bigg]{\\frac{1-bn}{n^k}} & \\leq 0\n\\end{align*}\nAs $\\ceil[\\Bigg]{\\dfrac{C-bn}{n^k}} \\in \\{1, 2, 3, \\ldots\\}$ and $\\ceil[\\Bigg]{\\dfrac{1-bn}{n^k}} \\in \\{0, -1, -2, \\ldots \\}$, hence from inequality \\eqref{step7:ineq6}, $\\ceil[\\Bigg]{\\dfrac{d_0 - bn}{n^k}}$ does not yield any specific integer for any $k$. Thus, $d_k = mj + b + \\ceil[\\Bigg]{\\dfrac{d_0 - bn}{n^k}}$ is also not a specific integer for any $k$. Therefore, $(C, n, j, k)$ is not solvable.\n\n\\end{proof}\n\nNote that from theorems \\ref{thm:step3}, \\ref{thm:step5}, \\ref{thm:step6}, \\ref{thm:step7}, we see that the value of a solvable trick $(C, n, j, k)$ does not depend on $k$. Hence, two solvable tricks $(C_1, n_1, j_1, k_1)$ and $(C_2, n_2, j_2, k_2)$ will be considered the \\textbf{same trick} if $C_1 = C_2, n_1 = n_2,$ and $j_1 = j_2$.\n\n\\section{Other Interesting Results}\n\n\\noindent\nNow that we have a complete mathematical understanding of the trick, we ask ourselves some interesting questions. \n\\begin{enumerate}\n\\item Given a number of cards $C$, how how many solvable magic tricks are there?\n\\item How does our algorithm solve the 21CT?\n\\item Assuming $C, n > 1$, what choices of $C$ and $n$ will guarantee that $(C, n, j, k)$ is solvable for all $0 \\leq j \\leq n-1$ and an appropriate $k$.\n\\item Assuming $C, n > 1$, what choices of $C$ and $n$ will guarantee that $(C, n, j, k)$ is not solvable for any $0 < j < n-1$.\n\\end{enumerate}\nWe answer these questions in the form of a theorem and several corollaries. \n\\begin{theorem}\\label{thm:interestingresult1}\nFor a given number of cards $C \\geq 1$, the number of solvable magic tricks $p$, is given by \n$$p =\n\\begin{cases}\n1, & C = 1 \\\\\n\\sum\\limits_{\\substack{n>1 \\\\ n \\mid C }} \\Big(n+1 - \\gcd\\Big(\\frac{C}{n}, n-1\\Big)\\Big) & C>1, n>1\n\\end{cases}\n$$\n\\end{theorem}\n\n\\begin{proof}\n\\noindent\n\\underline{Case 1 ($C=1$):} If $C= 1$, then $n = 1$, therefore, there is only one trick $(1, 1, 0, k)$ which is solvable for any $k >0$ by Theorem \\ref{thm:step3}. Thus $p=1$.\n\n\\enspace\n\\noindent\n\\underline{Case 2 ($C>1$):} If $n = 1$, then there is again only one trick $(C, 1, 0, k)$ which is not solvable for any $k>0$ by Theorem \\ref{thm:step4}. Therefore, $p=0$. \nNow suppose $C>1$ and $n>1$. We choose a fixed $n$ that is a divisor of $C$. We will first find out how many tricks are not solvable. By Theorem \\ref{thm:step7}, if $0 < j < n-1$, then $(C, n, j, k)$ is not solvable for any $k$, if $(n-1) \\mid mj$ where $m = \\dfrac{C}{n}$. As $0 < j < n-1$, thus $(n-1) \\nmid j$. The possible values of $mj$ are $m, 2m, 3m, \\ldots, (n-2)m$. There are $n-2$ such values. We thus need to find out which multiples of $m$ are also multiples of $n-1$ in order for the trick $(C, n, j, k)$ to be not solvable. The answer happens to be the multiples of $\\lcm(m, n-1)$ from the definition of the least common multiple. However, we need to find out how many multiples of $\\lcm(m, n-1)$ exist in the set $\\{m, 2m, 3m, \\ldots, (n-2)m\\}$. This is given by $\\dfrac{m(n-1)}{\\lcm(m, n-1)} - 1$. We subtract 1 as $m(n-1)$ is one such multiple that does not belong in the set. From elementary number theory \\cite{Burton}, we see that $\\dfrac{m(n-1)}{\\lcm(m, n-1)} = \\gcd(m, n-1)$. Thus, the number of tricks $(C, n, j, k)$ that are not solvable for $0 < j < n-1$ is $\\gcd(m, n-1) - 1$. So, the number of tricks that are solvable for $0 < j < n-1$ is $(n-2) - (\\gcd(m, n-1) - 1) = n-1 - \\gcd(m, n-1)$. Now for $j = 0$ and $j=n-1$ we have seen that $(C, n, j, k)$ is solvable by Theorems \\ref{thm:step5} and \\ref{thm:step6}. These add 2 more solvable tricks. Hence, the total number of solvable tricks $(C, n, j, k)$ for a fixed $n$, and an appropriate $k$ is \n$$n-1 - \\gcd(m, n-1) + 2 = n+1 -\\gcd\\Big(\\frac{C}{n}, n-1\\Big)$$\nAs $n > 1$ cycles through the divisors of $C$, we can see the total number of solvable tricks for a particular $C$ to be \n$$p = \\sum\\limits_{\\substack{n>1 \\\\ n \\mid C }} \\Big(n+1 - \\gcd\\Big(\\frac{C}{n}, n-1\\Big)\\Big)$$\n\\end{proof}\n\n\\begin{corollary}\nThe 21CT, $(21, 3, 1, 3)$ is solvable.\n\\end{corollary}\n\n\\begin{proof}\nWe use Step (7) of our algorithm as $C, n >1$ and $0 < j < 2$. Here $m = 21/3 = 7$, $mj = 7\\cdot 1 = 7$, and $n-1 = 2$. As, $ 2 \\nmid 7$, hence $(21, 3, 1, k)$ is solvable for any integer $k > \\log_nt$. We now find $t$. In order to find $t$, we need to know $b$ and $\\{b\\}$. $b = \\dfrac{mj}{n-1} = \\dfrac{7}{2}$. Hence, $\\floor{b} = 3$, and $\\{b\\} = \\dfrac{1}{2}$.\n\n\\noindent\n$t = \\max \\Bigg\\{ \\dfrac{21-\\Big(\\dfrac{7}{2}\\Big)\\cdot 3}{1-\\dfrac{1}{2}}, \n \\dfrac{\\Big(\\dfrac{7}{2}\\Big)\\cdot 3 - 1}{\\dfrac{1}{2}}\\Bigg\\} = \\max \\{21, 19\\} = 21$. As $k > \\log_3 21 = 2.771$, hence, $k = 3$. Thus, the 21CT, $(21, 3, 1, 3)$ is solvable. Moreover, $d_3 = mj + \\floor{b} + 1 = 7 + 3 + 1 = 11$. Therefore, $(21, 3, 1, 3) = 11$.\n\\end{proof}\n\n\n\\begin{corollary}\\label{cor:intresults3}\nSuppose $C, n > 1$. If $\\gcd\\Big((n-1), \\dfrac{C}{n}\\Big) = 1$, then $(C, n, j, k)$ is solvable for all $0 \\leq j \\leq n-1$ and an appropriate $k$. \n\\end{corollary} \n\n\\begin{proof}\nBy Theorem \\ref{thm:interestingresult1}, the number of solvable tricks for a specific $n$, is equal to $n+1-\\gcd\\Big(\\dfrac{C}{n}, n-1\\Big)$. As $\\gcd\\Big(\\dfrac{C}{n}, n-1\\Big) = 1$, hence the number of solvable tricks is $n+1 - 1 = n$ which are for all the possible values of $j = 0, 1, 2, \\ldots, n-1$. Hence, $(C, n, j, k)$ is solvable for all $0 \\leq j \\leq n-1$ and an appropriate $k$.\n\n\\end{proof}\n\n\\begin{comment}\n\\begin{corollary}\nSuppose $C, n > 1$. If $n=C$, then $(C, n, j, k)$ is solvable for all $0 \\leq j \\leq n-1$ and $k \\geq 1$. In this case, $(C, n, j, k) = j+1$. \n\\end{corollary}\n\n\\begin{proof}\nHere $\\gcd(n-1, \\dfrac{C}{n}) = \\gcd(C-1, 1) = 1$. Hence, by Corollary \\ref{cor:intresults3}, $(C, n, j, k)$ is solvable for all $0 \\leq j \\leq n-1$. \n\\end{proof}\n\n\\end{comment}\n\n\\begin{corollary}\nSuppose $C, n > 1$. If $C = n(n-1)$, then $(C, n, j, k)$ is not solvable for any $0 < j < n-1$. To find \n\\end{corollary}\n\n\\begin{proof}\nUsing previous notation, $m = \\dfrac{C}{n}$, we see that $C = n(n-1)$ is the same as $m = n-1$. Thus, $n-1 \\mid mj$ for all $0 < j < n-1$. Hence, by Theorem \\ref{thm:step7}, $(C, n, j, k)$ is not solvable.\n\\end{proof}\n\n\\section{Conclusion}\n\nWe have found a complete answer to the \\hyperref[gen-ques]{general question} posed at the start of Section 4 (Main Result). The \\hyperref[algorithm]{algorithm} presented in that section shows us a way of determining if a trick is solvable while starting with $C$ cards, split into $n$ stacks, with $j$ stacks going on top of the stack with the desired card after every iteration. In brief, here is a summary of questions that are answered in this paper.\n\\begin{itemize}\n\\item If Magi is handed $C$ $(>1)$ cards and is asked to split them into $n$ $(>1)$ stacks, without any specific $j$, then he can perform the trick $(C, n, 0, k)$ or $(C, n, n-1, k)$ for appropriate $k$'s as they are solvable due to Theorem \\ref{thm:step5}, and \\ref{thm:step6} respectively.\n\\item If Magi is handed $C$ $(>1)$ cards and is asked to split them into $n$ $(>1)$ stacks, with a specific $0 |C|^5 - |C|^4$, then $C$ is a minimal complement in $G$. \n\\end{theorem}\n\n\\begin{corollary}\\label{Cor:SmallInNon-abelian}\nAny nonempty finite subset of any infinite group is a minimal complement. \n\\end{corollary}\n\nThe above corollary generalizes \\cite[Theorem 9]{Kwon}, \\cite[Theorem B]{CoMin1}, \\cite[Theorem 2]{AlonKravitzLarson}. \nMoreover, increasing our understanding of which sets occur as minimal complements, the following result is also shown in section \\ref{Sec:MinComp}.\n\n\n\\begin{theorem}\\label{Thm:UnionOfCosetsInZ^d}\nLet $d, n, k$ be a positive integers such that \n$$k \\leq \\frac{n^{d/3}} {2(d \\log_2 n)^{2/3}}.$$\nLet $\\ensuremath{\\mathcal{X}}_1, \\cdots, \\ensuremath{\\mathcal{X}}_k$ be subsets of $\\ensuremath{\\mathbb{Z}}^d$ which are minimal complements in $\\ensuremath{\\mathbb{Z}}^d$. Let $c_1, \\cdots, c_k$ be elements of $\\ensuremath{\\mathbb{Z}}^d$ which are pairwise distinct modulo $n\\ensuremath{\\mathbb{Z}}^d$. Then \n$\\cup_{1 \\leq i \\leq k} (c_i + n\\ensuremath{\\mathcal{X}}_i)$ \nis a minimal complement in $\\ensuremath{\\mathbb{Z}}^d$. \nMoreover, if any nonempty subset of $\\ensuremath{\\mathbb{Z}}^d$ having finite symmetric difference with any one of $\\ensuremath{\\mathcal{X}}_1, \\cdots, \\ensuremath{\\mathcal{X}}_k$ is a minimal complement in $\\ensuremath{\\mathbb{Z}}^d$, then any nonempty subset of $\\cup_{1\\leq i \\leq k} (c_i + n\\ensuremath{\\mathbb{Z}}^d)$ having finite symmetric difference with \n$\\cup_{1 \\leq i \\leq k} (c_i + n\\ensuremath{\\mathcal{X}}_i)$ \nis a minimal complement in $\\ensuremath{\\mathbb{Z}}^d$. \n\\end{theorem}\n\n\n\nLet $G$ be a finite group of order $n$ and consider the collection $\\mathcal{C}$ of all non-empty subsets of $G$ which occur as minimal complements. There are several immediate questions about the elements of $\\mathcal{C}$, e.g., what are the sizes of the elements of $\\mathcal{C}$, what are the asymptotic properties of the sizes as $n\\rightarrow \\infty$, what are the integers $k$ between $1$ and $n$ such that any subset (or some subset) of $G$ of size $k$ is a minimal complement? \nIn a prior work of the authors, some such questions were asked \\cite[Question 1]{CoMin1}. \nOne can also study these questions by restricting to particular classes of groups, e.g., in the context of cyclic groups, or abelian groups, or non-abelian groups. We shall investigate these questions and provide partial answers to some of them in section \\ref{Sec:Limit}. In section \\ref{Sec:AsyCoMin}, we shall study the above questions for co-minimal pairs (see Definition \\ref{Def:CoMin}). \n\n\\section{Background literature} \n\\label{Sec:BackLit}\nTo study the asymptotic behaviour of minimal complements and that of co-minimal pairs we shall repeatedly use some previous results. Most of them are very recent and not yet in widespread use, so it is worthwhile to collect them here.\n\n\\begin{theorem}\n[{\\cite[Theorem 1]{AlonKravitzLarson}}]\\label{Thm:SmallWorks}\nLet $G$ be a group of order $n\\geq 2$. If $C$ is a nonempty subset of $G$ of size \n$$\\leq \\frac{n^{1/3}}{2(\\log_2 n)^{2/3}},$$\nthen $C$ is a minimal left complement to some subset in $G$ and it is a minimal right complement to some subset in $G$. \n\\end{theorem}\n\n\\begin{proposition}\n[{\\cite[Proposition 13]{AlonKravitzLarson}}]\n\\label{Prop:23rdBdd}\nLet $G$ be a finite group. If a subset $C$ of $G$ is a minimal complement to some subset $W$, \nthen \n$$|C| \n\\leq |G| \\frac{|W|} {2|W| -1}.$$\n\\end{proposition}\n\n\nAlon, Kravitz and Larson showed the above results in the context of abelian groups, but it can be seen that their proof extends to the setting when $G$ is not assumed to be abelian (by replacing the sums of the form $a+b$ (resp. $a-b$) by $a\\cdot b$ (resp. $a\\cdot b^{-1}$) and the sets of the form $a+B$ (resp. $a-B$) by $a\\cdot B$ (resp. $a\\cdot B^{-1}$). Recently, Burcroff and Luntzlara have proved a result which is more general than \nProposition \\ref{Prop:23rdBdd} in the context of abelian groups \n\\cite[Lemma 5]{BurcroffLuntzlara}. \n\n\n\n\\begin{theorem}\n[{\\cite[Theorem B]{CoMin1}}]\nGiven any nonempty subset $S$ of a group $G$ with $|S|\\leq 2$, there are subsets $L, R$ of $G$ such that $(S, R), (L, S)$ are co-minimal pairs. \n\\end{theorem}\n\n\\begin{proposition}\n[{\\cite[Proposition 2.17]{CoMin4}}]\n\\label{Prop:fini}\nLet $G$ be a finite group and $C$ be a subset of a subgroup $H$ of $G$ satisfying \n$$ |H| > |C| > 2[G:H] |H\\setminus C|.$$\nThen $C$ is not a minimal complement to any subset of $G$. \n\\end{proposition}\n\nProposition \\ref{Prop:fini} was first established by Alon, Kravitz and Larson in the context of abelian groups \\cite[Proposition 17]{AlonKravitzLarson}. \n\n\n\\begin{proposition}\n\\label{Prop:CoMinCartesian}\nIf $G_1, G_2$ are groups and $(A_1, B_1)$ (resp. $(A_2, B_2)$) is a co-minimal pair in $G_1$ (resp. $G_2$), then $(A_1 \\times A_2, B_1\\times B_2)$ is a co-minimal pair in $G_1\\times G_2$. \n\\end{proposition}\n\n\nThe above result is a special case of \\cite[Proposition 3.2]{CoMin1}. \n\n\\section{Sets occurring as minimal complements}\n\\label{Sec:MinComp}\n\nIn this section, we exhibit several sets that occur as minimal complements. \n\n\n\\begin{theorem}\n\\label{Thm:MinCompTwoSidedTra}\nLet $C$ be a nonempty subset of a group $G$. Assume that the set $C$ contains a right translate of itself only if it is equal to $C$. If for each $c\\in C$, there exists an element $g_c\\in G$ such that the sets $g_c \\cdot C^{-1} \\cdot C$ are pairwise disjoint, then $C$ is a minimal right complement in $G$. \nMoreover, if $C$ is finite and \n$$|G| > |C|^5 - |C|^4,$$\nthen $C$ is a minimal complement in $G$. \n\\end{theorem}\n\n\n\n\n\n\\begin{proof}\nFor $c\\in C$, let $w_c$ denote the element $g_c \\cdot c^{-1}$. It follows that the sets $w_c\\cdot c \\cdot C^{-1} \\cdot C$ are pairwise disjoint. \nIt also follows that the sets $w_c\\cdot c \\cdot C^{-1}$ are pairwise disjoint. \nLet $W$ denote the union of the sets $\\{w_c\\,|\\,c\\in C\\}$ and $G \\setminus (\\cup_{c\\in C} w_c \\cdot c \\cdot C^{-1})$. \nChoose an element $z$ of $G$. If $G \\setminus (\\cup_{c\\in C} w_c \\cdot c \\cdot C^{-1})$ contains some element of $z\\cdot C^{-1}$, then $W\\cdot C$ contains $z$. Suppose $z\\cdot C^{-1}$ is contained in $\\cup_{c\\in C} w_c \\cdot c \\cdot C^{-1}$. \nSince the sets $w_c\\cdot c \\cdot C^{-1} \\cdot C$ are pairwise disjoint, it follows that $z\\cdot C^{-1}$ is contained in $w_c \\cdot c \\cdot C^{-1}$ for exactly one $c\\in C$. By the hypothesis, we obtain $z\\cdot C^{-1} = w_c\\cdot c \\cdot C^{-1}$. So $z$ belongs to $W\\cdot C$. \nIt follows that $W\\cdot C = G$. Since the sets $w_c\\cdot c \\cdot C^{-1}$ are pairwise disjoint, it follows that $w_c \\cdot c \\notin w_d \\cdot C$ for any two distinct $c, d \\in C$. So, $C$ is a minimal right complement to $W$. \n\nNote that given finite subsets $A_1, \\cdots, A_r$ of $G$, there exist elements $g_1, \\cdots, g_r$ in $G$ such that $g_1\\cdot A_1, \\cdots, g_r\\cdot A_r$ are pairwise disjoint \nif \n$$|G| > |A_1 \\cdot A_s^{-1} | + \\cdots + |A_{s-1} \\cdot A_s^{-1}| $$\nholds for any $1< s \\leq r$. \nIf $|G| > |C|^5 - |C|^4$, then there exist $w_1, \\cdots, w_k$ in $G$ such that the sets $w_i\\cdot c_i \\cdot C^{-1} \\cdot C$ are pairwise disjoint. \nLet $W$ denote the union of the sets $\\{w_1, \\cdots, w_k\\}$ and $G \\setminus (\\cup_{1\\leq i \\leq k} w_i \\cdot c_i \\cdot C^{-1})$. \nChoose an element $z$ of $G$. \nSince the sets $w_i\\cdot c_i \\cdot C^{-1} \\cdot C$ are pairwise disjoint, at most one of $w_1 \\cdot c_1 \\cdot C^{-1}, \\cdots, w_k \\cdot c_k \\cdot C^{-1}$ intersects with $z\\cdot C^{-1}$. \nIf $w_i\\cdot c_i \\cdot C^{-1}$ intersects with $z\\cdot C^{-1}$ for some $i$, \nthen $z$ belongs to $W\\cdot C$ if $z\\cdot C^{-1} = w_i\\cdot c_i \\cdot C^{-1}$, \nand $z$ belongs to $W\\cdot C$ if $z\\cdot C^{-1} \\neq w_i\\cdot c_i \\cdot C^{-1}$. \nIf none of $w_1 \\cdot c_1 \\cdot C^{-1}, \\cdots, w_k \\cdot c_k \\cdot C^{-1}$ intersects with $z\\cdot C^{-1}$, then $z\\in W\\cdot C$. It follows that $W\\cdot C = G$. Since the sets $w_i\\cdot c_i \\cdot C^{-1} \\cdot C$ are pairwise disjoint, it follows that the sets $w_i \\cdot c_i \\cdot C^{-1}$ are pairwise disjoint, and hence $w_i \\cdot c_i \\notin w_j \\cdot C$ for any two distinct $i, j$. So, $C$ is a minimal right complement to $W$. \n\n\\end{proof}\n\n\n\\begin{proof}[Proof of Theorem \\ref{Thm:SmallInNon-abelian}]\nIt follows from Theorem \\ref{Thm:MinCompTwoSidedTra}. \n\\end{proof}\n\n\n\\begin{proof}[Proof of Corollary \\ref{Cor:SmallInNon-abelian}]\nIt follows from Theorem \\ref{Thm:MinCompTwoSidedTra}. \n\\end{proof}\n\n\n\\begin{corollary}\nFor any $d\\geq 1$, any nonempty bounded subset of $\\ensuremath{\\mathbb{Q}}^d$ is a minimal complement. \n\\end{corollary}\n\n\\begin{proof}\nIt follows from Theorem \\ref{Thm:MinCompTwoSidedTra}. \n\\end{proof}\n\n\n\n\n\n\nOne of the crucial steps of the proof of \\cite[Theorem 16]{AlonKravitzLarson} is to establish the following. \n\n\\begin{proposition}\n\\label{Prop:ExistenceOfSKTuple}\nGiven any finite group $\\Gamma$ of order $n$, two integers $s\\geq 2, k\\geq 1$ satisfying \n$$\\frac{s^2 k^3}{n} + \\frac{e^s k^{3s}} {n^{s-1}} + k \n\\left(\n\\frac{s^2\nk^3}{n}\n\\right)^s\n<1$$\nand a subset $C = \\{c_1, \\cdots, c_k\\}$ of $\\Gamma$ of size $k$, there exists an $sk$-tuple \n$$\n\\mathbf w :=\n\\begin{pmatrix}\nw_1^{(1)} & w_1^{(2)} & \\cdots & w_1^{(s)}\\\\\nw_2^{(1)} & w_2^{(2)} & \\cdots & w_2^{(s)}\\\\\n\\vdots & \\vdots & \\ddots & \\vdots \\\\\nw_k^{(1)} & w_k^{(2)} & \\cdots & w_k^{(s)}\\\\\n\\end{pmatrix}\n$$\nwith values in $\\Gamma$ such that each of the following statements is false. \n\\begin{enumerate}\n\\item \nthere exist distinct pairs $(i, p), (j, q)$ with $1\\leq i, j\\leq k, 1\\leq p,q\\leq s$ such that $w_i^{(p)} \\cdot c_i \\in w_j^{(q)} \\cdot C$. \n\\item \nthere exist at least $s$ distinct pairs $(i, p)$ such that the corresponding sets $w_i^{(p)} \\cdot c_i \\cdot C^{-1} \\cdot C$ have a nonempty intersection.\n\\item \nthere exists an integer $1\\leq i \\leq k$ such that for any $1\\leq p \\leq s$, there exist $z\\in \\Gamma$, $1\\leq j \\leq k, j \\neq i, 1\\leq q\\leq s$ such that the following conditions hold. \n\\begin{itemize}\n\\item \n$$\n\\frac ks < \n|(w_i^{(p)} \\cdot c_i \\cdot C^{-1}) \\cap (z\\cdot C^{-1})| < k$$\n\\item \n$w_j^{(q)} \\cdot c_j \\cdot C^{-1}$ contains the first element\\footnote{The elements of $z\\cdot C^{-1}$ are ordered according to the order of the elements of $C$.} of $(z\\cdot C^{-1}) \\setminus \n(w_i^{(p)} \\cdot c_i \\cdot C^{-1})$. \n\\end{itemize}\n\\end{enumerate}\n\\end{proposition}\n\nThe above result is established by Alon, Kravitz and Larson in the context of abelian groups. Moreover, their argument also works without the hypothesis that the underlying group is abelian and yields the above result. \n\n\n\\begin{theorem}\\label{Thm:UnionOfCosets}\nLet $G$ be a group and $H$ be a normal subgroup of $G$ of index $n\\geq 1$. \nLet $C = \\{c_1, \\cdots, c_k\\} \\cdot H$ be a subset of $G$, which is the union of $k$ distinct cosets of $H$ in $G$. Let $\\ensuremath{\\mathcal{C}}_1, \\cdots, \\ensuremath{\\mathcal{C}}_k$ be subsets of $c_1 H, \\cdots, c_k H$ respectively such that for any $1\\leq i \\leq k$, $c_i^{-1} \\ensuremath{\\mathcal{C}}_i$ is a minimal right complement in $H$. If there exists an integer $s\\geq 2$ satisfying \n$$\\frac{s^2 k^3}{n} + \\frac{e^s k^{3s}} {n^{s-1}} + k \n\\left(\n\\frac{s^2\nk^3}{n}\n\\right)^s\n<1,\n$$\nthen the set \n$$\\ensuremath{\\mathcal{C}}:=\\cup_{1\\leq i \\leq k} \\ensuremath{\\mathcal{C}}_i$$\nis a minimal right complement in $G$. \n\\end{theorem}\n\n\\begin{proof}\nFor $1\\leq i \\leq k$, let $\\ensuremath{\\mathcal{W}}_i$ be a subset of $H$ such that $c_i^{-1} \\cdot \\ensuremath{\\mathcal{C}}_i$ is a minimal right complement to $\\ensuremath{\\mathcal{W}}_i$ in $H$. We will assume that $\\ensuremath{\\mathcal{W}}_i$ contains the identity element. \n\nIn the following, the image of an element $x$ of $G$ under the mod $H$ reduction map $G\\to G/H$ is denoted by $\\overline x$. \nBy Proposition \\ref{Prop:ExistenceOfSKTuple}, there exists an $sk$-tuple \n$$\n\\mathbf w :=\n\\begin{pmatrix}\nw_1^{(1)} & w_1^{(2)} & \\cdots & w_1^{(s)}\\\\\nw_2^{(1)} & w_2^{(2)} & \\cdots & w_2^{(s)}\\\\\n\\vdots & \\vdots & \\ddots & \\vdots \\\\\nw_k^{(1)} & w_k^{(2)} & \\cdots & w_k^{(s)}\\\\\n\\end{pmatrix}\n$$\nwith values in $G$ such that the following conditions hold. \n\\begin{enumerate}[(i)]\n\\item \nfor any two distinct pairs $(i, p), (j, q)$ with $1\\leq i, j\\leq k, 1\\leq p,q\\leq s$, $\\overline w_i^{(p)} \\cdot \\overline c_i \\notin \\overline w_j^{(q)} \\cdot \\overline C$. \n\\item \nthe number of pairs $(i, p)$ such that the corresponding sets $\\overline w_i^{(p)} \\cdot \\overline c_i \\cdot \\overline C^{-1} \\cdot \\overline C$ have a nonempty intersection, is $\\frac ks$ elements, then by the third condition, it follows that for some $i$, the first element of $(\\overline z\\cdot \\overline C^{-1}) \\setminus \n(\\overline w_i^{(p_i)} \\cdot \\overline c_i \\cdot C^{-1})$ is not contained in $\\overline w_j^{(q)}\\cdot \\overline c_j \\cdot \\overline C^{-1}$ for any $j \\neq i$ and for any $1\\leq q \\leq s$, and hence the union of the $k$ sets $\\overline w_1^{(p_1)} \\cdot \\overline c_1 \\cdot \\overline C^{-1}, \\cdots, w_k^{(p_k)} \\cdot \\overline c_k \\cdot \\overline C^{-1}$ does not contain $\\overline z\\cdot \\overline C^{-1}$, and consequently, $z$ is contained in $W\\cdot \\ensuremath{\\mathcal{C}}$. So $\\ensuremath{\\mathcal{C}}$ is a minimal right complement to $W$ in $G$. \n\\end{proof}\n\n\\begin{proof}[Proof of Theorem \\ref{Thm:UnionOfCosetsInZ^d}]\nIt follows from Theorem \\ref{Thm:UnionOfCosets}. \n\\end{proof}\n\n\n\\section{Study of asymptotic behaviour}\n\\label{Sec:Limit}\n\nFor a finite group $G$ of order $n$, let $\\ensuremath{\\mathcal{A}}(G)$ (resp. $\\ensuremath{\\mathcal{S}}(G)$) denote the set of positive integers between $1$ and $n$ such that for any (resp. for some) integer $k\\in \\ensuremath{\\mathcal{A}}(G)$ (resp. $\\ensuremath{\\mathcal{S}}(G)$), any (resp. some) subset of $G$ of size $k$ is a minimal complement. Note that the inclusion \n$$\\ensuremath{\\mathcal{A}}(G) \\subseteq \\ensuremath{\\mathcal{S}}(G)$$\nholds. \nThere are several immediate questions about the structure of these sets, and the common structure of these sets when $G$ runs over a certain class of groups. We explain them below. \n\nFor $* = \\ensuremath{\\mathrm{cyc}}$ (resp. $\\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}$), a finite group $G$ is said to be a $*$-group if it is cyclic (resp. abelian, nilpotent, supersolvable, solvable). \nFor $* = \\emptyset$, a finite group $G$ is said to be a $*$-group if it is satisfies no additional condition other than being a group. \n\n\nFor $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$ and for any positive integer $n$, consider the following subsets of $\\{1, 2, \\cdots, n\\}$ defined as follows. \n\\begin{align*}\n\\ensuremath{\\mathcal{A}}_n^*\n& := \\bigcap_{G \\text{ is a $*$-group of order }n}\\ensuremath{\\mathcal{A}}(G) ,\\\\\n\\ensuremath{\\mathcal{S}}_n^*\n& := \\bigcap_{G \\text{ is a $*$-group of order }n}\\ensuremath{\\mathcal{S}}(G).\n\\end{align*}\nThe sets $\\ensuremath{\\mathcal{A}}_n^\\emptyset, \\ensuremath{\\mathcal{S}}_n^\\emptyset$ are also denoted by $\\ensuremath{\\mathcal{A}}_n, \\ensuremath{\\mathcal{S}}_n$ respectively. \nIn \\cite[Question 1]{CoMin1}, the authors asked to determine the structures of the sets $\\ensuremath{\\mathcal{A}}_n^\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathcal{S}}_n^\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathcal{A}}_n^\\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathcal{S}}_n^\\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathcal{A}}_n, \\ensuremath{\\mathcal{S}}_n$. \nVery recently, some parts of this question have been answered by some of the results of Alon, Kravitz and Larson. They established that the sizes of the minimal complements in the group $\\ensuremath{\\mathbb{Z}}/n\\ensuremath{\\mathbb{Z}}$ are exactly $1, 2, \\cdots, \\lfloor 2n/3\\rfloor, n$ \\cite[p. 5]{AlonKravitzLarson}. This shows that \n\\begin{equation}\n\\label{Eqn:SnCyc}\n\\ensuremath{\\mathcal{S}}^\\ensuremath{\\mathrm{cyc}}_n\n=\n\\left\\{1, 2, \\cdots, \\left\\lfloor 2n/3\\right\\rfloor, n\\right\\}\n\\end{equation}\nfor $n\\geq 2$. \n\n\n\nIt would be interesting to investigate the structure of the sets $\\ensuremath{\\mathcal{X}}_n^*$, and the asymptotic behaviour of the sets $\\frac 1n \\ensuremath{\\mathcal{X}}_n^*$ for $\\ensuremath{\\mathcal{X}}\\in \\{\\ensuremath{\\mathcal{S}}, \\ensuremath{\\mathcal{A}}\\}, *\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$. Using the results of Section \\ref{Sec:BackLit}, one can conclude several results, as we describe below. \nFor $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$, one has the inclusion\n$$\\ensuremath{\\mathcal{A}}_n^* \\subseteq \\ensuremath{\\mathcal{S}}_n^*,$$\nand the inclusions\n\\begin{equation}\n\\label{Eqn:Inclu*}\n\\ensuremath{\\mathcal{X}}_n \n\\subseteq \\ensuremath{\\mathcal{X}}_n^\\ensuremath{\\mathrm{sol}}\n\\subseteq \\ensuremath{\\mathcal{X}}_n^\\ensuremath{\\mathrm{ssol}}\n\\subseteq \\ensuremath{\\mathcal{X}}_n^\\ensuremath{\\mathrm{nil}} \n\\subseteq \\ensuremath{\\mathcal{X}}_n^\\ensuremath{\\mathrm{ab}} \n\\subseteq \\ensuremath{\\mathcal{X}}_n^\\ensuremath{\\mathrm{cyc}}\n\\end{equation}\nhold for $\\ensuremath{\\mathcal{X}} = \\ensuremath{\\mathcal{A}}, \\ensuremath{\\mathcal{S}}$. \nMoreover, for $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}\\}$, \n\\begin{equation}\n\\label{Eqn:Inclumn}\nm \\ensuremath{\\mathcal{S}}_n^* \\subseteq \\ensuremath{\\mathcal{S}}_{mn}^*\n\\end{equation}\nholds for any positive integers $n,m$ with $\\gcd (m,n)=1$, \nand \n\\begin{equation}\n\\label{Eqn:InclumnDiv}\n\\frac 1n \\ensuremath{\\mathcal{S}}_n^* \\subseteq \\frac 1m \\ensuremath{\\mathcal{S}}_m^*\n\\end{equation}\nholds for any positive integers $n,m$ with $n\\mid m$ and $\\gcd(n, m/n) =1$ (see Lemma \\ref{Lemma:SnInclusionSm}). \n\n\\begin{lemma}\n\\label{Lemma:SnInclusionSm}\nEquations \\eqref{Eqn:Inclumn}, \\eqref{Eqn:InclumnDiv} hold. \n\\end{lemma}\n\n\\begin{proof}\nFix $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}\\}$, and let $k$ be an element of $\\ensuremath{\\mathcal{S}}_n^*$. \nLet $G$ be a $*$-group of order $mn$ where $m$ is a positive integer with $\\gcd(m,n)=1$. Since $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}\\}$ and $\\gcd(m, n) = 1$, it follows that $G$ is isomorphic to $G_1 \\times G_2$ where $G_1$ (resp. $G_2$) is a group of order $m$ (resp. $n)$. \nNote that $G_2$ contains a subset $A$ of size $k$, which is a minimal complement in $G_2$. Then the subset $G_1 \\times A$ of $G_1 \\times G_2$ contains $mk$ elements. By Proposition \\ref{Prop:CoMinCartesian}, $G_1 \\times A$ is a minimal complement to some subset of $G_1\\times G_2$. Hence $mk$ is an element of $\\ensuremath{\\mathcal{S}}(G_1\\times G_2)$. Thus $mk$ lies in $\\ensuremath{\\mathcal{S}}(G)$ for any $*$-group $G$ of order $mn$. So $mk$ lies in $\\ensuremath{\\mathcal{S}}_{mn}^*$. This establishes Equation \\eqref{Eqn:Inclumn}. \nEquation \\eqref{Eqn:InclumnDiv} follows from Equation \\eqref{Eqn:Inclumn}. \n\\end{proof}\n\nFor $\\ensuremath{\\mathcal{X}}\\in \\{\\ensuremath{\\mathcal{S}}, \\ensuremath{\\mathcal{A}}\\}, *\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$, it follows from Theorem \\ref{Thm:SmallWorks} that any large finite group contains many minimal complements and thus \n$$\\lim_{n\\to \\infty} |\\ensuremath{\\mathcal{X}}_n^*| = \\infty.$$\n\n\\begin{questionIntro}\n\\label{Qn:XnGetsLarger}\nFor $\\ensuremath{\\mathcal{X}}\\in \\{\\ensuremath{\\mathcal{S}}, \\ensuremath{\\mathcal{A}}\\}$ and $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$, describe the asymptotic property of the sequence $|\\ensuremath{\\mathcal{X}}_n^*|$. \n\\end{questionIntro}\n\n\n\n\nFor $\\ensuremath{\\mathcal{X}}\\in \\{\\ensuremath{\\mathcal{S}}, \\ensuremath{\\mathcal{A}}\\}, *\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$, it follows from Theorem \\ref{Thm:SmallWorks} that the smallest positive integer lying outside $\\ensuremath{\\mathcal{X}}_n^*$ diverges to $\\infty$, i.e., \n$$\\lim_{n\\to \\infty} \\min (\\{1, 2, \\cdots, n\\} \\setminus \\ensuremath{\\mathcal{X}}_n^*)= \\infty.$$\n\n\n\n\\begin{questionIntro}\n\\label{Qn:LargeStringAt1}\nFor $\\ensuremath{\\mathcal{X}}\\in \\{\\ensuremath{\\mathcal{S}}, \\ensuremath{\\mathcal{A}}\\}$ and $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$, describe the asymptotic property of the sequence \n$\\min (\\{1, 2, \\cdots, n\\} \\setminus \\ensuremath{\\mathcal{X}}_n^*).$\n\\end{questionIntro}\n\nFor $\\ensuremath{\\mathcal{X}} = \\ensuremath{\\mathcal{A}}$, $* = \\ensuremath{\\mathrm{ab}}$, it follows from \\cite[Corollary 18]{AlonKravitzLarson} that \n$$\n\\liminf_{n\\to \\infty} \n\\frac{\\min (\\{1, 2, \\cdots, n\\} \\setminus \\ensuremath{\\mathcal{X}}_n^*)}{\\sqrt n} \n\\leq \n\\sqrt 2,$$\nand from \\cite[Theorem 3]{AlonKravitzLarson} that \n$$\n\\min (\\{1, 2, \\cdots, n\\} \\setminus \\ensuremath{\\mathcal{X}}_n^*)\n= \nO(n^{3/4 + \\varepsilon})$$\nfor any $\\varepsilon > 0$. \nMoreover, Alon, Kravitz and Larson conjectured that \n$$\n\\min (\\{1, 2, \\cdots, n\\} \\setminus \\ensuremath{\\mathcal{X}}_n^*)\n= \n\\widetilde \\Theta(\\sqrt n)$$\nfor $\\ensuremath{\\mathcal{X}} = \\ensuremath{\\mathcal{A}}, * = \\ensuremath{\\mathrm{ab}}$ \\cite[Conjecture 7]{AlonKravitzLarson}. \n\nFrom Propositions \\ref{Prop:23rdBdd}, \\ref{Prop:fini}, it follows that for $\\ensuremath{\\mathcal{X}}\\in \\{\\ensuremath{\\mathcal{S}}, \\ensuremath{\\mathcal{A}}\\}, *\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$,\n$$\\lim_{n\\to \\infty} |(\\{1, 2, \\cdots, n\\} \\setminus \\ensuremath{\\mathcal{X}}_n^*)| = \\infty.$$\n\n\n\n\\begin{questionIntro}\n\\label{Qn:XnComplementGetsLarger}\nFor $\\ensuremath{\\mathcal{X}}\\in \\{\\ensuremath{\\mathcal{S}}, \\ensuremath{\\mathcal{A}}\\}, *\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$, determine the asymptotic property of the sequence \n$$ |(\\{1, 2, \\cdots, \\lfloor 2n/3\\rfloor \\} \\setminus \\ensuremath{\\mathcal{X}}_n^*)|.$$\n\\end{questionIntro}\n\nFrom Propositions \\ref{Prop:23rdBdd}, \\ref{Prop:fini}, it follows that for $\\ensuremath{\\mathcal{X}}\\in \\{\\ensuremath{\\mathcal{S}}, \\ensuremath{\\mathcal{A}}\\}, *\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$, the maximum of $\\ensuremath{\\mathcal{X}}_n^*$ (excluding $n$) is $\\leq \\frac 23n$ (for $n\\geq 2$). Thus for $\\ensuremath{\\mathcal{X}}\\in \\{\\ensuremath{\\mathcal{S}}, \\ensuremath{\\mathcal{A}}\\}, *\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$,\n$$\\left(\\frac 23, 1\\right) \\cap \\frac 1n \\ensuremath{\\mathcal{X}}_n^* = \\emptyset.$$\nMoreover, it follows that for $\\ensuremath{\\mathcal{X}}\\in \\{\\ensuremath{\\mathcal{S}}, \\ensuremath{\\mathcal{A}}\\}, *\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$ and for any $\\varepsilon > 0$, \n$$(0, \\varepsilon) \\cap \\frac 1n \\ensuremath{\\mathcal{X}}_n^* \\neq \\emptyset$$\nfor large enough $n$ (since $1$ lies in $\\ensuremath{\\mathcal{X}}_n^*$). \nThis motivates the following question about the asymptotic behaviour of \n$$\\frac 1n \\ensuremath{\\mathcal{X}}^*_n$$\nas $n\\to \\infty$ for $\\ensuremath{\\mathcal{X}}\\in \\{\\ensuremath{\\mathcal{S}}, \\ensuremath{\\mathcal{A}}\\}, *\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$, \nand the asymptotic behaviour of these sets when $n$ ranges over an infinite set of positive integers (for instance, the set of primes, or the set of all prime powers, or the set of powers of a fixed prime, or the set of square-free integers etc.). \n\n\n\\begin{questionIntro}\n\\label{Qn:Measure}\nLet $\\ensuremath{\\mathcal{X}}\\in \\{\\ensuremath{\\mathcal{S}}, \\ensuremath{\\mathcal{A}}\\}, *\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$. \n\\begin{enumerate}[(i)]\n\\item \nLet $0 \\leq a < b \\leq \\frac 23$. Evaluate \n$$\n\\limsup \\frac{[a, b] \\cap \\frac 1n \\ensuremath{\\mathcal{X}}_n^*}\n{|\\ensuremath{\\mathcal{X}}_n^*|}, \n\\liminf \\frac{[a, b] \\cap \\frac 1n \\ensuremath{\\mathcal{X}}_n^*}\n{|\\ensuremath{\\mathcal{X}}_n^*|}.\n$$\nDoes the sequence \n$$\n\\frac{[a, b] \\cap \\frac 1n \\ensuremath{\\mathcal{X}}_n^*}\n{|\\ensuremath{\\mathcal{X}}_n^*|}\n$$\nconverge? Otherwise, what are its subsequential limits?\n\n\\item \nDoes there exist a probability measure $\\mu$ on $[0, \\frac 23]$ such that \n$$\n\\lim_{n\\to \\infty} \n\\frac{[a, b] \\cap \\frac 1n \\ensuremath{\\mathcal{X}}_n^*}\n{|\\ensuremath{\\mathcal{X}}_n^*|}\n= \\int_{[0, \\frac 23]} \\chi_{[a, b]} d\\mu\n$$\nfor any $0 \\leq a < b \\leq \\frac 23$, where $\\chi_A$ denotes the characteristic function of $A$ for $A\\subseteq [0, \\frac 23]$?\n\\end{enumerate}\n\\end{questionIntro}\n\n\\begin{questionIntro}\n\\label{Qn:BddMinCompDetailed}\nLet $\\ensuremath{\\mathcal{X}}\\in \\{\\ensuremath{\\mathcal{S}}, \\ensuremath{\\mathcal{A}}\\}, *\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$. \n\\begin{enumerate}[(i)]\n\n\\item \nDetermine the open subsets of $[0, \\frac 23]$ which do not intersect with $\\frac 1n \\ensuremath{\\mathcal{X}}_n^*$ for any/large enough/infinitely many $n$. \n\n\\item \nDetermine the open subsets of $[0, \\frac 23]$ which have nonempty intersection with $\\frac 1n \\ensuremath{\\mathcal{X}}_n^*$ for any/large enough/infinitely many $n$.\n\n\\end{enumerate}\n\n\\end{questionIntro}\n\n\\begin{remark}\nIn the above questions, one can restrict the integer $n$ from the set of positive integers to some smaller sets, for instance, the set of primes, or the set of all prime powers, or the set of powers of a fixed prime, or the set of square-free integers etc., and study these questions when $n$ varies over such a smaller subset. \n\\end{remark}\n\n\n\nUsing results from Section \\ref{Sec:BackLit}, we partially answer Question \\ref{Qn:XnComplementGetsLarger} in Proposition \\ref{Prop:Qn3Ans}.\n\n\\begin{lemma}\n\\label{Lemma:B1}\nLet $n$ be a positive integer and $d_1, \\cdots, d_k$ be distinct divisors of $n$ satisfying $d_1\\mid d_2 \\mid \\cdots \\mid d_k$. \nFor $1\\leq i \\leq k$, let $B_i$ denote the set defined by \n$$\nB_i\n:=\n\\left\\{\n\\frac n{d_i} -1 , \\frac n{d_i} - 2, \\cdots, \\frac n{d_i} - \n\\left(\\left\\lceil \\frac n{d_i(2d_i + 1)} \\right\\rceil -1\\right)\n\\right\\}\n.\n$$\nThe union \n$\n\\cup_{i=1}^k B_i \n$\nconsists of \n$$\\sum_{i=1}^k \n\\left(\\left\\lceil \\frac n{d_i(2d_i + 1)} \\right\\rceil -1\\right)\n$$\nelements. \n\\end{lemma}\n\n\\begin{proof}\nNote that the sets $B_1, \\cdots, B_k$ are pairwise disjoint since for any $i< j$, \n\\begin{align*}\n\\frac n{d_i} - \\left(\\left\\lceil \\frac n{d_i(2d_i + 1)} \\right\\rceil -1\\right)\n& \\geq \n\\frac n{d_i} - \\frac n{d_i(2d_i + 1)} \\\\\n& = \\frac {2nd_i}{d_i(2d_i + 1)} \\\\\n& = \\frac {2n}{2d_i + 1} \\\\\n& > \\frac n{2d_i} \\\\\n& \\geq \n\\frac n{d_j} \n\\end{align*}\nholds. \nThis proves the Lemma. \n\\end{proof}\n\n\\begin{lemma}\n\\label{Lemma:B2}\nLet $G$ be a group of order $n$. Let $d_1, \\cdots, d_k$ be distinct divisors of $n$ such that $d_1\\mid \\cdots \\mid d_k$. Assume that $G$ contains a subgroup of size $n/d_i$ for any $1\\leq i \\leq k$. \nThen the set $\\{1, 2, \\cdots, n\\} \\setminus \\ensuremath{\\mathcal{A}}(G) $ contains the set \n$$\n\\cup_{i=1}^k B_i ,\n$$\nand hence contains at least \n$$\\sum_{i=1}^k \n\\left(\\left\\lceil \\frac n{d_i(2d_i + 1)} \\right\\rceil -1\\right)\n$$\nelements. \n\\end{lemma}\n\n\n\n\\begin{proof}\n\nNote that for any integer $r$ satisfying \n$$\n1\\leq r \\leq \n\\left(\\left\\lceil \\frac n{d_i(2d_i + 1)} \\right\\rceil -1\\right),\n$$\nthe inequality \n$$\n\\frac{\\frac n{d_i} -r}{r} \n> 2d_i $$\nholds, \nand hence by Proposition \\ref{Prop:fini}, $G$ contains a subset of size $n/d_i -r$ which is not a minimal complement. \nThus for any $1\\leq i \\leq k$, the set $B_i$ does not intersect with $\\ensuremath{\\mathcal{A}}(G)$. So the set $\\{1, 2, \\cdots, n\\} \\setminus \\ensuremath{\\mathcal{A}}(G) $ contains\n$$\n\\cup_{i=1}^k B_i . \n$$\nIts cardinality is given by Lemma \\ref{Lemma:B1}. \n\\end{proof}\n\n\\begin{lemma}\n\\label{Lemma:Bdd}\nLet $n$ be a positive integer, $p$ be a prime number and $M$ be a positive integer such that $p^M$ divides $n$. \nThen there exists a sequence of $k$ distinct divisors $d_1, \\cdots, d_k$ of $n$ satisfying \n$1< d_1 \\mid d_2 \\mid \\cdots \\mid d_k$ \nsuch that \n$$\\sum_{i=1}^k \n\\left(\\left\\lceil \\frac n{d_i(2d_i + 1)} \\right\\rceil -1\\right)\n\\geq \n\\frac n{p^2(2 + \\frac 1{p})} - M.\n$$\n\n\\end{lemma}\n\n\n\\begin{proof}\nLet $k = M$ and $d_i = p ^{i}$ for $1\\leq i\\leq M$. \nNote that \n\\begin{align*}\n\\sum_{i=1}^k \n\\left(\\left\\lceil \\frac n{d_i(2d_i + 1)} \\right\\rceil -1\\right)\n& \\geq \n\\sum_{i=1}^M \n\\left(\\frac n{d_i(2d_i + 1)} -1\\right)\n\\\\\n& \\geq \n\\sum_{i=1}^M \n\\frac n{(2 + \\frac 1{p})d_i^2} - M\n\\\\\n& = \n\\frac n{2 + \\frac 1{p}} \n\\sum_{i=1}^M \n\\frac 1{d_i^2} - M\n\\\\\n& = \n\\frac n{2 + \\frac 1{p}} \\frac 1{p^2} \\frac {1 - \\frac 1{p^{2M}}}{1 -\\frac 1{p^2}} - M\n\\\\\n& \n\\geq \n\\frac n{p^2(2 + \\frac 1{p})}- M .\n\\end{align*}\n\\end{proof}\n\n\\begin{proposition}\n\\label{Prop:Qn3Ans}\nLet $\\{n_k\\}_{k\\geq 1}$ be a sequence of positive integers. Assume that no term of this sequence gets repeated infinitely often, i.e., it does not admits any constant subsequence. Let $\\ensuremath{\\mathcal{P}}$ be a finite set of primes such that all the prime divisors of any term of this sequence lie in this set. \nThen for any $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$, \n$$|(\\{1, 2, \\cdots, \\lfloor 2n_k/3\\rfloor \\} \\setminus \\ensuremath{\\mathcal{A}}_{n_k}^*)| \n\\geq \n\\frac {n_k} {(\\max \\ensuremath{\\mathcal{P}})^2(2 + \\frac 1{\\min \\ensuremath{\\mathcal{P}}})} - \\frac{\\log n_k}{\\log \\min \\ensuremath{\\mathcal{P}}}\n$$\nholds for large enough $k$, and consequently, \n$$\\lim_{k\\to \\infty} |(\\{1, 2, \\cdots, \\lfloor 2n_k/3\\rfloor \\} \\setminus \\ensuremath{\\mathcal{A}}_{n_k}^*)| = \\infty.$$\n\\end{proposition}\n\n\n\n\\begin{proof}\nBy Equation \\eqref{Eqn:Inclu*}, it suffices to prove the above inequality for $* = \\ensuremath{\\mathrm{cyc}}$. \nSince the terms of the sequence $\\{n_k\\}_{k\\geq 1}$ have prime factors from a finite set of primes and no term of this sequence gets repeated infinitely often, it follows that for any $M> 0$, there exists a positive integer $K$ such that for each $k\\geq K$, \nthe integer $n_k$ is divisible by $p_k^M$ for some $p_k\\in\\ensuremath{\\mathcal{P}}$. \nBy Lemmas \\ref{Lemma:B1}, \\ref{Lemma:B2}, \\ref{Lemma:Bdd}, \nfor any $k\\geq K$, it follows that \nthere exists a subset $B$ of $\\{1, 2, \\cdots, \\lfloor 2n_k/3\\rfloor \\}$ containing at least \n\\begin{align*}\n\\frac {n_k} {p_k^2(2 + \\frac 1{p_k})} - M\n& = \n\\frac {n_k} {p_k^2(2 + \\frac 1{p_k})} - \\frac{\\log n_k}{\\log p_k}\\\\\n& \\geq \n\\frac {n_k} {(\\max \\ensuremath{\\mathcal{P}})^2(2 + \\frac 1{\\min \\ensuremath{\\mathcal{P}}})} - \\frac{\\log n_k}{\\log \\min \\ensuremath{\\mathcal{P}}}\n\\end{align*}\nmany elements such that $\\ensuremath{\\mathcal{A}}(G)$ avoids $B$ for any nilpotent group $G$ of order $n_k$. This establishes the result.\n\\end{proof}\n\nUsing Equation \\eqref{Eqn:SnCyc}, we partially answer Question \\ref{Qn:Measure} in Proposition \\ref{Prop:Qn4Ans}. \n\n\\begin{proposition}\n\\label{Prop:Qn4Ans}\nFor $\\ensuremath{\\mathcal{X}} = \\ensuremath{\\mathcal{S}}, * = \\ensuremath{\\mathrm{cyc}}$, \n$$\n\\lim_{n\\to \\infty} \n\\frac{(a, b) \\cap \\frac 1n \\ensuremath{\\mathcal{X}}_n^*}\n{|\\ensuremath{\\mathcal{X}}_n^*|} = \\frac 32( b-a),\n$$\nholds for $0 \\leq a < b \\leq \\frac 23$, and Question \\ref{Qn:Measure}(ii) admits an answer in the affirmative. \n\\end{proposition}\n\n\\begin{proof}\nFrom Equation \\eqref{Eqn:SnCyc}, it follows that \n$$\n\\lim_{n\\to \\infty} \n\\frac{(a, b) \\cap \\frac 1n \\ensuremath{\\mathcal{X}}_n^*}\n{|\\ensuremath{\\mathcal{X}}_n^*|} = \\frac 32( b-a).\n$$\nNote that for the probability measure $\\mu$ corresponding to the Lebesgue measure on $[0, \\frac 23]$, it follows that \n$$\n\\lim_{n\\to \\infty} \n\\frac{(a, b) \\cap \\frac 1n \\ensuremath{\\mathcal{X}}_n^*}\n{|\\ensuremath{\\mathcal{X}}_n^*|}\n= \\int_{[0, \\frac 23]} \\chi_{[a, b]} d\\mu\n$$\nfor any $0 \\leq a < b \\leq \\frac 23$.\nThis answers part (ii). \n\\end{proof}\n\nUsing Proposition \\ref{Prop:fini}, we prove the following lemma, and then establish Proposition \\ref{Prop:OpenAvoidance} which partially answers part (i) of Question \\ref{Qn:BddMinCompDetailed}. \n\\begin{lemma}\n\\label{Lemma:OpenAvoidance}\nLet $G$ be a group of order $n$. Let $i$ be an integer such that $G$ admits a subgroup of index $i$. \nThen \n$$\\left(\\frac 2{2i+1} , \\frac 1i\\right) \\cap \\frac 1{|G|} \\ensuremath{\\mathcal{A}}(G) = \\emptyset.$$\n\\end{lemma}\n\n\\begin{proof}\nLet $H$ be a subgroup of $G$ of index $i$. \nFor any integer $m$ satisfying\n$$\\frac {2i}{2i+1} |H| < m < |H|,$$\nit follows from Proposition \\ref{Prop:fini} that no subset of $H$ containing $m$ elements is a minimal complement in $G$. Thus \n$$\\left(\\frac {2i}{2i+1} |H|, |H|\\right) \\cap \\ensuremath{\\mathcal{A}}(G) = \\emptyset,$$ \nwhich yields the result. \n\\end{proof}\n\n\n\n\n\n\\begin{proposition}\n\\label{Prop:OpenAvoidance}\nFor $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$, \n$$\n\\bigcup_{i=0}^\\infty \\left(\\frac 2{2p^i+1} , \\frac 1{p^i}\\right) \n$$ \ndoes not intersect with $\\ensuremath{\\mathcal{A}}_{p^n}^*$ for any $n\\geq 0$. \n\n\\end{proposition}\n\n\\begin{proof}\nFor any group $G$ of order $p^n$, Lemma \\ref{Lemma:OpenAvoidance} implies that the set \n$$\n\\bigcup_{i=0}^n \\left(\\frac 2{2p^i+1} , \\frac 1{p^i}\\right) \n$$ \ndoes not intersect with \n$\\frac 1{|G|} \\ensuremath{\\mathcal{A}}(G)$. Since the smallest element of $\\frac 1{|G|} \\ensuremath{\\mathcal{A}}(G)$ is $\\frac 1{|G|}$, it follows that \n$$\n\\bigcup_{i=0}^\\infty \\left(\\frac 2{2p^i+1} , \\frac 1{p^i}\\right) \n$$ \ndoes not intersect with \n$\\frac 1{|G|} \\ensuremath{\\mathcal{A}}(G)$. \nHence \n$$\n\\bigcup_{i=0}^\\infty \\left(\\frac 2{2p^i+1} , \\frac 1{p^i}\\right) \n$$ \ndoes not intersect with none of \n$\\ensuremath{\\mathcal{A}}_{p^n}^*$\nfor any $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$. \n\\end{proof}\n\n\nNote that it follows from Equation \\eqref{Eqn:SnCyc} that given any nonempty open subset $U$ of $[0, \\frac 23]$, it has nonempty intersection with $\\frac 1n \\ensuremath{\\mathcal{S}}_n^\\ensuremath{\\mathrm{cyc}}$ for large enough $n$. \nThis partially answers part (ii) of Question \\ref{Qn:BddMinCompDetailed}. \n\n\n\n\\section{Asymptotic behaviour of co-minimal pairs}\n\\label{Sec:AsyCoMin}\n\n\\begin{definition}\\label{Def:CoMin}\nA pair $(A, B)$ of two nonempty subsets $A, B$ of a group $G$ is called a co-minimal pair if $A \\cdot B = G$, and $A' \\cdot B \\subsetneq G$ for any $\\emptyset \\neq A' \\subsetneq A$ and $A\\cdot B' \\subsetneq G$ for any $\\emptyset \\neq B' \\subsetneq B$. \n\\end{definition}\n\nFor any finite group $G$, let \n$\\ensuremath{\\mathcal{S}}_2(G)$\ndenote the \nset of pairs of the form $(a, b)$ such that there is a co-minimal pair $(A, B)$ in $G$ with $|A| = a, |B| = b$. \nFor $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$ and for any positive integer $n$, consider the following subset of $\\{1, 2, \\cdots, n\\}\\times \\{1, 2, \\cdots, n\\}$ defined as follows. \n\\begin{align*}\n\\ensuremath{\\mathcal{S}}_{2, n}^*\n& := \\bigcap_{G \\text{ is a $*$-group of order }n}\\ensuremath{\\mathcal{S}}_2(G).\n\\end{align*}\nThe set $\\ensuremath{\\mathcal{S}}_{2,n}^\\emptyset$ is also denoted by $\\ensuremath{\\mathcal{S}}_{2,n}$. \nBy Theorem \\ref{Thm:SmallWorks}, it follows that \n$$\\lim_{n\\to \\infty} |\\ensuremath{\\mathcal{S}}_{2,n}^*| = \\infty$$\nfor $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$. \n\n\\begin{questionIntro}\n\\label{Qn:XnGetsLarger2}\nFor $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$, describe the asymptotic property of the sequence $|\\ensuremath{\\mathcal{S}}_{2,n}^*|$. \n\\end{questionIntro}\n\nLet $I$ denote the unit interval $[0,1]$ and $I^2$ denote the unit square $[0,1]\\times [0,1]$. The square $[0, 1/2]\\times [0, 1/2]$ is denoted by $I^2_{1/2}$. \nIt follows from Proposition \\ref{Prop:23rdBdd} that for a co-minimal pair $(A, B)$ in any finite group $G$, \n$$\n2|A||B| - |A| \\leq |G| |B|\n$$\nholds. \nSo, for any $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$ and any $(x, y)\\in \\frac 1n \\ensuremath{\\mathcal{S}}_{2,n}^*$, \n$$\\left(x - \\frac 1{2n} \\right)\n\\left(y - \\frac 12\\right) \\leq \\frac 1 {4n}$$\nand \n$$\\left(x - \\frac 1{2} \\right)\n\\left(y - \\frac 1{2n}\\right) \\leq \\frac 1 {4n}$$\nhold. For $n\\geq 1$, define \n$$U_n \n= \n\\left\\{(x, y) \\,|\\, 0 \\leq x \\leq 1 , 0 \\leq y \\leq 1, \n2xy \n\\leq x + \\frac 1n y, \n2xy \n\\leq y + \\frac 1n x\\right\\}.$$\nNote that $U_n$ contains $\\frac 1n \\ensuremath{\\mathcal{S}}_{2,n}^*$ for any $n\\geq 1$ and any $*\\in \\{\\ensuremath{\\mathrm{cyc}},\\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$. \n\n\n\\begin{lemma}\n\\label{Lemma:Avoidance}\nFor any $0 < \\varepsilon < 1/2$, let $R_\\varepsilon$ denote the subset of $I^2$ defined by \n$$\nR_\\varepsilon = \n([\\varepsilon, 1] \\times [1/2 + \\varepsilon, 1] )\n\\cup \n([1/2 + \\varepsilon, 1] \\times [\\varepsilon, 1] ).\n$$\nThe region $R_\\varepsilon$ does not intersect with $U_n$ for large enough $n$. \n\\end{lemma}\n\n\\begin{proof}\nLet $N$ be a positive integer such that $1/2N + 1/2\\sqrt N < \\varepsilon$. \nLet $(x, y)$ be an element of $R_\\varepsilon$. \nIf $(x, y)$ lies in $[\\varepsilon, 1] \\times [1/2 + \\varepsilon, 1]$, then for any $n\\geq N$, \n\\begin{align*}\n\\left(x - \\frac 1{2n} \\right)\\left(y - \\frac 12 \\right) \n& \\geq \n\\varepsilon \n\\left(x - \\frac 1{2n} \\right)\\\\\n& \\geq \n\\varepsilon \n\\left(x - \\frac 1{2N} \\right)\\\\\n& > \n\\frac 1{2\\sqrt N} \\cdot \\frac 1{2\\sqrt N} \\\\\n& \\geq \\frac 1 {4n}.\n\\end{align*}\nIf $(x, y)$ lies in $[1/2 + \\varepsilon, 1] \\times [\\varepsilon, 1]$, then for any $n\\geq N$, \n\\begin{align*}\n\\left(x - \\frac 12 \\right)\\left(y - \\frac 1{2n}\\right) \n& \\geq \n\\varepsilon \n\\left(y - \\frac 1{2n} \\right)\\\\\n& \\geq \n\\varepsilon \n\\left(y - \\frac 1{2N} \\right)\\\\\n& > \n\\frac 1{2\\sqrt N} \\cdot \\frac 1{2\\sqrt N} \\\\\n& \\geq \\frac 1 {4n}.\n\\end{align*}\nSo no element of $R_\\varepsilon$ lies in $U_n$ for $n\\geq N$. \n\\end{proof}\n\n\nNote that for any $\\varepsilon >0$ and $*\\in \\{\\ensuremath{\\mathrm{cyc}},\\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$, the set $R_\\varepsilon$ avoids \n$\\frac 1n \\ensuremath{\\mathcal{S}}_{2,n}^*$ for large enough $n$. \n\n\n\\begin{questionIntro}\n\\label{Qn:XnComplementGetsLarger2}\nFor $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$ and for $0 < \\varepsilon < 1/2$, determine the asymptotic property of the sequence \n$$ |(([1/n, 2/n, \\cdots, n/n]^2\\setminus R_\\varepsilon)\\setminus \\ensuremath{\\mathcal{S}}_{2,n}^*)|.$$\n\\end{questionIntro}\n\nNote that \n$$|(([1/n, 2/n, \\cdots, n/n]^2\\setminus R_\\varepsilon)\\setminus \\ensuremath{\\mathcal{S}}_{2,n}^*)|\n\\geq \n\\left\\lfloor \\frac n2 \\right\\rfloor^2 -1$$\nsince $(1/n, i/n), (i/n, 1/n)$ belongs to $([1/n, 2/n, \\cdots, n/n]^2\\setminus R_\\varepsilon)\\setminus \\ensuremath{\\mathcal{S}}_{2,n}^*$ for $1\\leq i \\leq \\lfloor n/2\\rfloor$. Thus a more interesting question would be to study the asymptotic property of the sequence \n$$ |(([1/n, 2/n, \\cdots, n/n]^2\\setminus (R_\\varepsilon \\cup R'_n))\\setminus \\ensuremath{\\mathcal{S}}_{2,n}^*)|$$\nwhere $R'_n$ consists of those points of $[1/n, 2/n, \\cdots, n/n]^2$ which are avoided by $\\ensuremath{\\mathcal{S}}_{2,n}^*$ for ``obvious reasons''. Thus, $R'_n$ contains those points satisfying $xy < 1/n$. \n\n\\begin{questionIntro}\n\\label{Qn:Measure2}\nLet $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$. \n\\begin{enumerate}[(i)]\n\\item \nLet $0 \\leq a < b \\leq 1, 0 \\leq c < d \\leq 1$. Evaluate \n$$\n\\limsup \\frac{([a, b]\\times [c,d]) \\cap \\frac 1n \\ensuremath{\\mathcal{S}}_{2,n}^*}\n{|\\ensuremath{\\mathcal{S}}_{2,n}^*|}, \n\\liminf \\frac{([a, b]\\times [c,d]) \\cap \\frac 1n \\ensuremath{\\mathcal{S}}_{2,n}^*}\n{|\\ensuremath{\\mathcal{S}}_{2,n}^*|}.\n$$\nDoes the sequence \n$$\n\\frac{([a, b]\\times [c,d]) \\cap \\frac 1n \\ensuremath{\\mathcal{S}}_{2,n}^*}\n{|\\ensuremath{\\mathcal{S}}_{2,n}^*|}\n$$\nconverge? Otherwise, what are its subsequential limits?\n\n\\item \nDoes there exist a probability measure $\\mu$ on $I^2$ such that \n$$\n\\lim_{n\\to \\infty} \n\\frac{([a, b] \\times [c,d])\\cap \\frac 1n \\ensuremath{\\mathcal{S}}_{2,n}^*}\n{|\\ensuremath{\\mathcal{S}}_{2,n}^*|}\n= \\int_{I^2} \\chi_{[a, b]\\times [c,d]} d\\mu\n$$\nfor any $0 \\leq a < b \\leq 1, 0 \\leq c < d \\leq 1$, where $\\chi_A$ denotes the characteristic function of $A$ for $A\\subseteq I^2$?\n\\end{enumerate}\n\\end{questionIntro}\n\n\\begin{questionIntro}\n\\label{Qn:BddMinCompDetailed2}\nLet $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$. \n\\begin{enumerate}[(i)]\n\n\\item \nDetermine the open subsets of $I^2$ which do not intersect with $\\frac 1n \\ensuremath{\\mathcal{S}}_{2,n}^*$ for any/large enough/infinitely many $n$. \n\n\\item \nDetermine the open subsets of $I^2$ which have nonempty intersection with $\\frac 1n \\ensuremath{\\mathcal{S}}_{2,n}^*$ for any/large enough/infinitely many $n$.\n\n\\end{enumerate}\n\n\\end{questionIntro}\n\nNote that Lemma \\ref{Lemma:Avoidance} partially answers part (i) of the above question. \n\n\n\\begin{definition}\nA $k$-tuple $(A_1, \\cdots, A_k)$ of non-empty subsets of a group $G$ is said to be a co-minimal $k$-tuple if \n$$A_1 \\cdot A_2 \\cdot \\cdots \\cdot A_k = G$$\nand for any $1\\leq i \\leq k$ and for any $a_i \\in A_i$, \n$$A_1 \\cdot A_2 \\cdot \\cdots \\cdot(A_i \\setminus \\{a_i\\}) \\cdot \\cdots \\cdot A_k \\neq G.$$\n\\end{definition}\n\n\n\nNote that one can define an analogue of $\\ensuremath{\\mathcal{S}}_{2, n}^*$ as follows. \nLet $k\\geq 2$ be an integer. \nFor any finite group $G$, let \n$\\ensuremath{\\mathcal{S}}_k(G)$\ndenote the \nset of pairs of the form $(a_1, \\cdots, a_k)$ such that there is a co-minimal $k$-tuple $(A_1, \\cdots, A_k)$ in $G$ with $|A_i| = a_i$ for $1\\leq i \\leq k$. \nFor $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$ and for any positive integer $n$, consider the following subset of $\\{1, 2, \\cdots, n\\}^k$ defined as follows. \n\\begin{align*}\n\\ensuremath{\\mathcal{S}}_{k, n}^*\n& := \\bigcap_{G \\text{ is a $*$-group of order }n}\\ensuremath{\\mathcal{S}}_k(G).\n\\end{align*}\nThe set $\\ensuremath{\\mathcal{S}}_{k,n}^\\emptyset$ is also denoted by $\\ensuremath{\\mathcal{S}}_{k,n}$. \nBy Theorem \\ref{Thm:SmallWorks}, it follows that \n$$\\lim_{n\\to \\infty} |\\ensuremath{\\mathcal{S}}_{k,n}^*| = \\infty$$\nfor $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$. \nOne could ask the following questions, which are analogous to Questions \\ref{Qn:XnGetsLarger2}, \n\\ref{Qn:XnComplementGetsLarger2}, \n\\ref{Qn:Measure2}, \n\\ref{Qn:BddMinCompDetailed2}.\n\n\n\n\\begin{questionIntro}\nFor $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$, describe the asymptotic property of the sequence $|\\ensuremath{\\mathcal{S}}_{k,n}^*|$. \n\\end{questionIntro}\n\n\n\n\\begin{questionIntro}\nFor $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$ and for $0 < \\varepsilon < 1/2$, determine the asymptotic property of the sequence \n$$ |(([1/n, 2/n, \\cdots, n/n]^k\\setminus R_n)\\setminus \\ensuremath{\\mathcal{S}}_{k,n}^*)|.$$\nwhere $R_n$ consists of those points of $[1/n, 2/n, \\cdots, n/n]^k$ which are avoided by $\\ensuremath{\\mathcal{S}}_{k,n}^*$ for ``obvious reasons''. For instance, $R'_n$ contains those points satisfying $x_1\\cdots x_k < 1/n^{k-1}$. \n\\end{questionIntro} \n\n\\begin{questionIntro}\nLet $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$. \n\\begin{enumerate}[(i)]\n\\item \nLet $a_1, b_1, \\cdots, a_k, b_k$ be real numbers satisfying $0 \\leq a_i < b_i \\leq 1$ for all $1\\leq i \\leq k$. Evaluate \n$$\n\\limsup \\frac{([a_1, b_1] \\times \\cdots \\times [a_k,b_k]) \\cap \\frac 1n \\ensuremath{\\mathcal{S}}_{k,n}^*}\n{|\\ensuremath{\\mathcal{S}}_{k,n}^*|}, \n\\liminf \\frac{([a_1, b_1] \\times \\cdots \\times [a_k,b_k]) \\cap \\frac 1n \\ensuremath{\\mathcal{S}}_{k,n}^*}\n{|\\ensuremath{\\mathcal{S}}_{k,n}^*|}.\n$$\nDoes the sequence \n$$\n\\frac{([a_1, b_1] \\times \\cdots \\times [a_k,b_k]) \\cap \\frac 1n \\ensuremath{\\mathcal{S}}_{k,n}^*}\n{|\\ensuremath{\\mathcal{S}}_{k,n}^*|}\n$$\nconverge? Otherwise, what are its subsequential limits?\n\n\t\\item \nDoes there exist a probability measure $\\mu$ on $I^k$ such that \n$$\n\\lim_{n\\to \\infty} \n\\frac{([a_1, b_1] \\times \\cdots \\times [a_k,b_k])\\cap \\frac 1n \\ensuremath{\\mathcal{S}}_{k,n}^*}\n{|\\ensuremath{\\mathcal{S}}_{k,n}^*|}\n= \\int_{I^k} \\chi_{[a_1, b_1] \\times \\cdots \\times [a_k,b_k]} d\\mu\n$$\nfor any real numbers $a_1, b_1, \\cdots, a_k, b_k$ satisfying $0 \\leq a_i < b_i \\leq 1$ for all $1\\leq i \\leq k$, where $\\chi_A$ denotes the characteristic function of $A$ for $A\\subseteq I^k$?\n\\end{enumerate}\n\\end{questionIntro}\n\n\\begin{questionIntro}\nLet $*\\in \\{\\ensuremath{\\mathrm{cyc}}, \\ensuremath{\\mathrm{ab}}, \\ensuremath{\\mathrm{nil}}, \\ensuremath{\\mathrm{ssol}}, \\ensuremath{\\mathrm{sol}}, \\emptyset\\}$. \n\\begin{enumerate}[(i)]\n\n\\item \nDetermine the open subsets of $I^k$ which do not intersect with $\\frac 1n \\ensuremath{\\mathcal{S}}_{k,n}^*$ for any/large enough/infinitely many $n$. \n\n\\item \nDetermine the open subsets of $I^k$ which have nonempty intersection with $\\frac 1n \\ensuremath{\\mathcal{S}}_{k,n}^*$ for any/large enough/infinitely many $n$.\n\n\\end{enumerate}\n\n\\end{questionIntro}\n\n\n\\section{Acknowledgements}\nThe first author is supported by the ISF Grant no. 662/15. He wishes to thank the Department of Mathematics at the Technion where a part of the work was carried out. The second author would like to acknowledge the Initiation Grant from the Indian Institute of Science Education and Research Bhopal, and the INSPIRE Faculty Award from the Department of Science and Technology, Government of India.\n\n\\def\\cprime{$'$} \\def\\Dbar{\\leavevmode\\lower.6ex\\hbox to 0pt{\\hskip-.23ex\n \\accent\"16\\hss}D} \\def\\cfac#1{\\ifmmode\\setbox7\\hbox{$\\accent\"5E#1$}\\else\n \\setbox7\\hbox{\\accent\"5E#1}\\penalty 10000\\relax\\fi\\raise 1\\ht7\n \\hbox{\\lower1.15ex\\hbox to 1\\wd7{\\hss\\accent\"13\\hss}}\\penalty 10000\n \\hskip-1\\wd7\\penalty 10000\\box7}\n \\def\\cftil#1{\\ifmmode\\setbox7\\hbox{$\\accent\"5E#1$}\\else\n \\setbox7\\hbox{\\accent\"5E#1}\\penalty 10000\\relax\\fi\\raise 1\\ht7\n \\hbox{\\lower1.15ex\\hbox to 1\\wd7{\\hss\\accent\"7E\\hss}}\\penalty 10000\n \\hskip-1\\wd7\\penalty 10000\\box7}\n \\def\\polhk#1{\\setbox0=\\hbox{#1}{\\ooalign{\\hidewidth\n \\lower1.5ex\\hbox{`}\\hidewidth\\crcr\\unhbox0}}}\n\\providecommand{\\bysame}{\\leavevmode\\hbox to3em{\\hrulefill}\\thinspace}\n\\providecommand{\\MR}{\\relax\\ifhmode\\unskip\\space\\fi MR }\n\\providecommand{\\MRhref}[2]{%\n \\href{http://www.ams.org/mathscinet-getitem?mr=#1}{#2}\n}\n\\providecommand{\\href}[2]{#2}\n", "meta": {"timestamp": "2020-07-29T02:23:49", "yymm": "2007", "arxiv_id": "2007.14389", "language": "en", "url": "https://arxiv.org/abs/2007.14389", "abstract": "The notion of minimal complements was introduced by Nathanson in 2011 as a natural group-theoretic analogue of the metric concept of nets. Given two non-empty subsets $W,W'$ in a group $G$, the set $W'$ is said to be a complement to $W$ if $W\\cdot W'=G$ and it is minimal if no proper subset of $W'$ is a complement to $W$. The inverse problem asks which sets may or not occur as minimal complements. We show some new results on the inverse problem and investigate how the study of the inverse problem naturally gives rise to questions about the asymptotic behaviour of these sets, providing partial answers to some of them.", "subjects": "Combinatorics (math.CO)", "title": "Asymptotic behaviour of minimal complements", "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.9867771786365549, "lm_q2_score": 0.8633916011860785, "lm_q1q2_score": 0.8519751282768961}} {"url": "https://arxiv.org/abs/1504.07156", "title": "On consecutive sums in permutations", "abstract": "We study the number of values taken by the sums $\\sum_{i=u}^{v-1} a_i$, where $a_1,a_2,\\dots,a_n$ is a permutation of $1,2,\\dots,n$ and $1 \\leq u < v \\leq n+1$. In particular, we show that for a random choice of a permutation, with high probability there are $(\\frac{1+e^{-2}}{4} +o(1)) n^2$ such sums. This answers an old question of Erdős and Harzheim. We also obtain non-trivial bounds on the maximum possible number of distinct sums, ranging over all permutations of $1,2,\\dots,n$. We close with some questions concerning the minimal possible number of distinct sums.", "text": "\\section{Introduction}\n\n \nFor a sequence $a = (a_i)_{i=1}^n $ with $a_i \\in \\mathbb{Z}$, let $S(a)$ denote the set of all distinct sums $\\sum_{i=u}^v a_i$ with $1 \\leq u \\leq v \\leq n$. We shall mostly be interested in the size of $S(a)$, where $a_i$ is a permutation of the set $[n] := \\{1,2,\\dots,n\\}$. A trivial upper bound $\\abs{S(a)} \\leq \\binom {n+1}{2}$ follows from counting the number of choices of $u$ and $v$ (or, incidentally, from computing $\\max S(a)$). \n\nThe special case when $a_i = i$ for all $i$ was considered by Erd\\H{o}s and Harzheim \\cite{Erdos-1977-27}. It can be shown that for such $a$ one has $\\abs{S(a)} = o(n^2)$. Because of the elementary formula \n$$\\sum_{i=u}^v i = \\frac{(v-u-1)(v+u)}{2},$$\nwe have $S(a) \\subset [n] \\cdot [2n] \\subset [2n]\\cdot [2n]$, where we use the notation $A \\cdot B = \\{ab \\ : \\ a \\in A, b \\in B\\}$. Hence, obtaining good bounds on $\\abs{S(a)}$ is essentially equivalent to obtaining good bounds on $\\abs{[n]\\cdot[n]}$. The latter question, known as the \\emph{multiplication table problem}, has been extensively studied. \n\nThe first proof that $\\abs{[n] \\cdot [n] } = o(n^2)$ is due to Erd\\H{o}s \\cite{Erdos-1955}, with further improvements by the same author \\cite{Erdos-1960}. The exact asymptotics were only recently obtained by Ford \\cite{Ford-2008}, who shows that:\n\\begin{equation}\n\\abs{ [n] \\cdot [n] } = \\Theta\\bra{ n^2 (\\log n)^{-c} (\\log \\log n)^{-3/2}},\n\\end{equation}\n\t where $c=1-\\frac{1+\\log \\log 2}{\\log 2}$.\n\nThis special case lead Erd\\H{o}s to pose the following question \\cite{Erdos-1977-27}:\n\\begin{question}\\label{question:Erdos-Harheim}\n\tIs it true that for any $\\varepsilon$ there exists $n_0$ such that for any $n \\geq n_0$ and for any permutation $a_1,\\dots,a_n$ of $[n]$ we have $\\abs{S(a)} \\leq \\varepsilon n$?\n\\end{question}\n\nThe purpose of this note is to show that the answer to this question is an emphatic \\emph{``no''}. Without further ado, we give the simplest example we are aware of where the conjecture fails.\n\n\\begin{proposition}\\label{I:prop:cexple-basic}\n\tFor any $n$ one can find a permutation $a$ of $[n]$ such that $\\abs{S(a)} \\geq \\frac{1}{4}n^2. $\n\\end{proposition}\n\\begin{proof}\n\tLet $a$ be the permutation $1,n,2,n-1,3,n-2,\\dots$, so that for each odd $i$ one has $a_i + a_{i+1} = n+1$. \n\t\n\tLet $S_1 \\subset S(a)$ be the set of the sums $s = \\sum_{i=u}^v a_i$ with $u \\equiv v \\pmod{2}$.\tIf $v = u+2l$ then $s = (n+1)l + k$, where $k = a_v$ if $u,v$ are odd and $k = a_u$ otherwise. If $s$ has another representation $s = \\sum_{i=u'}^{v'} a_i = (n+1)l' + k'$ like before, then it follows that $l'=l,\\ k'=k$, hence $u,v \\equiv u',v' \\pmod{2}$ and finally $u'=u,\\ v'=v$. Thus, all $ \\ceil{ \\frac{n+1}{2} }\\cdot \\floor{ \\frac{n+1}{2} } \\geq \\frac{n^2}{2}$ sums in $S_1$ are distinct.\n\\end{proof}\n\n\nThe constant $\\frac{1}{4}$ in the above Proposition \\ref{I:prop:cexple-basic} can be improved slightly, using a randomised variant of the construction above.\n\n\\begin{proposition}\\label{I:prop:cexple-improved}\n\tFor any $n$ there exists a permutation $a$ of $[n]$ such that $\\abs{S(a)} \\geq (c_1+o(1)) n^2$, where $c_1 = \\frac{3}{2}-\\frac{2}{\\sqrt{e}} = 0.286\\dots$. \n\\end{proposition}\n\nIn the other direction, we have the following slight improvement over the trivial bound $S(a) \\leq (\\frac{1}{2}+o(1))n^2$. \n\\begin{proposition}\\label{I:prop:upper-bound}\n\tFor any permutation $a$ of $[n]$, we have $\\abs{S(a)} \\leq (c_2+o(1)) n^2$ where $c_2 = \\frac{1}{4} + \\frac{\\pi}{16} = 0.446\\dots$.\n\\end{proposition}\n\nThese bounds are proven in Sections \\ref{section:EXT-low} and \\ref{section:EXT-up} respectively. It would be very surprising if either of these results were optimal. We expect that $\\max_{a} S(a) = (c+o(1)) n^2$ for some constant $c$ with $c_1 < c < c_2$. It may be interesting to find the exact value of $c$.\n\nWhile the above examples resolve the original question of Erd\\H{o}s, they do not say what happens for a ``typical'' permutation. Our main result shows that the the answer to the Question \\ref{question:Erdos-Harheim} is still negative ``on average'' in a rather strong sense.\n\n\\begin{proposition}\\label{I:prop:exact}\n\tLet $a$ be a permutation of $[n]$ chosen uniformly at random.\n\tThen $\\EE(\\abs{S(a)}) = (c + o(1)) n^2$ where $c= \\frac{1+e^{-2}}{4} = 0.283\\dots$ as $n \\to \\infty$.\n\tMoreover,\n\t for any $\\delta >0$ we have $\\mathbb{P}\\bra{ \\abs{ \\frac{\\abs{S(a)}}{n^2} - c } > \\delta} \\to 0$ as $n \\to \\infty$.\n\\end{proposition}\n\nThe proof of this proposition is carried out in Section \\ref{section:EXC}, dealing with the expected value of $\\abs{S(a)}$, and Section \\ref{section:HM}, dealing with the second moment.\n\nTo close this section, we remark that a somewhat similar problem was considered by Hegyv{\\'a}ri in \\cite{Hegyvari}. Instead of a permutation of $[n]$, \\cite{Hegyvari} deals with a sequence $(a_i)_{i = 1}^k$ where $k \\leq n$ and $a_i \\in [n]$, and asks for which $k$ it is possible that \\emph{all} sums $\\sum_{i=u}^v a_i$ are distinct. This turns out to be possible for $k \\geq (\\frac{1}{3}-o(1)) n$, which gives an alternative proof of Proposition \\ref{I:prop:cexple-basic} with a slightly worse constant $\\frac{1}{18}$ in place of $\\frac{1}{4}$. Conversely, our Proposition \\ref{I:prop:exact} implies a non-trivial bound $k \\leq \\bra{\\sqrt{\\frac{\\pi}{8} + \\frac{1}{2}} + o(1)}n$.\n\n\\subsection*{Notation} \n\nWe use a variety of asymptotic notation. We write $X = O(Y)$, or $X \\ll Y$ if there exists a constant $C > 0$ such that $\\abs{X} \\leq C Y$. If the constant $C$ is allowed to depend on a parameter $M$, we occasionally write $X = O_M(Y)$. Likewise, we write $X = \\Omega(Y)$ if there exists a constant $c > 0$ such that $X > c Y$. Throughout the paper, we work in the regime $n \\to \\infty$. We write $X = o(Y)$ if $Y > 0$ for $n$ large enough an $\\frac XY \\to 0$ as $n \\to \\infty$, and similarly $X = \\omega(Y)$ if instead $\\frac XY \\to \\infty$. Finally, if $X = O(Y)$ and $X = \\Omega(Y)$ we write $X \\sim Y$ or $X = \\Theta(Y)$. Occasionally, we this notation with a different limit in mind, with the obvious modifications.\n\nIn particular, we freely use expressions such as $O(Y)$, $\\Omega(Y)$, $o(Y)$, $\\omega(Y)$, $\\Theta(Y)$ to denote unspecified functions with the asymptotic behaviour as just described.\n\n\n\n\\subsection*{Acknowledgements} The author wishes to thank Ben Green to pointing out this problem, and for much advice during the work on it. The author is also grateful to Sean Eberhard, Freddie Manners, Przemek Mazur, Rudi Mrazovi\\'{c} and Aled Walker for many fruitful discussions. Finally, the author thanks Jozsef Solymosi and Norbert Hegyv\\'{a}ri for helpful comments.\n\n \n\n\n\n\n\n\n\n\n \\section{Expected value of $\\abs{S(a)}$}\\label{section:EXC}\n\n\\newcommand{{\\mathrm{uni}}}{{\\mathrm{uni}}}\n\nIn this section we study the asymptotic behaviour of $\\EE\\abs{S(a)}$. Our main goal is to prove the first part of Proposition \\ref{I:prop:exact}, namely the asymptotic formula for $\\EE\\abs{S(a)}$. We will reuse much of our work here in the subsequent section to compute the second moment $\\EE\\abs{S(a)}^2$.\n\nThroughout the argument, $n$ is a large integer, and $a$ is a permutation selected uniformly at random from the set of the permutations of $[n]$. \n\n\n\nOne of our main tools is an inequality due to Hoeffding. We will mostly use the slightly less well-known variant of it, pertaining to random variables chosen from a finite set without replacement, such as the $a_i$.\n \n\\begin{theorem}[Hoeffding \\cite{Hoeffding-1963}]\n\tLet $X_i$ ($i=1,\\dots,m)$ be independent random variables with $\\a \\leq X_i \\leq \\b$ for some constants $\\a,\\b$ and $\\mu = \\sum_{i=1}^k \\EE X_i$. We then have for any $t > 0$\n\t\\begin{equation}\\label{EXP:eq:01}\n\t\\mathbb{P}\\bra{ \\abs{{\\sum_{i=1}^k} X_i - \\mu } \\geq t } \\leq 2\\exp\\bra{- \\frac{2 t^2}{k(\\b-\\a)^2}}.\n\t\\end{equation}\n\tMoreover, the same inequality \\eqref{EXP:eq:01} is satisfied if instead of independence we assume that $X_i$ are drawn without replacement from a finite (multi-)set. \n\\end{theorem}\n\nFor any $\\varepsilon > 0$, let $S_\\varepsilon(a)$ denote the set of all the sums $s$ of the form $\\sum_{i=u}^v a_i$ with $\\varepsilon n \\leq u \\leq n - \\frac{2s}{n} - \\varepsilon n$. It would be sufficient to work with a concrete choice such as $\\varepsilon = \\frac{1}{\\log n}$, but many other choices are equally valid.\n\nWe first reduce the problem of determining $\\EE\\abs{S(a)}$ to the dealing with a single (putative) sum in $S_\\varepsilon(a)$\n\n\\begin{proposition}\\label{EXC:prop:P-s-in-S}\n\tFix a choice of $\\varepsilon = \\varepsilon(n)$ with $\\varepsilon = o(1)$ and $\\varepsilon = n^{-o(1)}$ as $n \\to 0$. Let $s = \\sigma \\frac{n^2}{2}$ with $\\varepsilon < \\sigma < 1-\\varepsilon$. Then \n\t\\begin{equation}\\label{EXC:eq:P-s-in-S}\n\t\\mathbb{P}(s \\not\\in S_{\\varepsilon}(a)) = e^{-2+2\\sigma} + o(1).\n\t\\end{equation}\n\tHere, the error term is uniform with respect to $s$, but may depend on the implicit rates of convergence in the remaining usages of $o(1)$ notation.\n\\end{proposition}\n\n\\begin{proof}[Proof of Proposition \\ref{I:prop:exact}, assuming \\ref{EXC:prop:P-s-in-S}]\n\tTake, for instance, $\\varepsilon(n) := \\frac{1}{\\log n}$. If the bound in equation \\eqref{EXC:eq:P-s-in-S} holds for each $s$, then we have:\n\t\\begin{align*}\n\t\t\\EE\\abs{S_{\\varepsilon}(a)} &= \\sum_{\\varepsilon \\frac{n^2}{2} < s < (1 - \\varepsilon) \\frac{n^2}{2} } \\mathbb{P}(s \\in S_\\varepsilon(a)) + O(\\varepsilon n^2) \\\\\n\t\t&= \\frac{n^2}{2} \\int_{\\varepsilon}^{1-\\varepsilon} (1-e^{-2+2\\sigma}) d \\sigma + o( n^2) \n\t\t\\\\ &= n^2 \\bra{ \\frac{1+e^{-2}}{4} + o(1) }.\n\t\\end{align*}\n\tHoeffding's inequality implies that $\\abs{S(a) \\setminus S_{\\varepsilon}(a) } = o(n^2)$ with probability $1-o(1)$. Indeed, if $s = \\sum_{i=u}^v a_i \\in S(a) \\setminus S_\\varepsilon(a)$ then one of the three possibilities holds. Either $u < \\varepsilon n$ or $v > n-2\\varepsilon n$ or $v-u \\leq \\frac{2s}{n}-\\varepsilon n$. The first two cases account for $O(\\varepsilon n^2)$ elements of $S(a) \\setminus S_\\varepsilon(a)$. For the third case, note that by Hoeffding's inequality and the union bound, we have:\n$$\t\n\t\\mathbb{P}\\bra{(\\exists u,v) \\ : \\ \\sum_{i=u}^v a_i > \\frac{n}{2}(v-u) + \\varepsilon n^2 } \\ll\n\t{n^2 \\exp\\bra{-2\\varepsilon^2 n}} = o(n^2).\n$$\nOf course, both sets $S(a), S_\\varepsilon(a)$ have size $O(n^2)$. Hence,\n\t\\begin{equation}\n\t\t{\\EE\\abs{S(a)}} = (1+o(1)){\\EE\\abs{S_\\varepsilon(a)}} = (1+o(1)) \\frac{1+e^{-2}}{4} n^2. \n\t\t\\qedhere\n\t\\end{equation}\n\\end{proof}\n\nWe devote most of the remainder of this section to proving Proposition \\ref{EXC:prop:P-s-in-S}. \n\nGiven an index $u$, we define the set \n$$S_u(a) := \\left\\{ \\sum_{i=u}^v a_i \\ : \\ u \\leq v \\leq n\\right\\}.$$ \nClearly, $s \\in S_\\varepsilon(a)$ precisely when $s \\in S_{u}(a)$ for some $\\varepsilon n \\leq u \\leq (1 - \\sigma - \\varepsilon)n$ (where $\\sigma = \\frac{2s}{n}$).\n\nLet $N$ be a large even integer. By the inclusion/exclusion formula, we have:\n\\begin{equation}\\label{EXC:eq:39}\n\t\\mathbb{P}( s \\not \\in S_\\varepsilon(a)) \\leq \\sum_{\\abs{U} \\leq N} (-1)^{\\abs{U}} \\mathbb{P}\\bra{ \\bigwedge_{u \\in U} s \\in S_u(a)},\n\\end{equation}\nwhere the sum is taken over all sets $U \\subset [n]$ with $\\min U \\geq \\varepsilon n$ and $\\max U \\leq (1 - \\sigma - \\varepsilon)n$. An analogous formula holds for odd $N$, except that the inequality sign is reversed. Hence, Proposition \\ref{EXC:prop:P-s-in-S} will follow easily once a good estimate is found for each of the sums $\\sum_{\\abs{U} = M}\\mathbb{P}\\bra{ \\bigwedge_{u \\in U} s \\in S_u(a)}$. \n\nWe next prove a lemma which will allow us to eliminate a proportion of terms in such sums. The statement is slightly more general to accommodate latter applications to the second moment. We will only need it in the special case when $s_k$ take at most two distinct values, but this additional assumption does not simplify the argument.\n\n\\newcommand{\\mathcal{S}}{\\mathcal{S}}\n\n\\begin{lemma}\\label{EXC:lem:kill-bad-U}\n\tFix a choice of $\\varepsilon = \\varepsilon(n)$ with $\\varepsilon = o(1)$ and $\\varepsilon = n^{-o(1)}$ as $n \\to 0$. Let $M$ be an integer, for $1 \\leq k \\leq M$ let $s_k = \\sigma_k \\frac{n^2}{2}$ with $\\varepsilon < \\sigma_k < 1-\\varepsilon$ and let $u_k \\in [n]$ be distinct. Then\n\t\\begin{equation}\\label{EXC:eq:40a}\n\t\t\\mathbb{P}\\bra{ \\bigwedge_{k=1}^M s_k \\in S_{u_k}(a)} = O \\bra{ \\frac{ {\\frac{1}{\\varepsilon}\\log n} }{n}}^M ,\n\t\\end{equation}\n\twhere the implicit constant depends only on $M$.\n\\end{lemma}\n\\begin{proof}\n\n\n\tBy Hoeffding's inequality, there is a constant $C_M$ such that with probability $1-O\\bra{\\frac{1}{n^M}}$ any interval $[u,v]$ of length $C_M \\log n$ has $\\sum_{i=u}^v a_i \\geq n$. \n\t\n\tLet us now select $a$ by selecting $a_i$ in the order of decreasing $i$. For each $k$, the event $s_k \\in S_{u_k}(a)$ becomes determined at the time when $a_{u_k}$ is chosen. Either there are at most $C_M \\log n$ choices of $a_{u_k}$ leading $s_k \\in S_{u_k}(a)$, or there is an interval of length $C_M \\log n$ and sum less than $ n$. Since the number of choices of $a_{u_k}$ is at least $\\varepsilon n$ in each step, we obtain the sough bound by a standard inductive argument.\n\\end{proof}\n\n\tOur next goal is to obtain bounds on the probabilities in \\eqref{EXC:eq:39} for ``generic'' $U$. The general strategy at this point is to select $a_i$ in several ``stages''. For each $u \\in U$, we shall select $a_i$ with $u \\leq i < u+m$ from a suitably constructed set $A_u$ of size $\\abs{A_u} \\sim \\frac{1}{m} n$, where $m$ is an integer to be specified later. The rationale is that (assuming some genericity conditions on $A_u$) it should be relatively easy to understand the distribution of the sum $\\sum_{i=u}^{u+m-1} a_i$, and conditioning on the choice of $A_u$, the sums $\\sum_{i=u}^{u+m-1} a_i$ are independent. The main reason why this approach succeeds is contained in the following innocuous lemma.\n\nRecall that for two functions $f,g \\colon \\mathbb{Z} \\to \\mathbb{R}$ with finite support, we define their convolution $f \\ast g$ by $f \\ast g (x) = \\sum_{y \\in \\mathbb{Z}} f(x-y) g(y)$. We shall denote $\\norm{f}_{\\infty} := \\max_{x \\in \\mathbb{Z}} \\abs{f(x)}$. Also recall that we use $\\omega(1)$ do denote an arbitrary function tending to $\\infty$ as $n \\to \\infty$.\n\n\\begin{lemma}\\label{EXC:lem:f-A,2=uni}\n\tFix a choice of integer $m = m(n)$ with $m = n^{o(1)}$ and $m = \\omega(\\log n)$ as $n \\to \\infty$.\n\n\tLet $A \\subset [n]$ be a randomly chosen set with $\\abs{A} = \\floor{n/m^2}$. Then the bound\n\t\\begin{equation}\n\t\t\\norm{\\frac{1_{A}}{\\abs{A}} \\ast \\frac{1_{A}}{\\abs{A}} - \\frac{1_{[n]}}{n} \\ast \\frac{1_{[n]}}{n} }_{\\infty} \\leq \\frac{1}{n}\\cdot \\frac{m^5}{n^{1/2}}\n\t\\label{EXC:eq:41}\n\t\\end{equation}\n\tholds with probability $1-n^{-\\omega(1)}$.\n\\end{lemma}\n\nAgain, we could work with a specific choice of $m$, such as $m \\sim n^{\\frac{1}{\\log \\log n}}$ or (with minor modifications) $m \\sim n^{\\frac{1}{100}}$. The slightly unnatural looking condition $\\abs{A} = \\floor{n/m^2}$ will simplify notation later on. \n\n\\begin{proof}\n\tTake arbitrary $x \\in [n]$. It will suffice to show that the inequality\n\t$$\n\t\t\\abs{ 1_{A} * 1_{A}(x) - \\frac{1}{m^4} 1_{[n]} * 1_{[n]}(x) } \\leq \\sqrt{n} m\n\t$$\n\tholds with probability $1-n^{-\\omega(1)}$, and apply the union bound. We may assume without loss of generality that $x \\leq n/2$. For ease of notation, we additionally assume that $x$ is odd. In this case, $1_{[n]} * 1_{[n]}(x) = x-1$.\n\t\n\tDefine $A_1 = A\\cap [1,x/2)$ and $A_2 = (x/2,x]$ so that we have $1_A * 1_A(x) = 2\\abs{A_1 \\cap (x-A_2)}$. An application of Hoeffding's inequality shows that the bounds:\n\t$$\n\t\t\\abs{ \\abs{A_1} - \\frac{x-1}{2m} } \\leq \\sqrt{x m}, \\qquad\n\t\t\\abs{ \\abs{A_2} - \\frac{x-1}{2m} } \\leq \\sqrt{x m} \t\t\n\t$$\n\thold with probability $\\geq 1 - 2 \\exp(-2m) = 1-n^{-\\omega(1)}$. Similarly, the bound\n\t$$\n\t\t\\abs{\n\t\t\t\\abs{ A_1 \\cap (x-A_2)} - \\frac{\\abs{A_1}\\abs{A_2}}{x/2}\n\t\t} \\leq \\sqrt{ x m }\n\t$$\n\tholds with probability $\\geq 1 - 2 \\exp(-2m) = 1-n^{-\\omega(1)}$. Hence, by the union bound, we have with probability $1-n^{-\\omega(1)}$ that\n\t\\begin{align*}\n\t\t\\abs{ 1_{A} * 1_{A}(x) - \\frac{1}{m^4} 1_{[n]} * 1_{[n]}(x) } \n\t\t &= \\abs{2 \\abs{A_1 \\cap (x-A_2)} - \\frac{x-1}{m^4} } \n\t\t\\\\& \\leq \\abs{ \\frac{4 \\abs{A_1} \\abs{A_2}}{x} - \\frac{x-1}{m^4} } + \\sqrt{n m} \n\t\t\\\\& \\leq 2 \\abs{ \\abs{A_1} - \\frac{x}{2m} } + 2\\abs{ \\abs{A_1} - \\frac{x}{2m} } + \\sqrt{nm}\n\t\t\\\\& \\leq 5 \\sqrt{nm} < m \\sqrt{n},\t\t \n\t\\end{align*}\n\tas required.\n\\end{proof}\n\nAs the reader will have noticed, we make no serious attempt to optimize the dependence on $m$. The exponent $5$ does not play a role; we merely need it to be a constant\n\n\\newcommand{{r}}{{r}}\nFor a set $A \\subset [n]$ and integer $k$, let us agree to denote by ${r}_{A,k}$ the distribution of $Y = X_1 + X_2 + \\dots\nX_k$, where $X_i$ are chosen uniformly without replacement, that is \n$${r}_{A,k}(x) = \\mathbb{P}(Y = x) = \\mathbb{P}\\bra{\\sum_{i=1}^k X_i = x}.$$ Also, let ${r}_{{\\mathrm{uni}},k}$ denote the distribution of $V = U_1 + U_2 + \\dots + U_k$ where $U_i \\in [n]$ are chosen uniformly and independently, hence in particular ${r}_{{\\mathrm{uni}},k} = {r}_{{\\mathrm{uni}},1} * \\cdots * {r}_{{\\mathrm{uni}},1}$.\n\nFor instance, if $\\abs{A} = \\floor{n/m^2}$ then ${r}_{A,1} = \\frac{1_A}{\\abs{A} }$, and since we have the bound $\\norm{ {r}_{A,1} \\ast {r}_{A,1} - {r}_{A,2}}_\\infty = O\\bra{\\frac{m^4}{n^2}}$, the above lemma says that with overwhelming probability $\\norm{ {r}_{A,2} - {r}_{{\\mathrm{uni}},2} }_{\\infty} \\ll \\frac{1}{n} \\frac{m^5}{n^{1/2}}$. More generally, we have the following.\n\n\\begin{corollary}\\label{EXC:lem:f-A,m=uni}\n\tFix $m$ with $m = n^{o(1)}$ and $m = \\omega(\\log n)$, even. Let $A \\subset [n]$ be a randomly chosen set with $\\abs{A} = m\\floor{n/m^2}$. Then the bound\n\t\\begin{equation}\n\t\t\\norm{ {r}_{A,m} - {r}_{{\\mathrm{uni}},m} }_{\\infty} \\leq \\frac{1}{n} \\cdot \\frac{m^6}{n^{1/2}}\n\t\t\\label{EXC:eq:42}\n\t\\end{equation}\n\tholds with probability $1-n^{-\\omega(1)}$.\n\\end{corollary}\n\\begin{proof}\n\t\n\tWe consider a two-stage random process. First, we select disjoint sets $A_1,\\dots,A_m \\subset A$ with $\\abs{A_i} = \\floor{n/m^2}$ uniformly at random. Second, we select $b_{2j-1},b_{2j} \\in A_j$ uniformly at random without replacement. It follows that\n\t\\begin{equation}\n\t{r}_{A,m} = \\EEbig_{A_1,\\dots,A_{m/2}} {r}_{A_1,2} \\ast {r}_{A_2,2} \\ast \\dots {r}_{A_{m/2},2}.\n\t\\end{equation}\n\twhere the expectation is taken over all partitions of $A$.\n\t\t\n\tIt is a standard fact that if $f$ is a probability distribution then for any bounded $g$ one has $\\norm{ f \\ast g}_{\\infty} \\leq \\norm{g}_\\infty$. Hence, we obtain by a simple inductive argument\n\t\\begin{equation}\n\t\t\\norm{ {r}_{A_1,2} \\ast {r}_{A_2,2} \\ast \\dots \\ast {r}_{A_{m/2},2} - {r}_{{\\mathrm{uni}},m} }_\\infty \\leq \\sum_{j=1}^{m/2} \\norm{{r}_{A_j,2} - {r}_{{\\mathrm{uni}},2}}_\\infty.\n\t\\end{equation}\n\t\t\n\tNote that if $A$ is selected uniformly at random (subject to cardinality), then so is each of the sets $A_j$.\nUsing Lemma \\ref{EXC:lem:f-A,2=uni} and the union bound, with probability $1\n\t-n^{-\\omega(1)}$, the bound \\eqref{EXC:eq:41} holds for each $A_j$. Hence, with probability $1-n^{-\\omega(1)}$, also the bound \\eqref{EXC:eq:42} holds.\n\n\\end{proof}\n\n\n\tWe shall exploit the fact that the sums $\\sum_{i=u}^v a_i$ tend to have roughly the correct magnitude if $v-u$ is larger than a power of $\\log n$. To make this idea precise, we introduce the following piece of notation. Let $I \\subset [n]$ be a (connected) interval, and let $b = (b_i)_{i \\in I}$ be a $[n]$-valued sequence indexed by $I$. Then $\\mathscr{R}(b)$ is the event that for any non-empty interval $J \\subset I$ one has \n\t\\begin{equation} \n\t\t\\abs{ \\sum_{j \\in J}b_i - \\abs{J} \\frac{n}{2} } \\leq \\sqrt{\\abs{J}} n \\log n.\n\t\t\\label{EXC:eq:R} \n\t\\end{equation}\nIt is an easy consequence of the Hoeffding inequality that for any $I$ we have\n$$\\mathbb{P}\\brabig{\\mathscr{R}(a|_{I})} = 1 - n^{-\\omega(1)},$$ where we use the notation $a|_I$ to denote the restriction of $a$ to $I$. (In fact, we could replace $\\log n$ in \\eqref{EXC:eq:R} above with any function $= \\omega(\\sqrt{\\log n})$). Slightly more generally, we have the following result.\n\n\\begin{lemma}\\label{EXC:lem:P-R-uniform} Fix a choice of even integer $m$ with $m = n^{o(1)}$ and $m = \\omega(\\log n)$. \n\tLet $A \\subset [n]$ be a randomly chosen set with $\\abs{A} = m \\floor{n/m^2}$. Let $b = (b_1,\\dots,b_m)$ be chosen uniformly at random without replacement from $A$. Then there is a function $h$ with $h(n) \\to \\infty$ as $n \\to \\infty$ such that\n\\begin{equation}\\label{EXC:cond:R1}\n\t\t\\mathbb{P}_{A}\\bra{ \\mathbb{P}_{b_i \\in A}( \\neg \\mathscr{R}(b)) \\geq n^{-h(n)} } \\leq n^{-h(n)}.\n\\end{equation}\n\\end{lemma}\n\\begin{proof}\n\tBy Hoeffding's inequality, for each interval $J$ with $\\abs{J} = l$ one has:\n\t$$\n\t\\mathbb{P}_{b_i \\in [n] }\n\t\\bra{ \\abs{ \\sum_{j \\in J} b_j - l \\frac{n}{2} } \\geq \\sqrt{l} n \\log n}\n\t\\leq 2\\exp\\bra{ - 2 \\log^2 n} \n\t= n^{-\\omega(1)},\n\t$$\n\twhere $b_i$ are chosen from $[n]$ without replacement, uniformly at random. Hence, after an application of the union bound over $O(n^2)$ choices of $J$, we also know that $$\\EEbig_{A}\\brabig{ \\mathbb{P}_{b_i \\in A}( \\neg \\mathscr{R}(b)) } = \\mathbb{P}_{b_i \\in [n]} \\bra{ \\neg \\mathscr{R}(b) } = n^{-\\omega(1)}.$$ The claim now follows from the Chebyshev bound.\n\\end{proof}\n\nThe above lemma can be restated by saying that for a randomly selected $A$, the bound\n\\begin{equation}\\label{EXC:cond:R2}\n\t\\mathbb{P}_{b_i \\in A}( \\mathscr{R}(b)) \\geq 1 - n^{-\\omega(1)}\n\\end{equation}\nholds with probability $1 - n^{-\\omega(1)}$.\n\nWe are now ready to estimate the probabilities in \\eqref{EXC:eq:39}. The only caveat is that we shall need to assume a certain genericity condition on $U$. Namely, we shall require that the set $U$ be well-separated:\n\\begin{equation}\\label{EXC:cond:U-reg}\n\t(\\forall u, v \\in U) \\ u = v \\vee \\abs{u-v} > m, \n\\end{equation}\nwhere $m$ is a parameter yet to be specified. Under this condition, we can now prove the needed bound. Again, we prove slightly more then necessary at the moment, since in this section we will only apply the following proposition with a constant sequence $s_k = s$. \n\n\\begin{proposition}\\label{EXC:prop:P-wedge-s-in-S}\n\tFix a choice of $\\varepsilon = \\varepsilon(n)$ and even $m = m(n)$ such that $\\varepsilon = o(1)$, $m = n^{o(1)}$ and $m = \\bra{ \\frac{1}{\\varepsilon} \\log n}^{\\omega(1)}$ as $n \\to 0$. \tLet $M$ be an integer. For $1 \\leq k \\leq M$ let $s_k = \\sigma_k \\frac{n^2}{2}$ with $\\varepsilon < \\sigma_k < 1-\\varepsilon$ and let $u_k \\in [n]$ be distinct, and such that $U := \\{u_k\\}_{k=1}^M$ satisfies the well-separatedness condition \\eqref{EXC:cond:U-reg}. Then \n\\begin{equation}\n\t\\mathbb{P}\\bra{ \\bigwedge_{u \\in U}^M \\bra{ s_k \\in S_{u_k}(a) } } = \\bra{\\frac{2}{n}}^M (1+ o(1)),\n\\end{equation}\nwhere the implicit error term depends only on $M$.\n\\end{proposition}\n\n\\begin{proof}\nDefine intervals $I_k = [u_k,u_k+ m)$ for $k \\in [M]$, and $I_{0} = [1, M m\\floor{n/m^2} ]$, and finally $I_{\\infty} = [n] \\setminus \\bigcup_{k=0}^M I_k$. Because of the well-separetedness \\eqref{EXC:cond:U-reg}, these are disjoint.\t\n\nWe can think of $a$ as being selected in two stages. Firstly, for each $1 \\leq k \\leq M$ we select a set $A_k$ of size $m\\floor{n/m^2}$ uniformly at random, and put $A_{\\infty} = [n] \\setminus \\bigcup A_k$. Next, for each $k \\in [M] \\cup \\{\\infty\\}$ for $i \\in I_{k}$, we select $a_i$ uniformly at random from $A_{k}$ without replacement. Finally, for $i \\in I_0$, we select $a_i$ from $\\bigcup_{k \\in [M]} A_k$ (except for values which have already been used) uniformly at random. It is not hard to see that this procedure produces the uniform distribution of $a$.\n\nBy Lemmas \\ref{EXC:lem:f-A,m=uni} and \\ref{EXC:lem:P-R-uniform}, for each $k \\in [M]$ the regularity\nconditions \\eqref{EXC:eq:42} and \\eqref{EXC:cond:R2} hold for $A_k$ with probability $1-n^{-\\omega(1)}$. Hence, it will suffice to prove the bound \n\\begin{equation}\\label{EXC:eq:49}\n\t\\mathbb{P}\\bra{ \\bigwedge_{k=1}^M (s_k \\in S_{u_k}(a)) \\middle| a_{i} \\in A_k \\text{ for all } k,\\ i \\in I_k } = \\bra{\\frac{2}{n}}^M \\bra{1+ o(1)}\n\\end{equation}\nwhere each of the sets $A_k$ satisfies \\eqref{EXC:eq:42} and \\eqref{EXC:cond:R2} (here, the error term does not depend on $s_k$).\n\nBecause for each fixed $k$ we have $\\mathbb{P}(\\neg \\mathscr{R}( a|_{I_k} ) | \\ a_{i} \\in A_k \\text{ for } i \\in I_k) = n^{-\\omega(1)}$ by \\eqref{EXC:cond:R2}, we may replace each event $s_k \\in S_{u_k}(a)$ in \\eqref{EXC:eq:49} with the event $s_k \\in S_{u_k}(a) \\wedge \\mathscr{R}(a|_{I_k})$ at the cost of introducing a negligible error term of the order $O(n^{-\\omega(1)})$. \n\nThe claim now follows by a standard inductive argument. It suffices to prove the following.\n\n\\begin{lemma*}\n\tLet $k \\in [M]$. For $k < l \\leq M$ and $l = \\infty$, let $a^*_{I_l} = (a^*_j)_{j \\in I_l}$ be $[n]$-valued sequences such that the well-distribution condition $\\mathscr{R}(a^*_{I_l})$ defined in \\eqref{EXC:eq:R} holds. Let $A \\subset [n]$ be a set with $\\abs{A} = m \\floor{n/m^2}$ satisfying the well-separatedness condition \\eqref{EXC:eq:42} and the well-distribution \\eqref{EXC:cond:R2}. Then\n\t\\begin{equation}\\label{EXC:eq:50}\n\t\t\\mathbb{P} \\left( s_k \\in S_{u_k}(a) \\wedge \\mathscr{R}(a_{I_k}) \n\t\t\\middle|\n\t\t \\begin{matrix}\n\t\ta_{i} \\in A \\text{ for } i \\in I_k \\text{ and } \\\\\n\t\t a_{I_{l}} = a_{I_l}^* \\text{ for } l = k+1,\\dots,M, \\infty\n\t\t \\end{matrix} \n\t\t \\right) = \\frac{2+o(1)}{n},\n\t\\end{equation}\n\t\\end{lemma*}\nwhere the error term does not depend on $s_k$. \n\\begin{proof}\n\tLet $b_1,\\dots,b_m$ be chosen uniformly at random without replacement from $A$. Then the probability in \\eqref{EXC:eq:50} can be rewritten as\n\t\\begin{equation}\\label{EXC:eq:51}\n\t\t\\mathbb{P}_{b_i \\in A} \\bra{ s_k - \\sum_{i=1}^m b_i \\in S_{u_k+m}(a^*) \\wedge \\mathscr{R}(b)}.\n\t\\end{equation}\n\n\tBecause of \\eqref{EXC:cond:R2}, we have $\\mathbb{P}_b(\\mathscr{R}(b)) = 1 - n^{-\\omega(1)}$. \t\n\tPutting $$R := s_k-S_{u_k+m}(a^*),$$ it will hence suffice to estimate $\\mathbb{P}_{b} \\bra{ \\sum_{i=1}^m b_i \\in R}$.\n\t\n\tThe assumption \\eqref{EXC:cond:R2} for $A_{k+1},A_{k+2},\\dots,A_M$ implies that for interval any $J \\subset \\bigcup_{l =k+1}^M I_l \\cup I_\\infty$ with $\\abs{J} \\gg \\frac{\\log^2n}{\\varepsilon^4}$ one has:\n\t\\begin{equation}\n\t\t\\sum_{i \\in J} a^*_j = \\frac{n}{2} \\abs{J}(1+O(\\varepsilon^2)).\n\t\\end{equation}\n\tNote in particular that \n\t$$\\max S_{u_k+m}(a^*) = \\frac{n}{2}(n-u_k-m)(1+O(\\varepsilon^2)) \\geq \\frac{1}{2}n^2(\\sigma_k+\\varepsilon - O(\\varepsilon^2)).$$ Thus, we have somewhat crude bounds $\\min R < 0$ and $ \\max R > \\frac{1}{2} \\varepsilon n^2 > mn$. (Recall that $\\varepsilon > \\frac{1}{\\sqrt{n}} > \\frac{m}{n}$.)\n\t\n\tIf $K \\subset [\\min R, \\max R]$ is an arbitrary interval of length at least $n \\frac{\\log^2n}{\\varepsilon^4}$ then \n\t\\begin{equation}\n\t\\abs{K \\cap R} \\geq \\frac{n}{2} \\abs{K} (1+ O(\\varepsilon^2)).\n\t\\label{EXC:eq:45}\t\n\t\\end{equation}\n\tIndeed, if $J$ denotes the set of $v$ such that $\\sum_{i=u_k+m}^v a_i^* \\in K$ then\n\t$$ \\abs{K} + O(n) = \\sum_{i \\in J} a_{i}^* = \\frac{n}{2} \\abs{K\\cap R}(1+O(\\varepsilon^2)),$$\n\twhich easily implies the claim.\n\n\nReturning to the probability we wish to estimate, we clearly have \n\\begin{equation}\\label{EXC:eq:53}\n\\mathbb{P}_{b}\\bra{ \\sum_{i=1}^m b_i \\in R} = \\sum_{x } 1_R(x) {r}_{A,m}(x),\n\\end{equation}\nwhere ${r}_{A,m}(x)$ is the distribution of $\\sum_{i \\in I_k} b_i$. The sum is formally taken over $\\mathbb{Z}$ but ${r}_{A,m}$ is identically $0$ outside the interval $[m,mn]$.\n\nIt follows from Corollary \\ref{EXC:lem:f-A,m=uni} that\n\\begin{align}\\label{EXC:eq:55}\n\t\\abs{\n\t\\sum_{x } 1_R(x) \\bra{ {r}_{A,m}(x) - {r}_{{\\mathrm{uni}},m}(x) }\n\t} \\leq mn \\norm{{r}_{A,m} - {r}_{{\\mathrm{uni}},m}}_\\infty \n\t= O\\bra{ \\frac{m^6}{n^{1/2}} } .\n\\end{align}\n\nWe note some basic properties of ${r}_{{\\mathrm{uni}},m}$. Firstly, ${r}_{{\\mathrm{uni}},m}$ takes its maximal value at a precisely one point, with respect to which it is symmetric. Indeed, this property holds for ${r}_{{\\mathrm{uni}},2}$ and is preserved under convolution with ${r}_{{\\mathrm{uni}},2}$.\n\nSecondly, for any integer $x$ we claim that ${r}_{{\\mathrm{uni}},m}(x) \\ll \\frac{1}{m^{1/2}n}$. For this purpose we introduce the exponential sum $\\phi(t) := \\frac{1}{n} \\sum_{k=1}^n e( k t) = \\frac{e(t)}{n} \\frac{e(tn)-1}{e(t)-1}$ where $e(t) := e^{2 \\pi i t}$, so that for each $x$ we have $${r}_{{\\mathrm{uni}},m}(x) = \\int_{0}^1 e(-xt) \\phi(t)^m dt \\leq \\int_0^1 \\abs{\\phi(t)}^m dt.$$\n\nWe have the trivial bound $\\abs{\\phi(t)} \\leq 1$ for each $t$. Slightly less trivially, $\\abs{\\phi(t)} = \\abs{ \\frac{\\sin n t\\pi}{n \\sin t \\pi}}$. Hence, for $\\frac{1}{2n} \\leq t \\leq 1 - \\frac{1}{2n}$ we find $\\abs{\\phi(t)} \\leq \\frac{1}{\\pi/2 - o(1)} < \\frac{2}{3}$ if $n$ is large enough, and hence \n$$\\int_{1/2n}^{1-1/2n} \\abs{\\phi(t)}^m dt < \\bra{\\frac{2}{3}}^m \\ll \\frac{1}{m^{1/2}n}.$$ \nFor $1 < t < \\frac{1}{2n}$ we have $\\abs{\\phi(t)} = \\abs{ \\frac{\\sin \\pi n t}{ n\\pi t(1 + O(1/n^2))}}$ and hence $\\abs{\\phi(t)}^m = \\abs{ \\frac{\\sin \\pi n t}{n \\pi t}}^me^{O(m/n^2)}$. One can find a universal constant $c$ such that $\\abs{ \\frac{\\sin u}{u}} \\leq e^{-c u^2}$ for $0 < u < \\pi/2$, and thus $\\abs{ \\phi(t) }^m \\leq e^{-c t^2 n^2 m}$ (possibly with a different constant). Consequently, \n$$\\int_{0}^{1/2n} \\abs{ \\phi(t) }^m dt < \\int_{0}^\\infty e^{-c t^2 n^2 m} dt < \\frac{C}{m^{1/2}n}$$ \nfor an absolute constant $C>0$. Applying the same argument to $1-\\frac{1}{2n} < t < 1$, we conclude that ${r}_{{\\mathrm{uni}},m}(x) \\leq \\frac{C}{m^{1/2}n}$, where $C$ is absolute.\n\n\n\nWe may now estimate $\\sum_{x } 1_R(x) {r}_{{\\mathrm{uni}},m}(x)$. Let us separate the interval $[0, m n]$ into pieces $\\dots,K_{-2},K_{-1},K_0,K_{1},K_{2},\\dots$, arranged in this order, disjoint, each of length $(1+o(1)) n \\frac{\\log^2n}{\\varepsilon^4}$, except for $K_0 := \\{\\frac{m(n+1)}{2}\\}$ of length $1$. Consider the function $g_{-}$ defined by $g_{-}(x) = \\frac{2}{n}(1-C \\varepsilon^2)$ for $x \\in {K_i}$ with $\\abs{i} > 1$ and $g_{-}(x) = 0$ for $x \\in K_{1},K_0,K_{-1}$. Similarly, consider $g_{+}$ defined by $g_{+}(x) = \\frac{2}{n}(1+C \\varepsilon^2)$ for $x \\in {K_i}$ with $\\abs{i} \\neq 0$ and $g_{+}\\bra{\\frac{m(n+1)}{2}} = 2 \\frac{\\log^2n}{\\varepsilon^4}$.\n\nThe density assumption \\eqref{EXC:eq:45} and monotonicity of ${r}_{{\\mathrm{uni}},m}$ imply that \n\\begin{equation}\n\t\\sum_{x} {r}_{{\\mathrm{uni}},m}(x) g_{-}(x) \\leq \\sum_{x} {r}_{{\\mathrm{uni}},m}(x) 1_{R}(x) \\leq \\sum_{x} {r}_{{\\mathrm{uni}},m}(x) g_{+}(x),\n\\end{equation}\nprovided that $C$ is sufficiently large. Moreover, we have \n$$\n\\sum_{\\substack{x \\in K_i, \\abs{i} \\leq 1 }}\n\t{r}_{{\\mathrm{uni}},m}(x) \\ll \\frac{\\log^{2} n}{\\varepsilon^2 m^{1/2}} \\ll \\varepsilon^2,\n$$\nand inserting this bound into the above inequality we find\n\\begin{equation}\n\t\\sum_{x} {r}_{{\\mathrm{uni}},m}(x) g_{-}(x) \\geq \\frac{2}{n}\\bra{1-C \\varepsilon^2}\\bra{1 - O(\\varepsilon^2)\n\t} = \\frac{2}{n}\\bra{1-O(\\varepsilon^2)}.\n\\end{equation}\nBy a similar argument, we have\n\\begin{equation}\n\t\\sum_{x} {r}_{{\\mathrm{uni}},m}(x) g_{+}(x) \\leq \\frac{2}{n}(1+C \\varepsilon^2) + {r}_{{\\mathrm{uni}},m}\\bra{\\frac{m}{2n}} \\cdot 2\\frac{\\log^2n}{\\varepsilon^4} = \\frac{2}{n}(1+O(\\varepsilon^2)).\n\\end{equation}\nHence, we obtain\n\\begin{equation}\n\\sum_{x} {r}_{{\\mathrm{uni}},m}(x) 1_{R}(x) = \\frac{2}{n}(1+O(\\varepsilon^2)).\n\\end{equation}\nCombined with \\eqref{EXC:eq:53} and \\eqref{EXC:eq:55}, this finishes the proof of the inductive step, and hence the entire proof.\n\n\n\\renewcommand{\\qedsymbol}{}\\end{proof}\n\n\\end{proof}\n\nWe are now in position to finish the proof of Proposition \\ref{EXC:prop:P-s-in-S} (and hence also the first part of Proposition \\ref{I:prop:exact}). \n\n\\begin{proof}[Proof of Proposition \\ref{EXC:prop:P-s-in-S}]\n\tLet $N$ be a large even integer. From \\eqref{EXC:eq:39} we have\n\t\\begin{equation}\n\t\t\\mathbb{P}(s \\not\\in S_\\varepsilon(a)) \\leq \\sum_{M \\leq N} (-1)^M \\sum_{\\abs{U} = M} \\mathbb{P}\\bra{ \\bigwedge_{u \\in U} s \\in S_u(a) }. \n\t\\end{equation}\n\tThe inner sum runs over $ \\binom{ (1-\\sigma + o(1))n}{M} = (1+o(1)) \\frac{(1-\\sigma)^M}{M!}n^M$ summands. The proportion of summands for which \\eqref{EXC:cond:U-reg} fails is $O(\\frac{m}{n})$, and by Lemma \\ref{EXC:lem:kill-bad-U} each of them contributes at most $O\\bra{\\frac{\\bra{ \\frac{1}{\\varepsilon}\\log n}^M}{n^M}}$. Hence, their total contribution is $O\\bra{ \\frac{m \\bra{ \\frac{1}{\\varepsilon}\\log n}^M}{n}} = o(1)$.\n\t\n\tAll remaining summands are by Proposition \\ref{EXC:prop:P-wedge-s-in-S} equal to $\\bra{\\frac{2}{n}}^M(1+o(1))$. Thus, the inner sum can be estimated as\n\\begin{align*}\n\t\\sum_{\\abs{U} = M} \\mathbb{P}\\bra{ \\bigwedge_{u \\in U} s \\in S_u(a) } &= \\frac{(1-\\sigma)^M}{M!}n^M \\cdot \\bra{\\frac{2}{n}}^M(1+o(1)) + o(1)\n\t\\\\&= \\frac{ (2-2\\sigma)^M }{M!} + o(1).\n\\end{align*}\n\n\tThis leads to the upper bound\n\\begin{align*}\n\\mathbb{P}(s \\not \\in S_\\varepsilon(a)) &\\leq \\sum_{M=0}^N (-1)^M \\frac{ (2-2\\sigma )^M }{M!} + o(1) \n\\leq e^{-2+2\\sigma} + \\frac{C}{N!} + o(1),\n\\end{align*}\t\t\n\t where the constant $C$ is absolute. Letting $N \\to \\infty$ slowly with $n$ we conclude that $\\mathbb{P}(s \\in S_\\varepsilon(a)) \\leq e^{-2+2\\sigma} + o(1)$.\n\t \n\t Running the same argument with $N$ odd we conclude the reversed bound $\\mathbb{P}(s \\in S_\\varepsilon(a)) \\geq e^{-2+2\\sigma} + o(1)$. Hence $\\mathbb{P}(s \\not\\in S_\\varepsilon(a)) = e^{-2+2\\sigma}(1+o(1))$, where the error term is uniform with respect to $s$.\n\t \n\\end{proof}\n\n\\section{Higher moments of $\\abs{S(a)}$}\\label{section:HM}\n\nIn this section, we (asymptotically) compute the second moment $\\EE\\abs{S(a)}^2$, where $a$ is a randomly selected permutation of $[n]$. Thus, we prove the concentration around the mean for $\\abs{S(a)}$.\n\nBy a standard application of the second moment method, we have for any $\\delta > 0$ that \n$$\n\t\\mathbb{P}\\bra{ \\abs{\\abs{S(a)}^{\\phantom{|}\\!\\!\\!} - \\EE(\\abs{S(a)}} > \\delta n^2 } \\leq \\frac{\\EE{\\abs{S(a)}^2} - \\EE(\\abs{S(a)})^2}{\\delta^2 n^4}.\n$$\n\nHence, the concentration around the mean in Proposition \\ref{I:prop:exact} will follow directly from the following result.\n\n\\begin{proposition}\\label{HM:prop:E(S(a)^p)}\n\tThe second moment of $\\abs{S(a)}$ is given by $$\\EE \\abs{S(a)}^2 = (1+o(1))c^2 n^{4}$$ as $n \\to \\infty$, where $c = \\frac{1+e^{-2}}{4}$. \n\\end{proposition}\n\nBecause $\\abs{S(a)}/n^2$ is bounded, the concentration around the mean, as stated in Proposition \\ref{I:prop:exact}, implies the asymptotic formula for the higher moments $\\EE \\abs{S(a)}^p$ for all $p\\geq 1$, namely $\\EE \\abs{S(a)}^p = (1+o(1)) c^p n^{2p}$. Similar formula also follows from a slight adaptation of the argument we give.\n\n\n\n\nWe will argue along the same lines as in Section \\ref{section:EXC}. The only missing ingredient we need in order to compute the higher moments of $\\abs{S(a)}$ is a strengthening of Lemma \\ref{EXC:lem:kill-bad-U} where $u_k$ are not necessarily distinct. This task is less trivial than it might appear at first.\n\nConsider an example where $u_1 = u_2$, $s_3 = s_1 - s_2$ (e.g. $s_3=s_2=\\frac{1}{2} s_1$) and $\\EE \\sum_{i=u_1}^{u_3} a_i = s_2$. For the condition $\\bigwedge_{k=1}^3 (s_k \\in S_{u_k}(a))$ to hold, it suffices that $s_2 = \\sum_{i=u_1}^{u_3-1} a_i $ and $s_3 \\in S_{u_3}(a)$. One expects the probability of the latter event to be roughly $\\frac{1}{n^{5/2}}$, at least when $s_k$ are not too small and $u_3$ is chosen to maximise $\\mathbb{P}(s_2 = \\sum_{i=u_1}^{u_3-1} a_i)$. This is significantly larger than the bound $\\frac{1}{n^3}$ which one might expect by analogy to Lemma \\ref{EXC:lem:kill-bad-U}. Hence, the direct generalisation of Lemma \\ref{EXC:lem:kill-bad-U} is not possible; instead we prove an averaged version.\n\nAs before, we will only apply the following Lemma in the case when $s_k$ take at most two distinct values.\n\n\\newcommand{\\mathcal{U}}{\\mathcal{U}}\n\\newcommand{\\operatorname{out-deg}}{\\operatorname{out-deg}}\n\\newcommand{\\operatorname{in-deg}}{\\operatorname{in-deg}}\n\n\\begin{lemma}\\label{EXC:lem:kill-bad-U-2}\n\tFix a choice of $\\varepsilon = \\varepsilon(n)$ with $\\varepsilon = o(1)$ and $\\varepsilon = n^{-o(1)}$ as $n \\to 0$. Let $M$ be an integer, for $1 \\leq k \\leq M$ let $s_k = \\sigma_k \\frac{n^2}{2}$ with $\\varepsilon < \\sigma_k < 1-\\varepsilon$. Denote by $\\mathcal{U}$ the set of all sequences $u= (u_k)_{k=1}^M$ such that $\\varepsilon n < u_k < (1-\\sigma_k - \\varepsilon)n$ for all $k$ and \\emph{not} all $u_k$ are distinct. Then\n\n\t\\begin{equation}\\label{EXC:eq:71}\n\t\t\\sum_{u \\in \\mathcal{U}} \\mathbb{P}\\bra{ \\bigwedge_{k=1}^M s_k \\in S_{u_k}(a)} = O\\bra{ \\frac{\\bra{\\frac{1}{\\varepsilon} \\log n}^M}{n}},\n\t\\end{equation}\n\twhere the implicit constant depends only on $M$.\n\\end{lemma}\n\\begin{proof}\n\tThroughout the argument, $M$ is fixed, and we allow all constructions to depend on $M$.\n\t\n\tWe introduce a structure which we call ``type graph'', intended to record possible linear dependencies between different sums of the form $\\sum_{i=u_k}^{v} a_{i}$ and target sums $s_k$. A type graph $G$ consists of the following data:\n\\begin{enumerate}\n\t\\item An initial interval $K = [N]$ together with a partition $K = I \\cup J$, $I \\cap J = \\emptyset$ and $\\abs{I},\\abs{J} \\leq M$.\n\t\\item A collection of $M$ edges $E \\subset K \\times K$, ordered so that if $\\bra{i,j} \\in E$ then $i < j$, such that $\\operatorname{out-deg}(i) \\geq 1$ for each $i \\in I$ and $\\operatorname{in-deg}(j) \\geq 1$ for each $j \\in J$.\n\t\\item A collection of labels $s_{{i,j}}$ for each $\\bra{i,j} \\in E$, where each $s_{{i,j}}$ has the form $s_{{i,j}} = \\sum_{k} q_k s_k$ with $q_k \\in \\mathbb{Z}$ and $\\sum_{k} \\abs{ q_k } \\leq 2^{M^2 - L} $ where $L = \\sum_{\\bra{i,j} \\in E} (j-i)$.\n\\end{enumerate}\n\nThe bound in the final condition is purely technical; it's significance will become clear in the course of the proof. Formally, $G$ is a 4-tuple consisting of $I,J,E$ and $(s_{{i,j}})_{\\bra{i,j} \\in E}$. Whenever we introduce a type graph $G$, we implicitly also introduce $I,J,K$, $E$ and $s_{{i,j}}$ appearing in the definition.\n\nIf $G$ is a type graph, and $(w_k)_{k \\in K}$ is a (strictly) increasing sequence, then we define the condition $\\mathscr{A}_G(a,w)$ to hold precisely when the system of equations:\n\\begin{equation}\\label{EXC:eq:78}\n\t\\sum_{i=w_k}^{w_l-1} a_i = s_{{k,l}}, \\quad \\bra{k,l} \\in E\n\\end{equation}\nis satisfied. \n\nWe will say that two type graphs $G$ and $G'$ are equivalent if the events $\\mathscr{A}_G(a,w)$ and $\\mathscr{A}_{G'}(a,w)$ are equivalent (in particular, $K = K'$). We will call a type graph $G$ minimal if it minimises the ``total edge length'' (which also appears in the final condition defining graph types) \n\\begin{align}\\label{EXC::eq:def-L}\nL = \\sum_{\\bra{k,l} \\in E} (l-k)\n\\end{align}\n within equivalence class. Clearly, each equivalence class contains a minimal element.\n\n\\begin{observation*}\n\tIf $G$ is a minimal type graph, then the underlying unlabelled graph $(K,E)$ is a union of disjoint paths.\n\\end{observation*}\n\\begin{proof}\nFirst note that if $G$ is minimal, then there is no pair of edges $\\bra{k,l}$, $\\bra{m,l} \\in E$ with equal endpoints.\n\nIndeed, if there was such a pair, without loss of generality with $k < m$, then we can replace the edge $\\bra{k,l}$ with $\\bra{k,m}$ and put $s_{k,m} = s_{{k,l}} - s_{{l,m}}$. (Note that this decreases the total edge length $L$ from \\eqref{EXC::eq:def-L} by at least $1$, and thus $s_{{k,l}} - s_{{l,m}}$ is a feasible label). Hence, if $G$ is minimal then for each $l$, $\\operatorname{in-deg}(l) \\leq 1$.\n\n By a similar argument, $\\operatorname{out-deg}(k) \\leq 1$ for each $k$. If follows that\n$G$ (as an unlabelled graph) is a union of disjoint paths.\n\\end{proof}\n\n\\begin{observation*}\n\tIf $(u_k)_{k=1}^M \\in \\mathcal{U}$ and if $a$ is a permutation such that $s_k \\in S_{u_k}(a) $ for each $k$, then there exists a minimal type graph $G$ and an increasing sequence $(w_k)_{k \\in K}$ such that $\\{w_i\\}_{i \\in I} = \\{u_k\\}_{k=1}^M$, $\\abs{I} < M$ and $\\mathscr{A}_G(a,w)$ holds.\n\\end{observation*}\n\\begin{proof}\n\tBy assumption, for each $k$ there is some $v_k$ such that $\\sum_{i=u_k}^{v_k-1} a_i = s_{k}$. We may define an increasing sequence $(w_k)_{k=1}^N$ so that $\\{w_k\\}_{k=1}^N = \\{u_k\\}_{k=1}^M \\cup \\{v_k\\}_{k=1}^M$ and put $K = [N]$. We next take $I = \\{i \\in K \\ : \\ (\\exists k \\in [M]) \\ w_i = u_k \\}$ and $J = K \\setminus I$. Finally, we define $E$ by putting $\\bra{i,j} \\in E$ precisely when $(w_i,w_j) = (u_k,v_k)$ for some $k \\in M$, and take $s_{{i,j}} := s_k$.\n\t\n\tThus defined $G$ satisfies all of the imposed conditions, possibly except minimality. Replacing $G$ by a minimal type graph within the equivalence class, we obtain the claim.\n\\end{proof}\n\nIt follows from the above Observation that the sum in \\eqref{EXC:eq:71} is at most:\n\t\\begin{equation}\\label{EXC:eq:75}\n\t\t\\sum_{\\substack{ G \\text{ -- m.t.g.} \\\\ \\abs{I} < M}}\n\t\t\\sum_{\\substack{ (w_k)_{k \\in K} \\\\ w_k \\geq \\varepsilon n}}\n\t\t\\mathbb{P}( \\mathscr{A}_G(a, w))\n\t\\end{equation}\n\twhere the sum $\\sum_G$ is taken over all minimal type graphs $G$ with $\\abs{I} < M$, and sum $\\sum_{w}$ is taken over all choices of $(w_k)_{k\\in K}$ with $\\varepsilon n \\leq w_k$ for all $k$.\n\n\tSince there are $O_M(1)$ type graphs, to prove the proposition, it will suffice to show that for each minimal type graph $G$ with $\\abs{I} < M$ we have\n\t\\begin{equation}\\label{EXC:eq:76}\n\t\t\\sum_{\\substack{ (w_k)_{k \\in K} \\\\ w_k \\geq \\varepsilon n}}\n\t\t\\mathbb{P}( \\mathscr{A}_G(a, w))\n\t\t\\leq C_M \\frac{ \\bra{ \\frac{1}{\\varepsilon} \\log n}^M}{n},\n\t\\end{equation}\n\twhere the constant $C_M$ depends only on $M$.\t\n\nLet $K_0$ be the set of $k \\in K$ such that $\\operatorname{in-deg}(k) = 0$, and $K_1 = K \\setminus K_0$. If $\\{w_k \\ : \\ k \\in K_0\\}$ is specified then for any permutation $a$ there is at most one choice of $\\{w_k \\ : \\ k \\in K_1\\}$ such that $\\mathscr{A}_G(a,w)$ holds. Hence, we may rewrite the probability in \\eqref{EXC:eq:76} as:\n\t\\begin{equation}\\label{EXC:eq:77}\n\t\t\\sum_{\\substack{ (w_k)_{k \\in K} \\\\ w_k \\geq \\varepsilon n}}\n\t\t\\mathbb{P}( \\mathscr{A}_G(a, w)) = \\sum_{\\substack{(w_m)_{m \\in K_0} \\\\ w_m \\geq \\varepsilon n }} \\mathbb{P}( (\\exists (w_m)_{m \\in K_1}) \\ : \\ \\mathscr{A}_G(a, w)).\n\t\\end{equation}\n\t\n\tWe will now select $a_i$ in the order of increasing $i$, starting at $i(0) = \\floor{ \\varepsilon n} $. Thus, at ``time'' $0 \\leq t \\leq n-i(0)$, we select $a_{i(t)}$, where $i(t) = i(0) + t$. We then select $a_{i}$ with $1 \\leq i < i(0)$ at times $n - i(0) < t \\leq n$, but these values will not play a significant role. To study time evolution, we let $\\mathcal{F}_t$ be the $\\sigma$-algebra generated by $a_{v(t')}$ with $0 \\leq t' \\leq t$. \n\t\n\tAt time $t = 0$, the values $w_m$ with $m \\in K_0$ are specified, but not the values $w_m$ with $m \\in K_1$. If $w_k$ has been specified and $\\bra{k,l} \\in E$ then as $t$ increases, so will $\\sum_{i=w_k}^{i(t)} a_{i}$. As long as $\\sum_{i=w_k}^{i(t)} a_{i} \\neq s_{{k,l}}$ we consider $w_l$ to be unspecified, but as soon $\\sum_{i=w_k}^{i(t)} a_{i} = s_{{k,l}}$ we put $w_l = i(t)+1$. If no such $t$ is found, then $w_l$ is unspecified indefinitely. Note hat whether or not $w_k$ is specified at $t$ is a $\\mathcal{F}_t$-measurable event.\n\t\n\tAt each time $t$, either there is a least value $l = l(t) \\in K_1$ such that $w_{l}$ has not yet been specified, or else all $w_{l}$ have been specified, and we set $l(t) := \\infty$. If $l(t) \\neq \\infty$, there is some $k = k(t)$, such that $\\bra { {k(t)}, {l(t)}} \\in E$ and hence $\\mathscr{A}_G(a,w)$ imposes the condition $\\sum_{i=w_{k(t)}}^{w_{l(t)}} a_i = s_{k(t),l(t)}$. If $l(t) = \\infty$, set $k(t) = \\infty$.\n\n\tWe introduce random variables $X_t,Y_t$, adapted to the filtration $\\mathcal{F}_t$. The variable $X_t$ counts equations required by $\\mathscr{A}_G(a,w)$ which have been satisfied: \n\t$$\n\tX_{t+1}=\n\t\\begin{cases}\n\t1 & \\text{ if $\\sum_{i=w_{k(t)}}^{i(t+1)} a_i = s_{k(t),l(t)}$ ,} \\\\\n\t0 & \\text{ otherwise (including $k(t) = \\infty$) }.\n\t\\end{cases} \n\t$$\t\n\t\nNote that the event $\\mathscr{A}_G(a,w)$ is equivalent to $\\sum_{t} X_t = \\abs{K_1}$. The variable $Y_t$ counts potential opportunities for $X_{t+1}$ to take value $1$: \n\t$$\n\tY_{t+1}=\n\t\\begin{cases}\n\t1 & \\text{ if $s_{k(t),l(t)}-\\sum_{i=w_{k(t)}}^{i(t)} a_i \\in [n]$,} \\\\\n\t0 & \\text{ otherwise (including $k(t) = \\infty$)}.\n\t\\end{cases} \n\t$$\t\n\tNote that if $X_{t+1} = 1$ then $Y_t = 1$. More precisely, we have:\n\t$$\n\t\t\\EE(X_{t+1} \\ | \\ \\mathcal{F}_t) \\leq \\frac{Y_t}{\\varepsilon n},\n\t$$\nsince at most one of more than $\\varepsilon n$ possible values of $a_{i(t+1)}$ guarantees that $X_{t+1} = 1$. \n\nLet $C = C_M$ be a large constant. We always have the following bound:\n\\begin{align*}\n\t\\mathbb{P}\\bra{ \\sum_t X_t = \\abs{K_1} }\n\t&\\leq \\mathbb{P}\\bra{ \\sum_t X_t = \\abs{K_1} \\ \\wedge \\ \\sum_t Y_t \\leq C \\log n }\n\t\\\\& + \\mathbb{P}\\bra{ \\sum_t Y_t > C \\log n }\n\\end{align*} \n(where the sums $\\sum_t$ can be assumed to run over $0 \\leq t \\leq n-\\varepsilon n$).\n\nThe first summand can estimated using a union bound on the choice of the set of $t$ with $X_t = 1$:\n\\begin{align*}\n\\mathbb{P}\\bra{ \\sum_t X_t = \\abs{K_1} \\ \\wedge \\ \\sum_t Y_t \\leq C \\log n } &\\leq \\binom{C \\log n}{\\abs{K_1}}\\bra{ \\frac{1}{\\varepsilon n} }^{\\abs{K_1}} \n\\\\ &= O_M \\bra{ { \\frac{ \\bra{ \\frac{1}{\\varepsilon}\\log n}^{\\abs{K_1}} }{n^{\\abs{K_1}} }}}.\n\\end{align*}\n\nFor the second sum, if $C$ is sufficiently large (with respect to $M$) we have by a standard application of Hoeffding's inequality\n$$\n\\mathbb{P}\\bra{ \\sum_t Y_t > C \\log n } = O_M \\bra{\\frac{1}{n^{\\abs{K_1}} } }.\n$$\n\nIn total, the sum in \\eqref{EXC:eq:77} is bounded from above by $O_M\\bra{ \\frac{\\bra{ \\frac{1}{\\varepsilon}\\log n}^{M}}{n^{\\abs{K_1} - \\abs{K_0}} }}$.\n\nTo finish the proof, it remains to verify that $\\abs{K_1} > \\abs{K_0}$. Recall that, being minimal, $G$ is a union of a certain number $P$ of paths of lengths $l_1,l_2,\\dots,l_P \\geq 2$ with startpoints in $I$ and endpoints in $J$. We have $\\abs{K_0} = P \\leq \\abs{I}$ and $\\abs{K_1} = \\sum_{j=1}^P (l_j -1) = \\abs{E} = M$. Since $M > \\abs{I}$ by assumption, we are done.\n\n\t\n\\end{proof}\n\nWe now have all the tools necessary to compute the second moment of $\\abs{S(a)}$. The argument is very similar to the one we used to compute $\\EE \\abs{S(a)}$. In places where the arguments are virtually identical, we give only the outline, and encourage the reader to look to Section \\ref{section:EXC} for details. \n\n\\begin{proof}[Proof of Proposition \\ref{HM:prop:E(S(a)^p)}]\n\n\tFix a choice of $\\varepsilon$, for instance $\\varepsilon(n) = \\frac{1}{\\log n}$. With the a Riemann integral approximation argument much as before, it will suffice to show that for any $s_1,s_2$ with $s_j = \\sigma_j \\frac{n^2}{2}$, $\\varepsilon < \\sigma_j < 1-\\varepsilon$ one has\n\\begin{equation}\\label{EXC:eq:61}\n\t\\mathbb{P}\\bra{ s_1,s_2 \\not \\in S_\\varepsilon(a) } = (1+o(1)) \\prod_{j=1}^2 e^{-2+2\\sigma_j}.\n\\end{equation}\nwhere the error term is uniform with respect to the choice of $s_1,s_2$. Once this is established, we also have \n$$\\mathbb{P}\\bra{ s_1,s_2 \\in S_\\varepsilon(a) } = (1+o(1)) \\prod_{j=1}^2 (1-e^{-2+2\\sigma_j}).$$\nand hence \n\\begin{align*}\n\t\\EE\\abs{S_\\varepsilon(a)}^2 &= \\sum_{s_1,s_2} \\mathbb{P}\\bra{ s_1,s_2 \\in S_\\varepsilon(a) } + o(n^4)\n\t\\\\ &= \\sum_{s_1,s_2} \\prod_{j=1}^2 (1-e^{-2+2 \\frac{2s}{n^2}}) + o(n^4) \n\t\\\\ &= n^4 \\int_{\\varepsilon}^{1-\\varepsilon} \\prod_{j=1}^2 (1-e^{-2+2\\sigma_j})d \\sigma_1 d\\sigma_2 + o(n^4) \n\t= n^4(c^2 + o(1)),\n\\end{align*} \nwhere the sums run over $\\varepsilon \\frac{n^2}{2} < s_1,s_2 < (1- \\varepsilon) \\frac{n^2}{2}$, and $c = \\frac{1 + e^{-2}}{4}$. Finally, we note that $\\abs{S(a)} = (1+o(1))\\abs{S_\\varepsilon(a)}$, as we have already shown in Section \\ref{section:EXC}. Hence, it remains to prove \\eqref{EXC:eq:61}.\n\nWe may rewrite the probability in \\eqref{EXC:eq:61} using inclusion/exclusion as:\n\\begin{equation}\\label{EXC:eq:62}\n\t\\sum_{M_1=0}^\\infty (-1)^{M_1} \\sum_{\\abs{U_1} = M_1} \\sum_{M_2=0}^\\infty (-1)^{M_2} \\sum_{\\abs{U_p} = M_p} \\mathbb{P}\\bra{ \\bigwedge_{j=1}^2 \\bigwedge_{k=1}^{M_j} \\bra{ s_j \\in S_{u_{j,k}}(a) }},\n\\end{equation}\nwhere the inner sums are taken over all choices of $U_{j} = \\{u_{j,k}\\}_{k=1}^{M_j} \\subset [n]$ such that $\\varepsilon n < u_{j,k} < (1-\\varepsilon-\\sigma_j) n$. \n\nLet $N$ be a large even integer, and put $N_1 = N$ and \n$$\t\n\tN_2 = N_2(M_1) = \n\t\\begin{cases}\n\t\tN & \\text{ if } M_1\\equiv 0 \\pmod{2} \\\\\n\t\tN+1 & \\text{ if } M_1 \\equiv 1 \\pmod{2}.\t\t\n\t\\end{cases}\n$$ \nThen the sum in \\eqref{EXC:eq:62} is bounded from above by the truncated sum\n\\begin{equation}\\label{EXC:eq:62a}\n\t\\sum_{M_1=0}^{N_1} (-1)^{M_1}\n\t\\sum_{M_2=0}^{N_2} (-1)^{M_2} \n\t\\sum_{\\substack{ \\abs{U_j} = M_j \\\\ j=1,\\dots,p}}\n\t\\mathbb{P}\\bra{ \\bigwedge_{j=1}^2 \\bigwedge_{k=1}^{M_j} \\bra{ s_j \\in S_{u_{j,k}}(a) }}.\n\\end{equation}\n\nWith the same definitions, for odd values of $N$ the expression above gives a lower bound. Hence, to find asymptotics for the sum \\eqref{EXC:eq:62}, it will suffice to find asymptotics for each of summands with $M_1,M_2$ fixed.\n\nWe now consider the innermost sum. Let $U = U_{1} \\cup U_2$. Using Lemmas \\ref{EXC:lem:kill-bad-U} and \\ref{EXC:lem:kill-bad-U-2}, we may disregard the contribution $O\\bra{\\frac{m \\bra{ \\frac{1}{\\varepsilon} \\log n}^M}{n}} =o(1)$ from $U$ with $U_1 \\cap U_2 \\neq \\emptyset$, and from $U$ which do not satisfy the regularity condition \\eqref{EXC:cond:U-reg} for a suitable choice of the parameter $m = m(n)$. For remaining $U$, we have from Proposition \\ref{EXC:prop:P-wedge-s-in-S} (or more precisely, the internal Lemma in its proof) that \n\\begin{equation}\\label{EXC:eq:63}\n\t\t\\sum_{\\substack{ \\abs{U_j} = M_j \\\\ j=1,2}}\n\t\\mathbb{P}\\bra{ \\bigwedge_{j=1}^2 \\bigwedge_{k=1}^{M_j} \\bra{ s_j \\in S_{u_{j,k}}(a) }}\n = (1+o(1))\\bra{ \\frac{2}{n} }^{M_1+M_2}.\n\\end{equation}\n\nThe number of choices of $(U_1,U_2)$ when $(M_1,M_2)$ are fixes is equal to $(1+o(1)) \\prod_{j=1}^2 \\frac{(1-\\sigma_j)^{M_j}}{M_j!}n^{M_j}$. Thus the inner sum is asymptotically equal to:\n\\begin{equation}\\label{EXC:eq:64}\n\\bra{ \\frac{2}{n} }^{M_1+M_2} \\prod_{j=1}^2 \\frac{(1-\\sigma_j)^{M_j}}{M_j!}n^{M_j} \n= \\prod_{j=1}^2 \\frac{(2-2\\sigma_j)^{M_j}}{M_j!}.\n\\end{equation}\n\nThus, for any large even integer $N$ the sum in \\eqref{EXC:eq:62a} is bounded from above by:\n\\begin{equation}\\label{EXC:eq:65}\n\t\\sum_{\\substack{ M_j=0 \\\\ j = 1,2}}^{\\infty} (-1)^{M_1+M_2}\n\t\\prod_{j=1}^2 \\frac{(2-2\\sigma_j)^{M_j}}{M_j!}\n\t+ \\frac{C}{N!} + o(1),\n\\end{equation}\nwhere $C$ is an absolute constant. The sum in \\eqref{EXC:eq:65} is simply the Taylor expansion of $\\prod_{j=1}^2 e^{-2+2\\sigma_j}$. Letting $N \\to \\infty$ slowly with $n$, and repeating the same argument for $N$ odd, we conclude that \\eqref{EXC:eq:61} holds.\n\n\n\\end{proof}\n\\section{Lower bound}\\label{section:EXT-low}\n\nWe will presently prove a lower bound on $\\max_{a} \\abs{S\\bra{a}}$ as $a$ runs over permutations of $[n]$, which is slightly better than $\\max_{a} \\abs{S\\bra{a}} \\geq \\EE_a \\abs{S\\bra{a}} = \\bra{ \\frac{1+e^{-2}}{4} +o(1)} n^2 $ (which in turn is slightly better than $\\max_{a} S(a) \\geq (\\frac{1}{4}+o(1))n^2$ from Proposition \\ref{I:prop:cexple-basic}). \n\nNamely, we will show that $\\max_{a} \\abs{S\\bra{a}} \\geq (\\frac{3}{2}-\\frac{2}{\\sqrt{e}} + o(1)) n^2$, thus proving Proposition \\ref{I:prop:cexple-improved}. We accomplish this by introducing a randomised variant of the construction from Proposition \\ref{I:prop:cexple-basic}.\n\nWe note that the numerical values of the two constants mentioned above are rather close: $\\frac{3}{2}-\\frac{2}{\\sqrt{e}} = 0.286\\dots$, while $\\frac{1+e^{-2}}{4} = 0.283\\dots$. Hence, the quality of the bound is hardly a reason for interest in the results of this section. Instead, we point out that our construction gives a fairly broad and explicit class of permutations with $\\abs{S(a)}$ significantly above the expected value. In particular, the bound $\\max_{a} \\abs{S\\bra{a}} \\geq \\EE_a \\abs{S\\bra{a}}$ is far from sharp.\n\n\\renewcommand{\\l}{\\lambda}\n\\begin{proof}[Proof of Proposition \\ref{I:prop:cexple-improved}]\\mbox{}\\\\\n\tLet $a$ be a permutation of $[n]$ selected uniformly at random, subject to the condition that $a_{i} + a_{i+1} = n+1$ for each odd $i$. We will show that $\\EE \\abs{S(a)} \\geq (c+o(1)) n^2$ where $c = \\frac{3}{2}-\\frac{2}{\\sqrt{e}}$.\n\t\n\tLet us fix a choice of $s = l\\bra{n+1} + r$, where $l = \\lambda \\frac{n}{2}$ and $r = \\rho n$ are integers and $0 \\leq \\lambda, \\rho \\leq 1$.\n\t\n\tOur first goal is to estimate $\\mathbb{P}\\bra{s \\in S(a)}$, where we fix the value of $s$.\n\t \nIt is clear that if $s = \\sum_{i=u}^v a_i$ then $v-u+1 \\in \\{2l,2l+1,2l+2\\}$. Let us enumerate all possible intervals of length $2l$, and let $\\mathscr{A}_i$ be the event that $s$ is the sum of $a_j$ over the $i$-th interval, where $1 \\leq i \\leq n-2l = \\bra{1-\\l + o(1)}n$. Likewise, let $\\mathscr{B}_j$ be the event that $s$ is the sum over the $j$-th interval of length $2l+2$, $1 \\leq j \\leq n-2l-2$. Finally, let $\\mathscr{C}$ be the event that $s$ is the sum over some interval of length $2l+1$; there can be at most one such interval.\n\nFor each $i$ there are some $u,v$ (namely the endpoints of the relevant interval) such that $\\mathscr{A}_i$ holds if and only if $r = a_u + a_v$. Hence $\\mathbb{P}\\bra{\\mathscr{A}_i} = \\frac{\\rho}{n} + O\\bra{\\frac{1}{n^2}}$. Similarly, each $\\mathscr{B}_j$ is equaivalent to $r + \\bra{n+1} = a_v + a_u$ for some $u,v$, and $\\mathbb{P}\\bra{\\mathscr{B}_j} = \\frac{1-\\rho}{n} + O\\bra{\\frac{1}{n^2}}$. Likewise, $\\mathscr{C}$ holds if and only if $r = a_u$ where either $u$ is even and $\\leq n-2l$ or $u$ is odd and $\\geq 2l$, and thus $\\mathbb{P}\\bra{\\mathscr{C}} = 1-\\l + O\\bra{\\frac{1}{n}}$.\n\nMoreover, we have the following asymptotic independence condition.\n\\begin{lemma*\nLet $k,l$ be integers. Then\n\\begin{equation}\\label{EXT:eq:11}\n\t\\EEE_{\\abs{I} = k} \\EEE_{\\abs{J}=l} \\mathbb{P}\\brabig{ \\bigwedge_{i \\in I} \\mathscr{A}_i \\wedge \\bigwedge_{j \\in J} \\mathscr{B}_j } = \\bra{\\frac{\\rho}{n}}^k \\bra{\\frac{1-\\rho}{n}}^l + O\\bra{\\frac{1}{n^{k+l+1}}},\n\\end{equation}\nwhere the expected values are taken over all possible choices of index sets of given size. \nLikewise, we have \n\\begin{equation}\\label{EXT:eq:12}\n\t\\EEE_{\\abs{I} = k} \\EEE_{\\abs{J}=l} \\mathbb{P}\\brabig{ \\bigwedge_{i \\in I} \\mathscr{A}_i \\wedge \\bigwedge_{j \\in J} \\mathscr{B}_j \\wedge \\mathscr{C} } \n\t= \\bra{\\frac{\\rho}{n}}^k \\bra{\\frac{1-\\rho}{n}}^l \\bra{1-\\l} + O\\bra{\\frac{1}{n^{k+l+1}}}.\n\\end{equation}\nHere, the implicit constants may depend on $k,l$.\n\\end{lemma*} \n\\begin{proof} \n\nWe only give the argument for the former equality, the latter being analogous. For given $I,J$ with $\\abs{I} = k,\\ \\abs{J} = l$, let $\\mathscr{E} = \\mathscr{E}_{I,J} = \\bigwedge_{i \\in I} \\mathscr{A}_i \\wedge \\bigwedge_{j \\in J} \\mathscr{B}_j$. Also, let $m = k+l$.\n\nRecall that each event $\\mathscr{A}_i$ is equivalent to $a_{u_i} + a_{v_i} = r$, and likewise $\\mathscr{B}_j$ is equivalent to $a_{u_j'} + a_{v_j'} = r+n+1$. Because of the restriction $a_{i}+a_{i+1} = n+1$ for odd $i$, the event $\\mathscr{E}$ is equivalent to a system of $m$ linear equations of the form \n\\begin{equation}\\label{EXT:eq:01}\n\\sum_{j=1}^{2m} M_{ij} b_j = c_i \\qquad (i \\in [m]),\n\\end{equation} \n where $M_{ij} \\in \\{0,-1,+1\\}$ with $M_{ij} \\neq 0$ for precisely $2$ values of $j$ for a fixed $i$, $b_j = a_{w_j}$ for some odd $w_j \\in [n]$ and $c_{i} \\in \\{r,r+(n+1),r-(n+1),r-2(n+1)\\}$. We may write \\eqref{EXT:eq:01} more briefly as $Mb = c$.\n \nWe will need a uniform bound on $\\mathbb{P}(\\mathscr{E}_{I,J})$, valid for all choices of $I,J$ with $\\abs{I} = k,\\abs{J} = l$. For this purpose, we introduce an (undirected) graph $G = G_{I,J}$ with vertex set $[2m]$ such that the edge $(j_1, j_2)$ is present precisely when there exists a row $i$ such that $M_{i,j_1}, M_{i,j_2} \\neq 0$.\n\nNote that if $\\mathbb{P}(\\mathscr{E}_{I,J}) \\neq 0$ then for any pair $j_1,j_2$ there can be at most one $i$ such that $M_{i,j_1}, M_{i,j_2} \\neq 0$; else one could find a pair of intervals with endpoints differing by $1$ and both sums equal to $s$, which is impossible. Hence, $G$ has precisely $m$ edges. Moreover, it follows from the construction that $G$ is acyclic. \nHence, $G$ has $2m-m = m$ connected components (possibly some of which are singletons). \n\nFor each connected component of $G$, one value of $b_j$ may be chosen freely, and then \\eqref{EXT:eq:01} determines the remaining ones. Hence, \n$$\\mathbb{P}(\\mathscr{E}_{I,J}) = \\mathbb{P}(Mb=c) = O\\bra{\\frac{1}{n^m}},$$ where the implicit constant depends at most on $m$. \n\nHaving obtained the universal upper bound above, we will restrict to suitably generic $I,J$. If $I, J$ are chosen randomly from all subsets of $[n]$ of suitable sizes $\\abs{I} = k,\\ \\abs{J} =l$, then with probability $1 - O\\bra{\\frac{1}{n}}$ the corresponding graph $G$ consist of $m$ disconnected intervals --- this happens precisely if no pair of indices $u_i,v_i,u'_j,v'_j$ introduced above is equal, nor is any pair of the form $w,w+1$ with $w$ odd.\n\nHence, at the cost of introducing an error term $O\\bra{\\frac{1}{n^{m+1}}}$, we may restrict our attention to $I,J$ such that the corresponding grahp $G$ is a union of $m$ connected, components as described above.\n\nWe now argue for a fixed choice of $I,J$. Suppose that $\\mathscr{D}$ is one of the events $\\mathscr{A}_i,\\mathscr{B}_j$, and let $u,v$ be the relevant coordinates $u_i,v_i$ or $u'_j,v'_j$ accordingly. For an integer $N$ (bounded in terms of $k,l$), let $d_1,\\dots,d_N \\in [n]$ and $i_{1},\\dots,i_{N} \\in [n]$ be arbitrary. Then \n\\begin{equation}\\label{EXT:eq:02a}\n\t\\mathbb{P}\\bra{\\mathscr{D} \\mid a_{i_1}=d_1,\\dots,a_{i_N} = d_N} = \\mathbb{P}\\bra{\\mathscr{D}} + O\\bra{\\frac{1}{n^2}},\n\\end{equation}\n\twhere the implicit constant depends only on $N$, provided that the event we condition on above has non-zero probability and that $i_1,\\dots,i_N$ differ from $u,v$ by at least $2$. By a standard argument, the above equality \\eqref{EXT:eq:02a} implies that for any $I' \\subset I,\\ J' \\subset J$ we have\n\\begin{equation}\\label{EXT:eq:02b}\n\t\\mathbb{P}\\bra{\\mathscr{D} \\mid \\bigwedge_{i \\in I'} \\mathscr{A}_i \\wedge \\bigwedge_{j \\in J'} \\mathscr{B}_j } = \\mathbb{P}\\bra{\\mathscr{D}} + O\\bra{\\frac{1}{n^2}}.\n\\end{equation}\nOrdering $\\mathscr{A}_i,\\mathscr{B}_j$ as $\\mathscr{D}_i$ with $i \\in K$, putting $\\mathscr{E}_j' = \\bigwedge_{i < j} \\mathscr{D}_j$ and using $\\mathbb{P}(\\mathscr{D}_j) = O\\bra{\\frac 1n}$ we find\n\\begin{align*}\\label{EXT:eq:02c}\n\t\\mathbb{P}\\bra{\\mathscr{E} } &= \\prod_{j \\in K} \\mathbb{P}\\bra{\\mathscr{D}_j \\mid \\mathscr{E}_j'} = \\prod_{j \\in K} \\bra{ \\mathbb{P}\\bra{\\mathscr{D}_j} + O\\bra{\\frac{1}{n^2}}}\n\t\\\\ &= \\prod_{j \\in K} \\mathbb{P}\\bra{\\mathscr{D}_j} + O\\bra{\\frac{1}{n^{m+1}}}.\n\\end{align*}\n\nThis is precisely the stated bound.\n\\end{proof}\n\nWe may now estimate $\\mathbb{P}\\bra{s \\not \\in S(a)}$ from inclusion-exclusion formula, truncated at level $N$, where $N$ is a large even integer:\n\\begin{align*}\n\t\\mathbb{P}\\bra{s \\not \\in S(a)} & \n\t\\leq \\sum_{k+l\\leq N}\\sum_{\\abs{I} = k} \\sum_{\\abs{J} = l} \\bra{-1}^{k+l} \\mathbb{P}\\bra{\\bigwedge_{i \\in I} A \\wedge \\bigwedge_{j \\in J} \\mathscr{B}_j} \n\t\\\\& - \n\t\\sum_{k+l < N}\\sum_{\\abs{I} = k} \\sum_{\\abs{J} = l} \\bra{-1}^{k+l} \\mathbb{P}\\bra{\\bigwedge_{i \\in I} A \\wedge \\bigwedge_{j \\in J} \\mathscr{B}_j \\wedge C}.\n\\end{align*}\nSubstituting bounds from the Lemma above we find:\n\\begin{align*}\n\\mathbb{P}\\bra{s \\not \\in S(a)} &\n\\leq \\sum_{k+l\\leq N} \\bra{-1}^{k+l} \\bra{ \\frac{\\rho}{n}}^k\n \\bra{ \\frac{1-\\rho}{n}}^l \\frac{\\bra{\\frac{1-\\lambda}{2}n}^{k+l} }{k! \\ l!} \n \\\\ &-\\sum_{k+l < N} \\bra{-1}^{k+l} \\bra{ \\frac{\\rho}{n}}^k\n \\bra{ \\frac{1-\\rho}{n}}^l \\frac{\\bra{\\frac{1-\\lambda}{2}n}^{k+l} }{k! \\ l!} \\bra{1-\\lambda} + O_N\\bra{\\frac{1}{n}} \n \\\\& = \\sum_{m \\leq N} \\bra{-1}^{m} \\frac{\\bra{\\frac{1-\\lambda}{2}}^{m} }{ m!} \\lambda\n + O\\bra{\\frac{1}{N!}} + O_N\\bra{\\frac{1}{n}} \n \\\\& = e^{-\\frac{1-\\lambda}2}\\lambda + O\\bra{\\frac{1}{N!}} + O_N\\bra{\\frac{1}{n}}, \n\\end{align*}\nwhere we use the notation $O_N\\bra{\\cdot}$ to signify that the implicit constant is allowed to depend on $N$. Letting $N \\to \\infty$ slowly with $n$, we conclude that:\n$$\n\\mathbb{P}\\bra{s \\not \\in S(a)} \\leq e^{-\\frac{1-\\lambda}{2}}\\lambda + o(1),\n$$\nwith the error term uniform with respect to the choice of $s$. It follows that:\n\\begin{align*}\n\t\\EE \\abs{S(a)} \n\t& \\geq n^2\\frac{1+o(1)}{2}\\int_0^{1} \\int_0^{1} \\bra{1 - e^{-\\frac{1-\\lambda}{2}}\\lambda} d\\lambda d\\rho \n\t\\\\ &= n^2 \\bra{\\frac{3}{2}-\\frac{2}{\\sqrt{e}} - o\\bra{1} }.\n\\end{align*}\nA symmetric argument yields $\\EE \\abs{S(a)} \\leq n^2 \\bra{\\frac{3}{2}-\\frac{2}{\\sqrt{e}} + o\\bra{1} }$, which finishes the proof.\n\\end{proof}\n\n\\section{Upper bound}\\label{section:EXT-up}\n\nAt this stage it seems that the distinct sums $\\sum_{i=u}^v a_i$ tend to be rather numerous. The trivial upper bound on their quantity is $\\binom{n+1}{2}$, which happens to be both the upper bound for any single sum and the number of distinct intervals.\n\nIt is natural to ask if this bound is sharp, and it turns out that it is not. In this section we obtain a slight improvement, namely\n\\begin{equation}\n{S(a)} \\leq \\bra{\\frac 14 + \\frac{\\pi}{16} + o(1)}{n^2},\n\\label{EXTup::eq:L-def}\n\\end{equation}\nthus proving Proposition \\ref{I:prop:upper-bound}.\n\nWe will consider the set of sums which are above average value. Define:\n\\begin{equation}\n\tL(a) := \\bbra{ (u,v) \\in [n]^2 \\ : \\ u \\leq v, \\ \\sum_{i=u}^v a_i \\geq \\frac{1}{2} \\binom{n+1}{2} }.\n\t\\label{EXTup:eq:La-def}\n\\end{equation}\n\t\nOur main idea is to show that $L(a)$ can never be too large. The following proposition easily implies Proposition \\ref{I:prop:upper-bound}.\n\t\n\\begin{proposition}\\label{EXTup:prop:L(a)-bound}\n\tLet $a$ be a permutation of $[n]$, and let $L(a)$ be defined as above. Then:\n\t$$\n\t\tL(a) \\leq \\bra{\\frac{\\pi}{16} + o(1)} n^2.\n\t$$\n\\end{proposition}\n\n\\begin{remark*}\n\tThe constant $\\frac{\\pi}{16}$ cannot be improved, as shown by the ``tent map'' permutation:\n\t$$\n\t\ta_i = \n\t\t\\begin{cases}\n\t\t2 i, & i \\leq \\frac{n}{2} \\\\\n\t\t2(n-i) + 1, & i > \\frac{n}{2}.\n\t\t\\end{cases}\n\t$$\n\tThis is essentially the only possible example, as will become clear in the course of the proof. Note, however, that for this particular permutation we have $S(a) = o(n^2)$ for similar reasons as for the trivial permutation; see also Proposition \\ref{LOW:prop:small-S}. We believe that the bound in \\ref{I:prop:upper-bound} is not sharp.\n\\end{remark*}\n\n\\begin{proof}[Proof of Proposition \\ref{I:prop:upper-bound} assuming Proposition \\ref{EXTup:prop:L(a)-bound}]\n\tPartitioning $S(a)$ in two parts at $\\frac{1}{2} \\binom{n+1}{2}$ we easily find that:\n\\begin{equation*}\n\\abs{S(a)} \\leq \\frac{1}{2} \\binom{n+1}{2} + \\abs{L(a)} \\leq\\bra{ \\frac{1}{4} + \\frac{\\pi}{16} + o(1)}n^2.\n\\end{equation*} \nThis is precisely the sought bound.\n\\end{proof}\n\nWe will devote the remainder of this section to proving Proposition \\ref{EXTup:prop:L(a)-bound}.\n\nTo begin with, we reduce the problem to the case when the permutation can be partitioned into two monotonous parts. This part of the argument restricts the sample space significantly, and will be crucial to ensure that the continuous version of the problem has enough compactness for the maximum to be realised.\n\n\\begin{observation}\\label{EXTup:obs:perm-mono}\n\tLet $a$ be a permutation on $[n]$. Then there exists another permutation $a'$ such that $\\abs{L(a)} \\leq \\abs{L(a')}$ and there exists $k$ such that $a_1',\\dots,a_k'$ is increasing and $a_k',a_{k+1}',\\dots, a_n'$ is decreasing.\n\\end{observation}\n\\begin{proof}\n\tChoose $k$ so that $\\sum_{i=1}^{k} a_i \\geq \\frac{1}{2} \\binom{n+1}{2}$ and $\\sum_{i=k}^{n} a_i \\geq \\frac{1}{2} \\binom{n+1}{2}$.\n\n\tConsider the permutation $a'$ obtained from $a$ by sorting $a_1,\\dots,a_k$ in increasing order and $a_{k+1},\\dots,a_n$ in the decreasing order. More precisely, let $a'$ be such that $\\{ a_i' \\ : \\ 1 \\leq i \\leq k\\} = \\{ a_i \\ : \\ 1 \\leq i \\leq k\\}$, $\\{ a_i' \\ : \\ k < i \\leq n\\} = \\{ a_i \\ : \\ k < i \\leq n\\}$, $a_1' < \\dots < a_k'$ and $a_{k+1}' > \\dots > a_n'$. Clearly, $a'$ has the required monotonicity property.\n\t\n\tWe claim that $L(a) \\subseteq L(a')$. Suppose that $(u,v) \\in L(a)$. By the choice of $k$, we have $u \\leq k \\leq v$. Thus,\n\t$$ \n\t\t\\frac{1}{2} \\binom{n+1}{2} \\leq \\sum_{i=u}^v a_i \n\t\t= \\sum_{i=u}^k a_i + \\sum_{i=k+1}^v a_i\n\t\t\\leq \\sum_{i=u}^k a_i' + \\sum_{i=k+1}^v a_i' = \\sum_{i=u}^v a_i'. \n\t$$\t\n\tThus, $(u,v) \\in L(a')$, as desired.\n\\end{proof}\n\n\tWe are now ready to introduce the continuous variant. The analogue of the space of all permutations of $[n]$ obeying the monotonicity condition in Observation \\ref{EXTup:obs:perm-mono} is the family ${ \\mathcal{F}_{\\mathrm{mon}} }$ of measurable functions $f\\colon [0,1] \\to [0,1]$ obeying the following conditions:\n\\begin{enumerate}\n\t\\item\\label{EXTup:cond:01} for any measurable $E \\subset [0,1]$, $\\int_{E} f(x)dx \\geq \\frac{\\abs{E}^2}{2}$, and $\\int_0^1 f(x) dx = \\frac{1}{2}$,\n\t\\item\\label{EXTup:cond:02} there exists $\\kappa=\\kappa_f$ such that $\\int_{0}^\\kappa f(x) dx = \\int_{\\kappa}^1 f(x) dx = \\frac{1}{4}$, and moreover $f|_{[0,\\kappa]}$ is increasing and $f|_{[\\kappa,1]}$ is decreasing.\n\\end{enumerate}\t\n(Here and elsewhere, if $E \\subset \\mathbb{R}$ is measurable then $\\abs{E}$ denotes the Lebesgue measure of $E$.) Note that for a permutation $a$ and any index set $I$ we have $\\sum_{i \\in I} a_i \\geq \\binom{\\abs{I}+1}{2}$, in analogy to condition \\eqref{EXTup:cond:01}.\n\nWe will also occasionally need to use the larger family $\\mathcal{F}$ of functions $f\\colon [0,1] \\to [0,1]$ which only satisfy the condition \\eqref{EXTup:cond:01} but not necessarily \\eqref{EXTup:cond:02}. We note in passing that $\\mathcal{F}$ is convex, and both ${ \\mathcal{F}_{\\mathrm{mon}} }$ and $\\mathcal{F}$ are closed in the $L^1$ topology.\n\nAnother thing we are missing is an continuous analogue of $L(a)$ from \\eqref{EXTup:eq:La-def}. For any $f \\in \\mathcal{F}$ we define:\n\\begin{align}\n\tL(f) &:= \\bbra{(x,y) \\in [0,1]^2 \\ : \\ x \\leq y,\\ \\int_x^y f(t) dt \\geq \\frac{1}{4} }\n\\end{align}\n\nThe continuous analogue of Proposition \\ref{EXTup:prop:L(a)-bound} is the following statement.\n\n\\begin{proposition}\\label{EXTup:prop:L(f)-bound}\n\tSuppose that $f \\in { \\mathcal{F}_{\\mathrm{mon}} }$. Then $\\abs{L(f)} \\leq \\frac{\\pi}{16}$.\n\\end{proposition}\n\nThis bound is sharp. The (unique) function $f$ with $\\Lambda(f) = \\frac{\\pi}{16}$ will turn out to be the ``tent map'':\n$$\n\t\tf(x) = \n\t\t\\begin{cases}\n\t\t\t2x & \\text{ if } x \\leq \\frac{1}{2}\\\\\n\t\t\t2(1-x) & \\text{ if } x \\geq \\frac{1}{2}.\n\t\t\\end{cases}\n$$\n\nWe defer the proof of Proposition \\ref{EXTup:prop:L(f)-bound}. Our immediate goal is to deduce Proposition \\ref{EXTup:prop:L(a)-bound} from Proposition \\ref{EXTup:prop:L(f)-bound}. Before we do that, we make some preliminary observations which will be useful in the course of the proof, as well as in the main body of the argument. For $f \\in \\mathcal{F}$, define \n\\begin{align}\n\t\tv_f(u) & := \\sup\\left\\{ v \\in [0,1] \\ : \\ \\int_{u}^v f(x) dx \\leq \\frac{1}{4} \\right\\}, \\\\\n\t\tu_f(u) & := \\inf\\left\\{ u \\in [0,1] \\ : \\ \\int_{u}^v f(x) dx \\leq \\frac{1}{4} \\right\\}\t, \\\\\n\t\t\\Lambda(f) &:= \\abs{L(f)}.\n\\end{align}\n\n\\begin{observation}\\label{EXTup:obs:prelim}\n\tWith the definitions as above, the following are true.\n\t\\begin{enumerate}\n\t\\item The maps $\\mathcal{F} \\ni f \\mapsto v_f \\in L^1([0,1])$ and $f \\mapsto u_f$ are well defined and continuous, where on $\\mathcal{F}$ we take the $L^1$ topology.\n\t\\item For any $u_0,v_0$ with $\\int_{u_0}^{v_0} f(x) dx = \\frac{1}{4}$ we have\n\t$$ \\Lambda(f) = \\int_{0}^{u_0} (1-v_f(x)) dx + \\int_{v_0}^1 u_f(x)dx- u_0 (1-v_0).$$\n\t\\item In particular, we have the formulas $$\\Lambda(f) = \\int_{0}^1 (1-v_f(u))du = \\int_{0}^1 u_f(v) dv.$$\n\t\\item The map $\\mathcal{F} \\ni f \\mapsto \\Lambda(f) \\in \\mathbb{R}$ is continuous.\n\t\\item If $f \\in { \\mathcal{F}_{\\mathrm{mon}} }$, then the set $L(f)$ is convex.\n\t\\end{enumerate}\n\\end{observation}\t\n\n\\begin{proof}\n\tThe integral formula for $\\Lambda(f)$ follows from partitioning $L(f)$ into three parts: $ L_- = L(f) \\cap \\{ u < u_0,\\ v< v_0 \\}$, ${L}_+ = {L}(f) \\cap \\{ u > u_0,\\ v > v_0 \\}$ and ${L}_+ = {L}(f) \\cap \\{ u \\leq u_0,\\ v \\leq v_0 \\} = [0,u_0] \\times [v_0,1]$.\n\t\n\tContinuity of $f \\mapsto \\Lambda(f)$ is clear once the previous points are established. It remains to prove continuity of $f \\mapsto v_f$ (argument for $f \\mapsto u_f$ is analogous).\n\t\n\tTake any $f,f_n \\in \\mathcal{F}$ with $f_n \\to f$ in $L^1$. Fix $u$ and let $v = v_f(u)$. Let us suppose that $v < 1$, since the case $v = 1$ is easier. For any $\\delta > 0$ we have\n\t$$\n\t\t\\int_u^{v-\\delta} f(x) dx + \\frac{1}{4} \\delta^2 < \n\t\t\\int_u^{v} f(x) dx = \\frac{1}{4} < \\int_u^{v+\\delta} f(x) dx - \\frac{1}{4} \\delta^2\n\t$$\n\tThus, for $n > n_0(\\delta)$ we have:\t\n\t$$\n\t\t\\int_u^{v-\\delta} f_n(x) dx + \\frac{1}{8} \\delta^2 < \\frac{1}{4} < \\int_u^{v+\\delta} f_n(x) dx - \\frac{1}{8} \\delta^2.\n\t$$\n\tConsequently, $v-\\delta < v_{f_n}(u) < v + \\delta$. Taking $\\delta \\to 0$ we conclude that $v_{f_n}(u) \\to v$ as $n \\to \\infty$. Hence, $v_{f_n} \\to v_f$ pointwise. Since all relevant functions are bounded, $v_{f_n} \\to v_f$ in $L^1$.\n\t\n\tFinally, we prove convexity of $L(f)$. Suppose that $(u_1,v_1), (u_2,v_2) \\in L(f)$ and $u = \\frac{u_1+u_2}{2}, v= \\frac{v_1+v_2}{2}$. We may without loss of generality assume that $u_1 < u_2 < \\kappa_f < v_1 < v_2$, and that $\\int_{u_1}^{v_1}f(x)dx = \\int_{u_2}^{v_2}f(x)dx = \\frac{1}{4}$. We then have \n\t$$\n\t\t\\int_{u_1}^{u_2}f(x)dx = \\int_{v_1}^{v_2}f(x)dx =: I,\n\t$$\n\tand because of monotonicity $\\int_{u_1}^{u}f(x)dx \\leq \\frac{1}{2} \\leq \\int_{u}^{u_2}f(x)dx$ and $\\int_{v_1}^{v}f(x)dx \\geq \\frac{1}{2} \\leq \\int_{v}^{v_2}f(x)dx$. Hence,\n\t$$ \\int_{u}^{v}f(x)dx = \\int_{u_1}^{v_1}f(x)dx - \\int_{u_1}^{u}f(x)dx + \\int_{v_1}^{v}f(x)dx \\geq \\frac{1}{4},$$\n\tand consequently $(u,v) \\in L(f)$. Since $L(f)$ is closed, this proves convexity.\n\\end{proof}\t\n\n\n\\begin{proof}[Proof of Proposition \\ref{EXTup:prop:L(a)-bound} assuming Proposition \\ref{EXTup:prop:L(f)-bound}]\n\n\t For each $n$, let $a^{(n)}$ be a permutation of $[n]$ which maximizes $\\abs{L(a^{(n)})}$.\n\t We may assume without loss of generality that $a^{(n)}_i$ are increasing for $i=1,\\dots,k^{(n)}$ and decreasing for $i = k^{(n)},\\dots,n$. \n\t\n\tTo the permutation $a^{(n)}$ we may associate a function $f_n\\colon[0,1] \\to \\mathbb{R}$ defined by\n\\begin{equation}\n \tf_n\\bra{\\frac{i + t}{n}} = \\frac{a_i^{(n)}}{n+1},\\quad \\text{ for } i \\in [n],\\ t \\in [-1,0),\n\\end{equation} \t\n(and for completeness, put $f(1) := \\frac{a_n^{(n)}}{n+1}$). \n\t \n\t We have the obvious bounds $0 \\leq f_n(x) \\leq 1$, as well as the formula $$\\sum_{i=u+1}^{v} a_i^{(n)} = n(n+1) { \\int_{u/n}^{v/n} f_n(x) dx}$$ for $u,v \\in [n]$. In particular, $\\int_0^1 f_n(x) dx = \\frac{1}{2}$. It is not difficult to see that for any measurable $E$ with $\\abs{E} = \\frac{m + \\mu}{n}$ with $m \\in \\mathbb{N}, \\mu \\in [0,1)$ we have \n\t $$\\int_E f_n(x) dx \\geq \\frac{1 + 2 + \\dots + m + (m+1)\\mu }{n(n+1)} = \\frac{(m+\\mu)(m+1)}{2n(n+1)^2} \\geq \\frac{\\abs{E}^2}{2},$$\nwhere the last inequality can be checked with elementary methods. It is also clear from the construction that $f_n$ is monotonous on $[0,\\frac{k}{n}]$ and on $[\\frac{k-1}{n},1]$.\n\n\tHence, $f_n \\in { \\mathcal{F}_{\\mathrm{mon}} }$. By Proposition \\ref{EXTup:prop:L(f)-bound}, $\\Lambda(f_n) \\leq \\frac{\\pi}{16}$. It remains to relate $\\Lambda(f_n)$ to $\\abs{L(a^{(n)})}$. For any $u,v$, if $(u,v) \\in L(a^{(n)})$ then the square $[\\frac{u}{n},\\frac{u+1}{n}] \\times [\\frac{v}{n},\\frac{v+1}{n}]$ is contained in ${L}(f_n)^c$, and conversely if $(u,v) \\not\\in L(a^{(n)})$ then square $[\\frac{u+1}{n},\\frac{u+2}{n}] \\times [\\frac{v-1}{n},\\frac{v}{n}]$ is contained in ${L}(f_n)$. Together with the observation that $L(f)$ is convex,\n\tthis shows that $\\frac{\\abs{L(a^{(n)})}}{n^2} - \\Lambda(f_n) = O(\\frac 1n)$.\n\t\n\tThus, $\\abs{L(a)} = (1+o(1)) \\Lambda(f_n) n^2 \\leq \\frac{\\pi}{16}n^2,$ which finishes the proof.\n\t\n\\end{proof}\n\t\n\tThe rest of this section will be devoted to proving Proposition \\ref{EXTup:prop:L(f)-bound}. Our first step in that direction is to show that the supremum $\\sup_{g \\in \\mathcal{F}} \\Lambda(g)$ is realised by some $f \\in { \\mathcal{F}_{\\mathrm{mon}} }$, where it is convenient to allow $g$ to range over the larger family $\\mathcal{F}$ to simplify perturbation arguments later.\n\n\t\\begin{observation}\n\n\tThere exists $f \\in { \\mathcal{F}_{\\mathrm{mon}} }$ such that $\\Lambda(f) = \\sup_{g \\in \\mathcal{F}} \\Lambda(g)$\n\\end{observation}\n\\begin{proof}\n\tFor any $g \\in \\mathcal{F}$, there exists $\\tilde g \\in { \\mathcal{F}_{\\mathrm{mon}} }$ such that $\\Lambda(\\tilde g) \\geq \\Lambda(g)$. This follows from an argument essentially equivalent to the one in Observation \\ref{EXTup:obs:perm-mono}.\n\tHence, it will suffice to show that the supremum $\\sup_{g \\in { \\mathcal{F}_{\\mathrm{mon}} }} \\Lambda(g)$ is realised by some $f$, which in turn will follow once we show that ${ \\mathcal{F}_{\\mathrm{mon}} }$ is compact.\n\n\tCompactness of ${ \\mathcal{F}_{\\mathrm{mon}} }$ is direct consequence of the classical Helly's selection theorem, see e.g. \\cite{Brunk-1956} for details.\t For a direct proof, consider any sequence $f_n \\in { \\mathcal{F}_{\\mathrm{mon}} }$. Passing to a subsequence, we may assume that $f_n$ converges pointwise on $\\mathbb{Q} \\cap [0,1]$. By motonicity, $f_n$ converges pointwise a.e. to some function $f$. Thus, by the dominated convergence, $f_n$ converges in $L^1$. It is clear that $f \\in { \\mathcal{F}_{\\mathrm{mon}} }$.\n\\end{proof}\n\nOnce we know that there is $f \\in { \\mathcal{F}_{\\mathrm{mon}} }$ which maximises $\\Lambda$, we may study such $f$ more closely. It comes as no surprise that the behaviour of $\\Lambda(f)$ under small distortions is relevant.\n\n\\begin{lemma}\\label{EXTup:lem:delta-Lambda}\n\tLet $f \\in \\mathcal{F}$, and suppose that $h \\in L^\\infty([0,1])$ is such that $f+\\tau h \\in \\mathcal{F}$ for sufficiently small $\\tau > 0$.\n\t For $\\tau > 0$, denote:\n\\begin{align}\n\t\\Delta_\\tau \\Lambda(f) &:= \\Lambda\\bra{f+\\tau h} - \\Lambda\\bra{f},\\\\\n\t\\delta_h \\Lambda(f) &:= \\lim_{\\tau \\to 0} \\frac{1}{\\tau} \\Delta_\\tau \\Lambda(f).\n\\end{align}\n\n\tWe then have the formula:\n\t\\begin{align*}\n\t\t\\delta_h \\Lambda(f) = \\int_{0}^1 h(x) w(x) dx, \\ \\text{where } \\ \n\t\tw(x) := \n\t\\begin{cases} \n\t\t\\int_{0}^x \\frac{du}{f(v(u))} & \\text{ if } x \\leq \\kappa, \\\\\n\t\t\\int_{x}^1 \\frac{dv}{f(u(v))} & \\text{ if } x \\geq \\kappa,\n\t\\end{cases}\n\t\\end{align*}\t\n\twhere $\\kappa=\\kappa_f$ is such that $\\int_{0}^\\kappa f(x) dx = \\int_{\\kappa}^1 f(x) dx = \\frac{1}{4}$.\n\\end{lemma}\n\n\\begin{proof}\nWe may assume without loss of generality that $\\norm{h}_\\infty \\leq 1$.\nFollowing the convention suggested above, for $\\tau > 0$ we denote:\n\\begin{align*}\n\t(\\Delta_\\tau u_f)(x) &:= u_{f+\\tau h}(x) - u_{f}(x),\\\n\t&(\\Delta_\\tau v_f)(x) &:= v_{f+\\tau h}(x) - v_{f}(x),\n\t\\\\\n\t(\\delta_h u_f)(x) &:= \\lim_{\\tau \\to 0} \\frac{1}{\\tau} (\\Delta_\\tau u_f)(x), \t\n\t&(\\delta_h v_f)(x) &:= \\lim_{\\tau \\to 0} \\frac{1}{\\tau} (\\Delta_\\tau v_f)(x).\n\\end{align*}\nSince $f$ is fixed, we will suppress dependence on $f$, writing $\\Delta_\\tau u$, $\\Delta_\\tau v$ and $\\delta_h u$ and $\\delta_h v$ whenever ambiguity does not arise.\n\nWe have a trivial estimate $\\abs{\\Delta_\\tau v (u)} \\leq 2 \\sqrt{ \\tau}$ for each $u$, which follows directly from the chain of inequalities\n$$\\tau \\norm{h}_{1} \\geq \\abs{\n\\int_{v(u)}^{v(u)+\\Delta_\\tau v(u)} (f-\\tau h)(x) dx }\n\\geq \\frac{(\\Delta_\\tau v(u))^2}{2} - \\tau \\norm{h}_{1}.\n$$\n(Here and elsewhere, we use the convention that $\\int_a^b \\equiv -\\int_b^a$ if $a > b$).\n\nIn the same way, we have $\\abs{\\Delta_\\tau u (v)} \\leq 2 \\sqrt{ \\tau}$ for each $v$. \n\nNow, fix $\\mu < \\kappa$. If $\\tau$ is small enough, then for $0 \\leq u \\leq \\mu$ we have that $v(u) + \\Delta_\\tau v(u) < 1$. For $u < \\mu$ we have, slightly more precisely than above:\n\\begin{align*}\n\t\\frac{1}{2} &= \\int_{u}^{v(u) + \\Delta_\\tau v(u)} (f + \\tau h)(x) dx \n\t\\\\ &=\n\t\\frac{1}{2} + \\tau \\int_u^{v(u)} h(x) dx + \n\t \\int_{v(u)}^{v(u) + \\Delta_\\tau v(u)} f(x) dx + O\\bra{ \\tau^{3/2}}.\n\\end{align*}\nHence\n\\begin{align*}\n \\frac{1}{\\Delta_\\tau v(u)} \\int_{v(u)}^{v(u) + \\Delta_\\tau v(u)} f(x) dx\n &= - \\frac{\\tau}{\\Delta_\\tau v(u)} \\bra{\n \\int_u^{v(u)} h(x) dx + O\\bra{\\sqrt{\\tau}}\n }.\n\\end{align*}\n\tFor a.e. $u$, the expression on the left hand side tends to $f(v(u))$ as $\\tau \\to 0$, by the Lebesgue density theorem. Hence, for a.e. $u$, $\\delta_h v(u)$ is well defined and:\n\t\\begin{align}\n\t\t\\delta_h v(u) = \\lim_{\\tau \\to 0} \\frac{\\Delta_\\tau v(u)}{\\tau} = -\\frac{\\int_u^{v(u)} h(x) dx }{f(v(u))}. \\label{EXTup:eq:54}\n\t\\end{align}\n\nBy a symmetric argument, if we fix $\\kappa < \\nu \\leq 1$ then for a.e. $v > \\nu$ we have:\n\t\\begin{align}\n\t\t\\delta_h u(v) = \\frac{\\int_{u(v)}^{v} h(x) dx }{f(u(v))}. \\label{EXTup:eq:55}\n\t\\end{align}\n\nFix $0 < \\mu < \\kappa $ and put $\\nu = v(\\mu)$. Using Observation \\ref{EXTup:obs:prelim} we have:\n$$\n\t\\Lambda(f) = \\int_0^{\\mu} (1-v(u)) du + \\int_{\\nu}^1 u(v) dv - \\mu (1-\\nu).\n$$\n\nNote that for any $\\varepsilon > 0$ and any sufficiently small $\\tau$ we have $\\Delta_\\tau u(v) \\leq \\frac{2\\tau \\norm{h}_1}{ f( \\nu + \\varepsilon)}$ for $u \\leq \\mu$, and hence $\\frac{ \\Delta_\\tau v(u)}{\\tau} $ is uniformly bounded. Likewise, $\\frac{ \\Delta_\\tau u(v)}{\\tau} $ if uniformly bounded for $v \\geq \\nu$. We may now compute:\n\\begin{align*}\n\t\\delta_h \\Lambda(f) &= \\lim_{\\tau \\to 0} \\frac 1\\tau \\Delta_\\tau \\Lambda(f) \n\t\\\\&= \\lim_{\\tau \\to 0} \n\t \\int_0^{\\mu} - \\frac{ \\Delta_\\tau v(u)}{\\tau} du + \\int_{\\nu}^1 \\frac{ \\Delta_\\tau u(v)}{\\tau} dv + \\mu \\frac{ \\Delta_\\tau v(\\mu) }{\\tau} + O(\\tau)\n\t\\\\&= -\\int_0^{\\mu} \\delta_h v(u) du + \\int_{\\nu}^1 \\delta_h u(v) dv + \\mu \\delta_h v(\\mu)\n\n\n\\end{align*}\nwhere the last equality uses the dominated convergence theorem. Note that this holds for any $0 < \\mu < \\kappa$. Passing to the limit $\\mu \\to 0$ or $\\mu \\to \\kappa$ we find simpler expressions:\n\\begin{align}\n\t\\delta_h \\Lambda(f) \n\t&= -\\int_0^{\\kappa} \\delta_h v(u) du = \\int_{\\kappa}^1 \\delta_h u(v) dv \\label{EXTup:eq:56}\n\\end{align}\n\nInserting \\eqref{EXTup:eq:54} into the first equation of \\eqref{EXTup:eq:56} and exchanging the order of integration, we now find:\n\\begin{align*}\n\t\\delta_h \\Lambda(f) &= \\int_0^\\kappa \\frac{\\int_u^{v(u)} h(x) dx }{f(v(u))} du\n\t= \\int_{0}^{1} h(x) \\int_{u(x)}^{\\min(x,\\kappa)} \\frac{du}{f(v(u))}dx.\n\\end{align*}\n\tLet $w(x)$ denote the value of the inner integral $\\int_{u(x)}^x \\frac{du}{f(v(u))}$. If $x \\leq \\kappa$ then $u(x) = 0$ so we obtain the sought formula $w(x) = \\int_{0}^x \\frac{du}{f(v(u))}$. Else, the formula follows from a change of variables $v = v( u)$ together with the observation that $\\frac{dv}{du} = \\frac{f(u)}{f(v)}$.\n\n\n\\end{proof}\n\n\tUsing standard techniques, we can extract from the above Lemma strong structural information about the function $f$ minimising $\\Lambda$. The proof is complicated by the fact that we need to account for a variety of pathological behaviour that $f$ may potentially exhibit. However, the key idea is simply to relate to each undesirable behaviour of $f$ a perturbation which increases $\\Lambda(f)$.\n\n\\newcommand{\\operatorname{sgn}}{\\operatorname{sgn}}\n\tFor $t \\in \\mathbb{R}$, we let $\\operatorname{sgn} t$ denote the sign of $t$, with the convention that $\\operatorname{sgn} t = +1$ for $t > 0$, $\\operatorname{sgn} t = -1$ for $t < 0$ and $\\operatorname{sgn} 0 = 0$. In particular, $f$ is continuous.\n\n\\begin{lemma}\\label{EXTup:lem:constraints}\n\tSuppose that $f \\in { \\mathcal{F}_{\\mathrm{mon}} }$ is such that $\\Lambda(f) = \\sup_{g \\in \\mathcal{F}} \\Lambda(g)$, and let $x, y \\in [0,1]$. Then:\n\t\\begin{enumerate}\n\t\t\\item\\label{EXTup:cond:11mono@lem:cons} For each $x,y$ we have $\\operatorname{sgn}( f(x) - f(y) ) = \\operatorname{sgn} ( w(x) - w(y) )$.\n\t\t\\item\\label{EXTup:cond:12cont@lem:cons} The function $f$ is continuous (except perhaps at $\\kappa_f$).\n\t\t\\item\\label{EXTup:cond:13meas@lem:cons} For each $E \\subset [0,1]$, measurable, we have $\\abs{f^{-1}(E)} = \\abs{E}$.\n\t\\end{enumerate}\t \n\\end{lemma}\n\\begin{proof}\n\n\tWe separate the proof into several steps.\n\t\n\t\\textit{Step 1.} If $f(x) > f(y)$ then $w(x) \\geq w(y)$.\n\t\n\tSuppose otherwise, so that $w(x) < w(y)$. Since $w$ is continuous, there is some $\\varepsilon > 0$ such that $w(x') < w(y')$ when $\\abs{x'-x}, \\abs{y'-y} < \\varepsilon$. Suppose for concreteness that $x < \\kappa_f < y$ (the remaining cases being fully analogous), so that $f$ is increasing in a neighbourhood of $x$ and decreasing in a neighbourhood of $y$.\n\t\n\tWe wish to consider a function $\\tilde f$ obtained by ``swapping'' the intervals $[x,x+\\varepsilon]$ and $[y-\\varepsilon,y]$. Hence, we define \n\t\\begin{align*}\n\t\t\\tilde f(x + t) &= f(y-\\varepsilon + t) & t \\in [0,\\varepsilon]& \\\\\n\t\t\\tilde f(y-\\varepsilon + t) &= f(x + t) &t \\in [0,\\varepsilon]& \\\\\n\t\t\\tilde f(x) &= f(x) \\qquad & x \\not \\in [x,x+\\varepsilon] \\cup [y-\\varepsilon,y]& \n\t\\end{align*}\n\t\n\tIt is clear that $\\tilde f \\in \\mathcal{F}$, since membership in $\\mathcal{F}$ is ``invariant under rearrangement'', in the sense that the condition $g \\in \\mathcal{F}$ can be phrased purely in terms of the values $\\abs{g^{-1}(E)}$ for $E \\subset [0,1]$, measurable.\n\t\n\tPut $h = \\tilde f - f$, so that $f + \\tau h \\in \\mathcal{F}$ for $t \\in [0,1]$ because of convexity. Using Lemma \\ref{EXTup:lem:delta-Lambda} (and the notation therein) we have:\n\t\\begin{align*}\n\t\\delta_h \\Lambda(f) &= \\int_0^1 h(z)w(z) dz\n\t\\\\&= \\int_0^\\varepsilon (w(x + t) - w(y-\\varepsilon+t)) ( f(y-\\varepsilon+t) - f(x+t)) dt > 0.\n\t\\end{align*}\n\tHence, for sufficiently small $\\tau$ we conclude that $\\Lambda( f + \\tau h) > \\Lambda(f)$, contradicting the choice of $f$.\n\t\n\t\\textit{Step 2.} There is no interval where $f$ is constant.\n\t\n\tSuppose otherwise, so that there exists some $t_0$ with $U := f^{-1}(\\{t_0\\})$ with positive measure. Then, we claim that for any test function $h$ such that $\\operatorname{supp} h \\subset U$, $\\norm{h}_{\\infty} \\leq 1$, $\\int_0^1 h(x) dx = 0 $, and for sufficiently small $\\tau > 0$ we have $f + \\tau h \\in \\mathcal{F}$. (Recall that $\\operatorname{supp} h = \\{ x \\ : \\ h(x) \\neq 0 \\}$.) The fact that $U$ will generally consist of two disjoint intervals contained in $[0,\\kappa_f]$ and $[\\kappa_f,1]$ respectively complicates the notation slightly. Instead, we argue a slightly simpler statement, and leave it to the interested reader to work out the details.\n\t\n\t\\textit{Claim.} Pick two points $0 \\leq x_0 < x_1 \\leq 1$. Let $G(x) = \\frac{x^2}{2}$ and let $F(x) = \\frac{x_0+x_1}{2} x - \\frac{x_0x_1}{2}$ (so that $F(x) = G(x)$ for $x = x_0,x_1$). Also, fix a differentialbe function $H$ with $\\operatorname{supp} H \\subset [x_0,x_1]$ and $\\norm{H'}_{\\infty} \\leq 1$. Then, for sufficiently small $\\tau$, for any $x \\in [x_0,x_1]$ we have $F(x) + \\tau H(x) \\geq G(x)$. \n\t\n\t\\textit{Proof of Claim.} Taking $\\tau$ sufficiently small, we may ensure that $F'(x) + \\tau H'(x) > G'(x)$ for $x_0 \\leq x \\leq x_0 + \\frac{x_1-x_0}{10}$ and $F'(x) + \\tau H'(x) < G'(x)$ for $ \\leq x_1 - \\frac{x_1-x_0}{10} \\leq x \\leq x_1$. Hence $F(x) + \\tau H(x) \\geq G(x)$ when $x$ is within $ \\frac{x_1-x_0}{10} $ of $x_0,x_1$. Since $F(x) - G(x)$ is bounded away from $0$ on for $x_0 + \\frac{x_1-x_0}{10} \\leq x \\leq x_1 - \\frac{x_1-x_0}{10}$ we can ensure $F(x) + \\tau H(x) \\geq G(x)$ for such $x$ by choosing sufficiently small $\\tau$. \n\t\n\tWe are now ready to derive a contradiction. Since for any fixed $h$ as above we have $f \\pm \\tau h \\in \\mathcal{F}$, we conclude that $\\delta_h \\Lambda(f) = 0$. Hence, by Lemma \\ref{EXTup:lem:delta-Lambda} we have $\\int_U h(z)w(z) dz = 0$. However, this is only possible if $w$ is not constant on $U$, which it clearly is not. Thus, $f$ is not constant on any interval.\n\t \t\n\\textit{Step 3.} For any $t > 0$, we have $\\inf_{\\abs{E}=t} \\int_{E} f(x) dx = \\frac{t^2}{2}$.\n\t\t\t\t\n\t Once this equality is proved, the condition \\eqref{EXTup:cond:13meas@lem:cons} follows. (Both are easily seen to be equivalent to the statement that the increasing rearrangement of $f$ is the map $\\tilde f(x) = x$.)\n\t \n\t Note that inequality in one direction follows directly from the fact that $f \\in \\mathcal{F}$. We just need to prove $\\inf_{\\abs{E}=t} \\int_{E} f(x) dx \\leq \\frac{t^2}{2}$. \n\t \n \tSuppose for the sake of contradiction that for some $t_0$ we have strict inequality $\\int_{E_0} f(x) dx > \\frac{1}{2}t_0^2$ where $E_0$ is chosen so that $\\abs{E_0} = t_0$ and the integral is minimised. From monotonicity of $f$ it follows that $E_0$ takes the form $E_0 = [0,x_0] \\cup [y_0,1]$ for some $0 \\leq x_0 \\leq \\kappa_f \\leq y_0 \\leq 1$, and because $f$ is nowhere constant, the choice of $x_0,y_0$ is unique.\n \t\n \tLet $t_1 > t_0$ be such that $\\int_{E_0} f(x) dx = \\frac{1}{2}t_1^2$, and let $E_1 = [0,x_1] \\cup [y_1,1]$ with $\\abs{E_1} = t_1$ minimise $\\int_{E_1} f(x) dx$. Take $F = E_1 \\setminus E_0 = [x_0,x_1] \\cup [y_1, y_0]$, and take a set $U$ such that $\\operatorname{cl} U \\subset \\operatorname{int} F$ and $\\abs{U} > 0$. Since for any $x \\in U$ and $x' \\in E_0$ we have $f(x) > f(x')$, it is not difficult to show that for any $h$ with $\\operatorname{supp} h \\subset U$, $\\norm{h}_\\infty \\leq 1$, $\\int_0^1 h(x)dx = 0$, for sufficiently small $\\tau$ we have $f + \\tau h \\in \\mathcal{F}$.\n \t\n \tBy the same argument as in Step 2., this means that $\\int_0^1 h(x) w(x)dx = 0$. Since $w$ is non-constant and $h$ is arbitrary, this is the sought contradiction.\n \t\n\\textit{Step 4.} The function $f$ is continuous (expect possibly at $\\kappa_f$).\n\t\n\tNote that the only way $f$ could be discontinuous is if it had a jump discontinuity at some point $x$. Suppose that this is the case, and assume for concreteness that $x < \\kappa_f$ so that $f$ is increasing at $x$. Since by Step 3., the image of $f$ is dense in $[0, 1]$, we may find some $y$ where $f$ is continuous such that $\\lim_{z \\to x-} f(z) < f(y) < \\lim_{z \\to x+} f(z)$. Assumption that $x < \\kappa_f$ forces $y > \\kappa_f$, so for $y' < y$, sufficiently close, we have $w(y') > w(y)$. Hence, using Step 1. we have $w(x) \\leq w(y) < w(y') \\leq w(x)$, which is a contradiction.\n\t\n\\textit{Step 5.} If $f(x) = f(y)$ then $w(x) = w(y)$.\n\nAssume for concreteness that $x < \\kappa_f$. Using Steps 1. and 2. we have $w(x) = \\lim_{z \\to x} w(z) \\leq w(y)$, and symmetrically $w(x) \\geq w(y)$.\n\nSteps 5. and 1. together imply condition \\eqref{EXTup:cond:11mono@lem:cons}, hence the proof is complete.\n\n\\end{proof}\n\n\tNote that modifying a function $f \\in \\mathcal{F}$ at a single point does not affect $\\Lambda(f)$. Hence, the above Lemma \\ref{EXTup:lem:constraints} implies in particular that $\\Lambda$ is maximised by a continuous, nowhere constant function $f \\in { \\mathcal{F}_{\\mathrm{mon}} }$.\n\n\tFor continuous $f \\in { \\mathcal{F}_{\\mathrm{mon}} }$ which are not constant on any interval, we introduce local inverse functions $\\alpha_f \\colon [0,1] \\to [0,\\kappa_f]$ and $\\beta_f\\colon [0,1] \\to [\\kappa_f,1]$ so that \n\t$$f(\\a_f(t)) = f(\\b_f(t)) = t.$$\n\tWhenever possible, we will suppress dependence on $f$, writing simply $\\a$ and $\\b$. We have the following, somewhat unexpected, relation.\n\\newcommand{\\tilde}{\\tilde}\n\n\\begin{lemma}\\label{EXTup:obs:at+bt=cons}\n\tSuppose that $f \\in { \\mathcal{F}_{\\mathrm{mon}} }$ is such that $\\Lambda(f) = \\sup_{g \\in \\mathcal{F}} \\Lambda(g)$ and that $f$ is continuous. Then, for any $t \\in [0,1]$ we have $v \\circ \\a(t) + u \\circ \\b(t) = 1$. In particular, $\\kappa = \\frac{1}{2}$.\n\t\n\\end{lemma}\n\\begin{proof}\n\tThe claim is clearly true for $t = 1$, and the value of $\\kappa$ follows from the remaining part of the statement by taking $t = 0$.\n\t\n\tIt is a direct consequence of Lemma \\ref{EXTup:lem:constraints} that $\\a$ and $\\b$ are Lipschitz continuous, with Lipschitz constant at most $1$. Moreover, $u$ and $v$ are Lipschitz continuous on $[0,1] \\setminus (\\k - \\varepsilon,\\k+\\varepsilon)$ for any $\\varepsilon > 0$ (with the Lipschitz constant dependent on $\\varepsilon$). Thus, $v \\circ \\a + u \\circ \\b$ is Lipschitz continuous on $[0,1 -\\varepsilon]$ for any $\\varepsilon$. In particular, $v \\circ \\a + u \\circ \\b$ is absolutely continuous on $[0,1 -\\varepsilon]$, and to prove that it is constant it will suffice to show that $(v \\circ \\a + u \\circ \\b)'(t)=0$ for almost all $t$. Passing to the limit $\\varepsilon \\to 0$ will then complete the proof.\n\n\tA simple computation yields:\n\t\\begin{equation}\n\t\t(u \\circ \\b)'(t) = \\b'(t) \\frac{ f \\circ \\b(t) }{f \\circ u \\circ \\b(t)} = \\frac{t \\b'(t) }{f \\circ u \\circ \\b(t)}, \\qquad\n\t\t(v \\circ \\a)'(t) = - \\frac{t \\a'(t) }{f\\circ v \\circ \\a(t)} \\label{EXTup:eq:43-a}\n\t\\end{equation}\n\talmost everywhere (where $\\a'(t)$ and $\\b'(t)$ are defined).\t\n\t\n\tAnother application of Lemma \\ref{EXTup:lem:constraints} implies that $w(\\a(t)) = w(\\b(t))$ (with $w$ defined as in Lemma \\ref{EXTup:lem:delta-Lambda}). Differentiating this equality, we conclude that:\n\t\\begin{equation}\n\t\t\\frac{\\b'(t)}{f \\circ u \\circ \\b(t)} = \\frac{\\a'(t)}{f\\circ v \\circ \\a(t)} = 0.\n\t\\label{EXTup:eq:43-b}\n\t\\end{equation}\n\n\tCombining \\eqref{EXTup:eq:43-a} and \\eqref{EXTup:eq:43-b} we conclude that indeed $(v \\circ \\a + u \\circ \\b)'(t)=0$ a.e., which finishes the proof.\n\n\\end{proof}\n\n\tWe are now ready to prove the final bit of information we need about the function maximising $\\Lambda$, namely the symmetry.\n\t\n\\begin{observation}\\label{EXTup:obs:symmetry}\n\tSuppose that $f \\in { \\mathcal{F}_{\\mathrm{mon}} }$, continuous, is such that $\\Lambda(f) = \\sup_{g \\in \\mathcal{F}} \\Lambda(g)$. \n\n\tThen any $s, t \\in [0,1]$, if $\\a(s) = {u \\circ \\b}(t)$ then also $\\a(t) = {u \\circ \\b}(s)$. In particular, the function $f$ is symmetric: $f(x) = f(1-x)$ for all $x$. \n\\end{observation}\n\\begin{proof}\n\t\t\n\tWe begin by proving the symmetry of $f$, assuming the former part of the claim. It will be enough to show that for any $t$, $\\a(t) + \\b(t) = 1$. Take any $t$, and let $s$ be such that $\\a(s) = {u \\circ \\b}(t)$. By assumption, $\\a(t) = {u \\circ \\b}(s)$. We have ${v \\circ \\a}(s) = v \\circ u \\circ \\b(t) = \\b(t)$. Hence, $\\a(t) + \\b(t) = {u \\circ \\b}(s) + {v \\circ \\a}(s) = 1$.\n\t\n\tFor the remaining part of the argument, it will be convenient to define a pair of transformations $T_\\a$ and $T_\\b$ on $[0,1]$ given by $T_\\a = f \\circ {u \\circ \\b}$ and $T_\\b = f \\circ {v \\circ \\a}$. We note several properties of these transformations.\n\t\\begin{enumerate}\n\t\\item\\label{EXTup:cond:T1} $T_\\a \\circ T_\\b (t) = t$ and $T_\\b \\circ T_\\a (t) = t$ for any $t$;\n\t\\item\\label{EXTup:cond:T2} $\\a(T_\\a(t)) = {u \\circ \\b}(t)$ and $\\b(t) = {v \\circ \\a}(T_\\a(t))$ for any $t$;\n\t\\item\\label{EXTup:cond:T3} $\\abs{ \\a(T_\\a^2(t)) - \\a(T_\\a^2(t')) } = \\abs{ \\b(t) - \\b(t')}$ for any $t,t'$;\n\t\\item\\label{EXTup:cond:T4} $T_\\a(t) > T_\\a(t')$ and $T_\\b(t) > T_\\b(t')$ for any $t < t'$.\n\t\\end{enumerate}\n\tAssertions \\eqref{EXTup:cond:T1} and \\eqref{EXTup:cond:T2} follow directly by substitution. For example, we have:\n\t$$\n\t\tT_\\a \\circ T_\\b (t) = f \\circ {u \\circ \\b} \\circ f \\circ {v \\circ \\a}(t)\n\t\t= f \\circ {u} \\circ {v \\circ \\a}(t) = f \\circ{ \\a} (t) = t,\n\t$$\n\twhere we use that $u \\circ v(x) = x$ and $\\beta \\circ f(x) = x$ in appropriate ranges of $x$. The remaining equalities follow along similar lines.\n\t\n\tAssertion \\eqref{EXTup:cond:T4} follows from known monotonicity properties of $f$, $\\a, \\b$ and $u, v$. For instance, if $t < t'$ then $\\b(t) > \\b(t')$, hence $u \\circ \\beta(t) > u \\circ \\beta(t')$ and $f\\circ u \\circ \\beta(t) > f \\circ u \\circ \\beta(t')$ (note that $f$ is increasing in the relevant range).\n\t\n\tAssertion \\eqref{EXTup:cond:T3} is the least obvious and the most crucial. Using Observation \\ref{EXTup:obs:at+bt=cons} and previously established properties of $T_\\a$ we have:\n\t\\begin{align*}\n\t\\abs{ \\a(T_\\a^2(t)) - \\a(T_\\a^2(t')) } &=\n\t\\abs{ u \\circ \\b (T_\\a(t)) - u \\circ \\b (T_\\a(t')) } \n\t\\\\ &= \\abs{ v \\circ \\a (T_\\a(t)) - v \\circ \\a (T_\\a(t')) }\n\t = \\abs{ \\b (t) - \\b(t') }.\n\t\\end{align*}\t\n\t\n\tOur main claim is equivalent to the statement that for each $t$, $T_\\a^2(t) = t$, where we may restrict our attention to $t$ with $T_\\a(t) < t$. For the sake of contradiction, suppose that for some $t_0$ we have $T_\\a^2(t_0) \\neq t_0$. For concreteness, we may suppose that $T_\\a^2(t_0) > t_0$, the other case being fully analogous. \n\t\n\tLet us consider the consecutive iterates $t_n := T_\\a^n(t_0)$ for $n \\in \\mathbb{Z}$ (for $n < 0$ we use $T_\\beta = T_\\a^{-1}$). By \\eqref{EXTup:cond:T4}, it is clear that $t_{2n}$ is monotonously increasing, while $t_{2n+1}$ is monotonously decreasing. Moreover, we may define:\n\t\\begin{align*}\n\t\tI_n &:= { \\int_{\\a(t_n)}^{\\a(t_{n+2})} f(x) dx } = { \\int_{\\b(t_{n-1})}^{\\b(t_{n+1})} f(x) dx }\\\\\n\t\\end{align*}\nwhere the two integrals are equal because by \\eqref{EXTup:cond:T2} we have\n\\begin{align*}\n\t0 &= \\int_{\\a(t_n)}^{v\\circ \\a(t_n)}f(x)dx - \\int_{\\a(t_{n+2})}^{v\\circ \\a(t_{n+2})} f(x)dx \n\t\\\\&= \\int_{\\a(t_n)}^{\\a(t_{n+2})}f(x)dx - \\int_{\\b(t_{n-1})}^{\\b(t_{n+1})}f(x)dx.\n\\end{align*}\n\nIn similar spirit, we may define\n\t\\begin{align*}\n\t\tl_n &:= {\\a(t_{n+2}) - \\a(t_{n}) } = {\\b(t_{n-2}) - \\b(t_{n}) }\n\t\\end{align*}\nwhere the two quantities are equal by a direct application of \\eqref{EXTup:cond:T3}.\n\n\tUsing monotonicity of $f$ and $t_n$, we find that for any $n \\in \\mathbb{Z}$ we have \n\t\\begin{align*}\n\tl_{2n} t_{2n+2} &\\geq I_{2n} \\geq l_{2n} t_{2n},\n\t&l_{2n+1} t_{2n+3} &\\leq I_{2n+1} \\leq l_{2n+1} t_{2n+1}, \\\\\n\tl_{2n+1} t_{2n-1} &\\geq I_{2n} \\geq l_{2n+1} t_{2n+1},\n\t&l_{2n+2} t_{2n} &\\leq I_{2n+2} \\leq l_{2n+2} t_{2n+2} .\n\t\\end{align*}\n\t\n\tCombining the above inequalities, we find\n\t$$\n\t\t\\frac{l_{2n}}{l_{2n+2}} \n\t\t= \\frac{l_{2n}}{l_{2n+1}} \\cdot \\frac{l_{2n+1}}{l_{2n+2}}\n\t\t\\geq \\frac{t_{2n+1}}{t_{2n+2}} \n\t\t\t\\cdot \\frac{t_{2n}}{t_{2n+1}}\n\t\t= \\frac{t_{2n}}{t_{2n+2}}.\n\t$$\t\n\t\n\tTaking the product as $n$ ranges over $-N+1,\\dots,0$ and letting $N \\to \\infty$ we conclude that \n\t$$\\liminf_{N\\to \\infty} l_{-2N} \\geq \\frac{ l_0}{t_0} \\lim_{N\\to \\infty} t_{-2N}. $$\n\tThe limit on the right hand side is defined, because the sequence $t_{-2N}$ is decreasing. \n\n\tWe claim that $\\lim_{N\\to \\infty} t_{-2N} > 0$. Clearly, we have $T_\\a(t_0) = t_1 < t_0$ and $T_\\a(t_1) = t_2 > t_1$. Thus, there is some $t_1 < t_* < t_0$ such that $T_\\a(t_*) = t_*$. By monotonicity of $T_\\a$ we have that for any $t$ that $t > t_*$ if and only if $T_\\a(t) < t_*$, which is further equivalent to $T_\\a^2(t) > t_*$. Hence, $t_{2n} > t_*$ for all $n \\in \\mathbb{Z}$, and in particular $\\lim_{N\\to \\infty} t_{-2N} \\geq t_* > 0$.\n\t\n\tIt follows that $l_{-2N}$ converges to a non-zero limit as $N \\to \\infty$. On the other hand, by construction of $l_n$, the interval $[0,1]$ contains a disjoint union of intervals with lengths $l_n$, $n \\in \\mathbb{Z}$. In particular, $\\sum_{n \\in \\mathbb{Z}} l_n < \\infty$, contradicting the above statement.\n\\end{proof}\n\n\\begin{corollary}\n\tThe unique (up to a.e. equality) function $f \\in { \\mathcal{F}_{\\mathrm{mon}} }$ maximising $\\Lambda$ is given by:\n\t$$\n\t\tf(x) = \n\t\t\\begin{cases}\n\t\t\t2x & \\text{ if } x \\leq \\frac{1}{2}\\\\\n\t\t\t2(1-x) & \\text{ if } x \\geq \\frac{1}{2}.\n\t\t\\end{cases}\n\t$$\n\\end{corollary}\n\nWe are now in position to finish the proof of Proposition \\ref{EXTup:prop:L(f)-bound}. Let $f$ be the function defined above. The region ${L}(f)$ can be described explicitly: \n$${L}(f) = \\bbra{ (x, y) \\in [0,1]^2 \\ : \\ x \\leq y, \\ x^2 + (1-y)^2 \\leq \\frac{1}{4} }.$$ \nThis gives $\\Lambda(f) = \\frac{\\pi}{16}$, which is precisely the needed bound.\n\n\\section{Closing remarks}\\label{section:END}\n\nIn the previous sections, we have obtained a fairly satisfactory understanding the $\\abs{S(a)}$ in for a random permutation, as well as in the ``best case scenario'' where $a$ is chosen to maximize $\\abs{S(a)}$. It is natural to ask about the behaviour of $\\abs{S(a)}$ in the ``worst case scenario'', when $a$ is chosen to minimize $\\abs{S(a)}$. We now address this problem, but we ask more questions than we answer.\n\nThe best lower bound we are aware of can be obtained by an argument in \\cite{Solymosi-2005} (also present in \\cite{Balog-2014}), a variant of which we sketch below for the convenience of the reader.\n\n\n\\begin{proposition}\n\tFor any permutation $a$ of $[n]$ we have $\\abs{S(a)} \\geq \\frac{n^{3/2}}{4\\sqrt{2}}$.\n\\end{proposition}\n\n\\begin{proof}\n\tFor any integer $k$, consider the set $S_k(a)$ of the sums $s \\in S(a)$ with $kn+1 \\leq s \\leq (k+1)n$. Clearly, $S(a) = \\bigcup_{k \\geq 0} S_k(a)$, and the union is disjoint.\n\t\n\tTake any $1 \\leq k \\leq \\frac{n+1}{4}$. For any $u$, at least one of the sums $\\sum_{i=u}^n a_i$ or $\\sum_{i=1}^u a_i$ exceeds $ \\frac{1}{2} \\binom{n+1}{2} \\geq k n $. For concreteness, suppose $\\sum_{i=u}^n a_i > kn$ and let $v$ be the smallest integer such that $\\sum_{i=u}^v a_i > k n$. \n\t\n\tBy the choice of $v$ we have $\\sum_{i=u}^v a_i \\in S_{k-1}(a)$ and $\\sum_{i=u+1}^v a_i \\in S_{k}(a)\\cup S_{k-1}(a)$. Hence, $a_u \\in S_k(a) - (S_k(a) \\cup S_{k-1}(a))$. Since $u$ was chosen arbitrarily, we conclude that \n\t$ S_k(a) - (S_k(a) \\cup S_{k-1}(a)) \\supset [n],$\n\twhich implies that $$\\abs{S_k(a)} + \\abs{S_{k-1}(a)} \\geq \\sqrt{2n}.$$\n\t\n\tSumming over $1 \\leq k \\leq \\frac{n+1}{4}$ and using $\\abs{S_0(a)} = n \\geq \\sqrt{2n} $ we conclude that:\n\t$$\\abs{S(a)} \\geq \\frac{1}{2} \\sum_{k=0}^{\\floor{\\frac{n+1}4}} (\\abs{S_k(a)} + \\abs{S_{k-1}(a)}) \\geq \\frac{1}{2} \\floor{\\frac{n+5}{4}} \\sqrt{2n} \\geq \\frac{n^{3/2}}{4\\sqrt{2}}.$$\n\\end{proof}\n\nThis bound is rather far from what one would expect. The trivial permutation $a_i = i$ is essentially the only known example with $\\abs{S(a)} = o(n^2)$. In this case, we have $\\abs{S(a)} \\sim n^{2-o(1)} $.\n\nSlightly more generally, we have a similar result for permutations of ``bounded complexity''. Because the result is rather standard, we only sketch the argument.\n\n\\begin{example}\\label{LOW:prop:small-S}\n\tFix an integer $M$. Let $a$ be a permutation of $[n]$ of complexity at most $M$, by which we mean that there is a partition $[n] = \\bigcup_{j=1}^{M} I_j$ into intervals such that for each $j \\in [M]$, $i \\in I_j$ we have $a_i = i b_j + c_j$, where $b_j,c_j$ are some constants and $\\abs{b_j} \\leq M$.\n\t\n\tThen $\\abs{S(a)} = o(n^2)$ but for any $\\delta > 0$ also $\\abs{S(a)} = \\Omega(n^{2-\\delta})$ (the implicit rates of convergence are allowed to depend on $M$ and $\\delta$).\n\\end{example}\n\\begin{proof}[Sketch of the proof]\n\tFor the lower bound, it is enough to observe that one of the intervals $I_j$ has large length $\\abs{I_j} \\geq n/M$. For $u,v \\in I_j$, $S(a)$ contains the sum\n\t$$ s = \\sum_{i=u}^{v} a_i = (v-u+1)\\bra{c_j + \\frac{u+v}2 b_j}.$$ \n\tThere are at least $\\binom{n/M}2$ choices of $u,v$, and each sum $s$ is represented in at most $d(s) \\ll n^{\\delta}$ possible ways. Hence, $\\abs{S(a)} = \\Omega(n^{2-\\delta})$.\n\t\n\tFor the upper bound, it suffices to show that for any $j,k \\in [M]$ the number of sums $\\sum_{i=u}^v a_i$ with $u \\in I_j,\\ v \\in I_k$ is $o(n^2)$. If we fix $j,k$, and let $u \\in I_j,\\ v \\in I_k$ then\n\t $$\\sum_{i=u}^v a_i = Av^2 + Bu^2 + C x + D y + E,$$\nwhere $A = \\frac{b_k}{2}$, $B = - \\frac{b_j}{2}$. Hence, the problem reduces to showing that for the polynomial $P(u,v) = Av^2 + Bu^2 + C x + D y + E$ we have\n$$\n\t\\abs{\\bbra{ P(u,v) \\ : \\ u,v \\in [n] }} = o(n^2).\n$$\nIt is not difficult to reduce to the case $C = D = E = 0$ and $\\gcd(A,B) = 1$. There are not two cases to consider.\n\n\\textit{Case 1.} Suppose that $-AB$ is a square. Then $P(u,v)$ factors as $(A'u + B'v)(A'u -B'u)$ and the bound follows from the theorem of Erd\\H{o}s cited in the introduction.\n\n\\textit{Case 2.} Suppose that $-AB$ is not a square. Then, by Chebotarev's density theorem, there is a family $\\mathcal{P}$ of primes $p$ with positive relative density such that $-AB$ is not a square modulo $p$. For any $p \\in \\mathcal{P}$ we then have $$P(u,v) \\not \\equiv p,2p,\\dots,(p-1) p \\pmod{p^2},$$\nand thus \n\\begin{align*} \\lim_{n \\to \\infty} \\frac{ \\abs{\\bbra{ P(u,v) \\ : \\ u,v \\in [n] }} }{ n^2 } \\ll\n \\prod_{\\substack{ p \\in \\mathcal{P}}} \\bra{1-\\frac{p-1}{p^2}} = 0. & \\qedhere\n\\end{align*}\n\\end{proof}\n\nIn the above example, we could equally well allow the ``complexity'' tend to $\\infty$ with $n$, but do so sufficiently slowly.\n\n\n\nHence, both for random and highly structured permutations $a$, we have $\\abs{S(a)} \\sim n^{2-o(1)}$. In absence of plausible counterexamples, we pose the following question.\n\n\\begin{question}\n\tIs it the case that for any $\\delta > 0$, there exists a constant $c_\\delta > 0$ such that for any $n$, and for any permutation $a$ of $[n]$ one has $\\abs{S(a)} \\geq c_\\delta n^{2-\\delta}$?\n\\end{question}\n\nIn fact, all examples with $\\abs{S(a)} = o(n^2)$ we are aware of exhibit some algebraic structure, much as in Proposition \\ref{LOW:prop:small-S}. It is not quite the case that $\\abs{S(a)}$ is minimised for the trivial permutation, but none of the known examples seems to be significantly worse. Hence, we may ask more boldly:\n\n\\begin{question}\n\tDoes there exists an absolute constant $c$ such that for any $n$, and for any permutation $a$ of $[n]$ one has $\\abs{S(a)} \\geq c \\abs{S(\\mathrm{id}_n)}$?\n\t\n\t(Here, $\\mathrm{id}_n$ denotes the trivial permutation on $[n]$).\n\\end{question}\n\nIn similar spirit, we may also ask if the only way for $S(a)$ to be small is if $a$ has some algebraic structure. To give an indication of just how much structure one may hope to find, we give the following examples.\n\n\\begin{example}\n\tFix a constant $M$, and suppose that $M \\mid n$. Consider the permutation \n\t$$a = (1,\\frac{n}{M}+1,\\dots,(M-1)\\frac{n}{M}+1, 2,\\frac{n}{M}+2,\\dots,(M-1)\\frac{n}{M}+2,3,\\dots),$$ that is $a_i = \\ceil{\\frac{i}{M}} + \\frac{n}{M} (i-1 \\mod M)$. Then $\\abs{S(a)} = o(n^2)$ (where the decay rate may depend on $M$). \n\t\n\\end{example}\n\nNote, however, that a similar looking permutation of Proposition \\ref{I:prop:cexple-basic} has $\\Theta(n^2)$ sums.\n\n\\begin{example}\n\tTake any permutation $a$ with $\\abs{S(a)} = o(n^2)$, and let $m = o(n)$. Consider a permutation $b$ obtained from $a$ by choosing $m$ pairs of consecutive indices $i,i+1$ and swapping $a_i$ with $a_{i+1}$. More precisely, pick a set $I$ with $\\abs{I} = m$ and such that $i \\in I$ implies $i+1 \\not \\in I$, and define:\n$$\n\tb_i = \n\t\\begin{cases}\n\t\ta_{i+1} & \\text{ if } i \\in I,\\\\\n\t\ta_{i-1} & \\text{ if } i+1 \\in I,\\\\\n\t\ta_{i} & \\text{ otherwise}.\\\\\t\t\n\t\\end{cases}\n$$\nThen $\\abs{S(b)} = o(n^2)$.\n\\end{example}\n\nIn view of the above examples, one cannot hope to ensure that $a_i$ looks structured on any large structured piece of $[n]$, nor that it looks structured on an long interval. We can, however, hope that the following should have a positive answer.\n \n\\begin{question}\n\tDoes there exists $\\varepsilon > 0$ such that the following is true?\n\t\n\tLet $a$ be a permutation of $[n]$ with $\\abs{S(a)} \\leq \\varepsilon n^2$. Then there exists an index set $I \\subset [n]$ with $\\abs{I} = \\omega(1)$ as $n \\to \\infty$ and constants $b,c$ such that for $i \\in I$ we have $a_i = b i + c$.\n\\end{question}\n\n\nOne may also ask a similar questions in a more general context. Let $a$ be a permutation of a set $A \\subset \\mathbb{N}$ of size $\\abs{A} = n$, not necessarily equal to $[n]$. Let $S(a)$ be defined just as before, to be the set $\\bbra{ \\sum_{i=u}^v a_i \\ : \\ u,v \\in [n]}$. How small can $\\abs{S(a)}$ be?\n\nIt is perhaps more natural to phrase this question in different terms. For a set $B = \\{b_i \\ : \\ i \\in [m] \\}$ with $b_1 < b_2 < \\dots < b_m$, define (following the terminology of \\cite{Solymosi-2005} and \\cite{Balog-2014}) the set of gaps $D(B) = \\{ b_{i+1} - b_{i} \\ : \\ i \\in [m-1]\\}$. Note that setting $B = \\{ \\sum_{i=1}^{v} a_i \\ : \\ v \\in [n] \\} \\cup \\{0\\}$ we recover $S(a) = (B-B) \\cap \\mathbb{N}$ and $D(B) = \\{a_i \\ : \\ i \\in [n] \\} = A$. \n\n\\begin{question}\n\tSuppose that $B \\subset \\mathbb{N}$ with $\\abs{B} = n$ is such that $\\abs{D(B)} = n-1$. For which $\\delta > 0$ does there exist $c_\\delta > 0$ (independent of $B$) such that $\\abs{B-B} \\geq c_\\delta n^{2-\\delta}$?\n\\end{question}\n\nThis question is already alluded to in \\cite{Solymosi-2005}, and resolved positively in the case $\\delta = \\frac{1}{2}$. For $\\delta \\in (0,\\frac{1}{2})$ to the best of our knowledge, the answer is not known. \n\n\n\n\n\n\\nocite{Alon-Spencer,Rudin}\n\\bibliographystyle{plain}\n", "meta": {"timestamp": "2015-05-21T02:00:46", "yymm": "1504", "arxiv_id": "1504.07156", "language": "en", "url": "https://arxiv.org/abs/1504.07156", "abstract": "We study the number of values taken by the sums $\\sum_{i=u}^{v-1} a_i$, where $a_1,a_2,\\dots,a_n$ is a permutation of $1,2,\\dots,n$ and $1 \\leq u < v \\leq n+1$. In particular, we show that for a random choice of a permutation, with high probability there are $(\\frac{1+e^{-2}}{4} +o(1)) n^2$ such sums. This answers an old question of Erdős and Harzheim. We also obtain non-trivial bounds on the maximum possible number of distinct sums, ranging over all permutations of $1,2,\\dots,n$. We close with some questions concerning the minimal possible number of distinct sums.", "subjects": "Combinatorics (math.CO); Number Theory (math.NT)", "title": "On consecutive sums in permutations", "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.9845754510863375, "lm_q2_score": 0.8652240756264639, "lm_q1q2_score": 0.851878384550685}} {"url": "https://arxiv.org/abs/2101.09600", "title": "Extremizing Temperature Functions of Rods with Robin Boundary Conditions", "abstract": "We compare the solutions of two one-dimensional Poisson problems on an interval with Robin boundary conditions, one with given data, and one where the data has been symmetrized. When the Robin parameter is positive and the symmetrization is symmetric decreasing rearrangement, we prove that the solution to the symmetrized problem has larger increasing convex means. When the Robin parameter equals zero (so that we have Neumann boundary conditions) and the symmetrization is decreasing rearrangement, we similarly show that the solution to the symmetrized problem has larger convex means.", "text": "\\section{Introduction: Physical Motivation and Main Results}\nOur paper is motivated by the following physical problem:\n\n\\begin{problem}\\label{Prob:1dRod}\nConsider a metal rod of length $\\ell$. To half the locations on the rod, heat is generated uniformly, while on the remaining half of the rod, heat is neither generated nor absorbed. If the rod's ends are frozen at zero temperature, where should we place the heat sources to maximize the hottest steady-state temperature across the rod?\n\\end{problem}\nSeveral possible arrangements appear in Figure 1 below. Heat is generated in the white regions while heat is neither generated nor absorbed in the gray regions.\n\n\n\\begin{figure}[h]\n\\centering\n\\begin{subfigure}[t]{.4\\textwidth}\n\\centering\n\\includegraphics[width=\\linewidth]{Arrangement1_.pdf}\n \\caption{}\\label{fig:fig_a}\n\\end{subfigure}\n\\begin{subfigure}[t]{.4\\textwidth}\n\\centering\n\\includegraphics[width=\\linewidth]{Arrangement2_.pdf}\n\\caption{}\\label{fig:fig_b}\n\\end{subfigure}\n\n\\medskip\n\\begin{subfigure}[t]{.4\\textwidth}\n\\centering\n\\vspace{0pt\n\\includegraphics[width=\\linewidth]{Arrangement3_.pdf}\n\\caption{}\\label{fig:fig_c}\n\\end{subfigure}\n\\begin{subfigure}[t]{.4\\textwidth}\n\\centering\n\\vspace{0pt\n\\includegraphics[width=\\linewidth]{Arrangement4_.pdf}\n\\caption{}\\label{fig:fig_c}\n\\end{subfigure}\n\\begin{minipage}[t]{1\\textwidth}\n\\caption{Four possible heat source arrangements for Problem \\ref{Prob:1dRod}.}\n\\end{minipage}\n\n\\end{figure}\n\nThe solution to Problem \\ref{Prob:1dRod} follows from a celebrated result in symmetrization known as Talenti's Theorem \\cite{Talenti}. To understand Talenti's solution\\footnote{Although Talenti's work in \\cite{Talenti} explicitly assumes that the dimension $n\\geq2$, the result still holds in dimension $1$. For a different approach to comparison theorems that yields the same result in all dimensions, see Corollary 3 of \\cite{Baernstein Cortona Volume} or Theorem 10.10 of \\cite{Barenstein Star Function in Complex Analysis} and Corollary \\ref{Cor:DComp} below.}, we write out the mathematical formulation of Problem \\ref{Prob:1dRod}. Suppose the rod is located along the interval $\\left[-\\frac{\\ell}{2},\\frac{\\ell}{2}\\right]$ and let $E\\subseteq [-\\frac{\\ell}{2},\\frac{\\ell}{2}]$ denote the locations of the heat sources. Then the steady-state temperature function $u$ satisfies the one-dimensional Poisson problem\n\\begin{equation}\\label{eq:1duTalenti}\n-u''=\\chi_E \\quad \\textup{in} \\quad \\left(-\\frac{\\ell}{2},\\frac{\\ell}{2}\\right), \\qquad u\\left(-\\frac{\\ell}{2}\\right)=u\\left(\\frac{\\ell}{2}\\right)=0,\n\\end{equation}\nwhere $\\chi_E$ denotes the characteristic function of the set $E$. Talenti's Theorem compares the solution $u$ in \\eqref{eq:1duTalenti} to the solution $v$ of a problem that has been ``symmetrized.'' Specifically, let $v$ solve the Poisson problem\n\\begin{equation*}\n-v''=\\chi_{\\left(-\\frac{|E|}{2},\\frac{|E|}{2}\\right)} \\quad \\textup{in} \\quad \\left(-\\frac{\\ell}{2},\\frac{\\ell}{2}\\right), \\qquad v\\left(-\\frac{\\ell}{2}\\right)=v\\left(\\frac{\\ell}{2}\\right)=0,\n\\end{equation*}\nwhere $|E|$ denotes the length of $E$; in this case $\\chi_{\\left(-\\frac{|E|}{2},\\frac{|E|}{2}\\right)}$ is called the \\emph{symmetric decreasing rearrangement} of $\\chi_E$ (for a precise definition, see Definition \\ref{def:decrearr}). Talenti showed that the temperature functions $u$ and $v$ compare through their convex means, that is,\n\\begin{equation}\\label{eq:convuv}\n\\int_{-\\frac{\\ell}{2}}^{\\frac{\\ell}{2}}\\phi(u)\\,dx\\leq \\int_{-\\frac{\\ell}{2}}^{\\frac{\\ell}{2}}\\phi(v)\\,dx\n\\end{equation}\nfor each convex increasing function $\\phi:\\mathbb{R} \\to \\mathbb{R}$. Since $u$ and $v$ are concave functions, they are minimized at the ends of the interval $\\left[-\\frac{\\ell}{2},\\frac{\\ell}{2}\\right]$, where both functions vanish. That is,\n\\begin{equation}\\label{eq:infuv}\n\\min_{\\left[-\\frac{\\ell}{2},\\frac{\\ell}{2}\\right]} u=\\min_{\\left[-\\frac{\\ell}{2},\\frac{\\ell}{2}\\right]} v=0.\n\\end{equation}\nThus $u$ and $v$ are nonnegative. Taking $\\phi(x)=\\chi_{[0,\\infty)}(x)\\cdot x^p$ in \\eqref{eq:convuv} gives\n\\[\n\\|u\\|_{L^p\\left[-\\frac{\\ell}{2},\\frac{\\ell}{2}\\right]}\\leq \\|v\\|_{L^p\\left[-\\frac{\\ell}{2},\\frac{\\ell}{2}\\right]}, \\qquad 1\\leq p<+\\infty,\n\\]\nand sending $p\\to +\\infty$ shows\n\\begin{equation}\\label{eq:supuv}\n\\max_{\\left[-\\frac{\\ell}{2},\\frac{\\ell}{2}\\right]} u \\leq \\max_{\\left[-\\frac{\\ell}{2},\\frac{\\ell}{2}\\right]} v.\n\\end{equation}\nTalenti's Theorem thus says that the hottest temperature in Problem \\ref{Prob:1dRod} is maximized when the heat sources are centrally gathered in the middle of the rod as in (A) of Figure 1. Note also that equations \\eqref{eq:infuv} and \\eqref{eq:supuv} show that \n\\[\n\\underset{\\left[-\\frac{\\ell}{2},\\frac{\\ell}{2}\\right]}{\\textup{osc}}\\ u \\leq \\underset{\\left[-\\frac{\\ell}{2},\\frac{\\ell}{2}\\right]}{\\textup{osc}}\\ v,\n\\]\nwhere $\\textup{osc}=\\max - \\min$ denotes the oscillation, or temperature gap (the difference between the rod's largest and smallest temperatures). Thus, arrangement (A) not only maximizes the rod's hottest temperature but also its temperature gap.\n\nThe present paper is motivated by a simple question: What happens if we consider analogues of Problem \\ref{Prob:1dRod} with other boundary conditions? To start, we might consider a situation where the ends of the bar interact with the outside environment. For example, imagine that each end of the bar is submerged in a large bath of fluid with temperature zero. Newton's law of cooling then says that the heat flux is proportional to the temperature at each end of the rod. This physical setting yields boundary conditions known as Robin boundary conditions. Thus, we ask:\n\n\\begin{problem}\\label{Prob:1dRodRobin}\nWith the same setup as Problem \\ref{Prob:1dRod} for a rod with Robin boundary conditions, where should we locate the heat sources to maximize the hottest steady-state temperature?\n\\end{problem}\n\nWe solve Problem \\ref{Prob:1dRodRobin} by proving a one-dimensional comparison principle for Robin problems in the spirit of Talenti. The result stated below is normalized so the length of the bar equals $2\\pi$, but the the result holds for any interval. Specifically, we prove:\n\n\\begin{theorem}[ODE Robin Comparison Principle]\\label{Th:RComp}\nLet $0\\leq f\\in L^1[-\\pi,\\pi]$ and $\\alpha>0$. Suppose $u$ and $v$ solve the Poisson problems\n\\begin{align*}\n-u''&=f \\quad \\textup{in} \\quad (-\\pi,\\pi), \\qquad -u'(-\\pi)+\\alpha u(-\\pi)=u'(\\pi)+\\alpha u(\\pi)=0,\\\\\n-v''&=f^{\\#} \\quad \\textup{in} \\quad (-\\pi,\\pi), \\qquad -v'(-\\pi)+\\alpha v(-\\pi)=v'(\\pi)+\\alpha v(\\pi)=0,\n\\end{align*}\nwith $f^{\\#}$ the symmetric decreasing rearrangement of $f$. Then\n\\[\n\\int_{-\\pi}^{\\pi}\\phi(u)\\,dx\\leq \\int_{-\\pi}^{\\pi}\\phi(v)\\,dx\n\\]\nfor each increasing convex function $\\phi:\\mathbb{R}\\to \\mathbb{R}$. In particular,\n\\begin{equation}\\label{eq:LpRobin}\n\\|u\\|_{L^p[-\\pi,\\pi]} \\leq \\|v\\|_{L^p[-\\pi,\\pi]},\\qquad 1\\leq p\\leq +\\infty.\\\\\n\\end{equation}\n\\end{theorem}\n\nTo resolve Problem \\ref{Prob:1dRodRobin}, let $E\\subseteq [-\\pi,\\pi]$ denote the locations of the heat sources and let $u$ and $v$ denote the corresponding temperature functions for the Robin problems of Theorem \\ref{Th:RComp}:\n\\begin{align*}\n-u''&=\\chi_E \\quad \\textup{in} \\quad (-\\pi,\\pi), \\qquad -u'(-\\pi)+\\alpha u(-\\pi)=u'(\\pi)+\\alpha u(\\pi)=0,\\\\\n-v''&= \\chi_{\\left(-\\frac{\\pi}{2},\\frac{\\pi}{2}\\right)} \\quad \\textup{in} \\quad (-\\pi,\\pi), \\qquad -v'(-\\pi)+\\alpha v(-\\pi)=v'(\\pi)+\\alpha v(\\pi)=0.\n\\end{align*}\nThe proof of Theorem \\ref{Th:RComp} shows that $u$ and $v$ are nonnegative, thus taking $p=+\\infty$ in \\eqref{eq:LpRobin} shows\n\\[\n\\max_{\\left[-\\pi,\\pi\\right]} u \\leq \\max_{\\left[-\\pi,\\pi\\right]} v.\n\\]\nThus, as in Problem \\ref{Prob:1dRod}, Problem \\ref{Prob:1dRodRobin} is resolved with an arrangement of heat sources analogous to (A) in Figure 1. In fact, in Corollary \\ref{Cor:DComp}, we prove that the corresponding Dirichlet result follows from Theorem \\ref{Th:RComp}. Unlike the Dirichlet setting, however, the temperature gap does not necessarily increase under symmetric decreasing rearrangement; see Example \\ref{ex:osc} and Proposition \\ref{prop:robingapint}.\n\nWe also address the situation where the ends of the rod are perfectly insulated. In this setting, we cannot consider a verbatim analogue of Problem \\ref{Prob:1dRod}, since perfect insulation requires the presence of both heat sinks and sources. The temperature function, moreover, is unique only up to an additive constant. Thus, we ask:\n\n\\begin{problem}\\label{Prob:1dRodNeumann}\nSuppose half of a given rod is heated uniformly, while on the complimentary half, heat is absorbed uniformly. If the rod's ends are perfectly insulated, where should we place the heat sources and sinks to maximize the hottest steady-state temperature across the rod, assuming the temperature has zero mean?\n\\end{problem}\n\n\n\n\n\nAgain, we solve Problem \\ref{Prob:1dRodNeumann} by proving a Talenti-style comparison principle. We normalize and assume the rod has length $\\pi$, but as before, the result holds for any interval. We prove:\n\n\n\\begin{theorem}[ODE Neumann Comparison Principle]\\label{Th:NComp}\nLet $f\\in L^1[0,\\pi]$ have zero mean and suppose $u$ and $v$ solve the Poisson problems\n\\begin{align*}\n-u''&=f \\quad \\textup{in} \\quad (0,\\pi), \\qquad u'(0)=u'(\\pi)=0,\\\\\n-v''&=f^{\\ast} \\quad \\textup{in} \\quad (0,\\pi), \\qquad v'(0)=v'(\\pi)=0,\n\\end{align*}\nwith $f^{\\ast}$ the decreasing rearrangement of $f$. If $u$ and $v$ both have zero mean, then\n\\[\n\\int_{0}^{\\pi}\\phi(u)\\,dx\\leq \\int_{0}^{\\pi}\\phi(v)\\,dx\n\\]\nfor each convex function $\\phi:\\mathbb{R}\\to \\mathbb{R}$. In particular,\n\\begin{align*}\n\\|u\\|_{L^p[0,\\pi]} &\\leq \\|v\\|_{L^p[0,\\pi]},\\qquad 1\\leq p\\leq +\\infty,\\\\\n\\max_{[0,\\pi]}u \\leq \\max_{[0,\\pi]}v,\\qquad \\min_{[0,\\pi]}v &\\leq \\min_{[0,\\pi]}u, \\qquad \\underset{[0,\\pi]}{\\textup{osc}} \\ u \\leq \\underset{[0,\\pi]}{\\textup{osc}} \\ v.\n\\end{align*}\n\\end{theorem}\n\nTo resolve Problem \\ref{Prob:1dRodNeumann}, let $E \\subseteq [0,\\pi]$ denote the locations of the heat sources and let $u$ and $v$ denote the temperature functions for the Neumann problems of Theorem \\ref{Th:NComp}:\n\\begin{align*}\n-u''&=\\chi_E-\\chi_{[0,\\pi]\\setminus E} \\quad \\textup{in} \\quad (0,\\pi), \\qquad u'(0)=u'(\\pi)=0,\\\\\n-v''&=\\chi_{\\left[0,\\frac{\\pi}{2}\\right)}-\\chi_{\\left[\\frac{\\pi}{2},\\pi\\right]} \\quad \\textup{in} \\quad (0,\\pi), \\qquad v'(0)=v'(\\pi)=0.\n\\end{align*}\nAccording to Theorem \\ref{Th:NComp}, the maximum temperature and temperature gap increase, and the minimal temperature decreases under decreasing rearrangement:\n\\[\n\\max_{[0,\\pi]}u \\leq \\max_{[0,\\pi]}v,\\qquad \\min_{[0,\\pi]}v \\leq \\min_{[0,\\pi]}u, \\qquad \\underset{[0,\\pi]}{\\textup{osc}} \\ u \\leq \\underset{[0,\\pi]}{\\textup{osc}} \\ v.\n\\]\nThus Problem \\ref{Prob:1dRodNeumann} is resolved by choosing an arrangement of sources and sinks analogous to (B) in Figure 1, only here, white areas represent heat sources and gray areas represent heat sinks.\n\nTaken in sum, Theorems \\ref{Th:RComp} and \\ref{Th:NComp} reveal a striking difference in the behavior of source functions that induce large temperature functions (interpreted in the sense of convex means). With the Neumann problem, one takes full advantage of the insulated ends, sweeping the greatest sources to one end of the bar and greatest sinks to the opposite end. However, the instant any heat energy is allowed to escape through the bar's ends and the Robin regime is entered, the arrangement switches and we instead move the greatest sources towards the middle of the bar and push the weakest sources out towards the ends.\n\n\nThe results of our paper are examples of comparison theorems for differential equations. To place our work in the existing literature, we recall that the first major comparison result, as mentioned above, is due to Talenti \\cite{Talenti}, who compared the solutions of two Poisson problems with Dirichlet boundary conditions and nonnegative source, $f$, namely\n\\[\n\\begin{array}{rclccccrclcc}\n-\\Delta u & = & f & \\text{in} & \\Omega, & & & -\\Delta v & = & f^{\\#} & \\text{in} & \\Omega^{\\#},\\\\\nu & = & 0 & \\text{on} & \\partial \\Omega, & & &v & = & 0 & \\text{on} & \\partial \\Omega^{\\#}.\n\\end{array}\n\\]\nHere, $\\Omega \\subseteq \\mathbb{R}^n$ is a bounded Lipschitz domain with $n\\geq 2$, $0\\leq f\\in L^2(\\Omega)$, $\\Omega^{\\#} \\subseteq \\mathbb{R}^n$ is the open ball centered at $0$ with the same volume as $\\Omega$, and $f^{\\#}$ denotes the symmetric decreasing rearrangement of $f$, a radially decreasing function on $\\Omega^{\\#}$ whose upper level sets have the same volume as those of $f$, meaning $|\\{x\\in \\Omega:f(x)>t\\}|=|\\{x\\in \\Omega^{\\#}:f^{\\#}(x)>t\\}|$ for $t\\in \\mathbb{R}$. Talenti showed that the solutions $u$ and $v$ compare via their symmetric decreasing rearrangements through the inequality\n\\[\nu^{\\#} \\leq v \\quad \\textup{in }\\Omega^{\\#}.\n\\]\n\n \nThe history of comparison phenomena that followed Talenti's original work is long and the results are the subject of many articles. Fortunately, Talenti has prepared a thorough survey of the material (up to 2016). We direct the reader interested in this important background to \\cite{TalentiSurvey} and the references therein. \n\nSince the publication of Talenti's survey, authors have begun to turn their attention to comparison principles for Robin problems. As an example relevant to our work, in \\cite{ANT} Alvino, Nitsch, and Trombetti consider the exact same setup addressed by Talenti \\cite{Talenti} and mentioned above, but impose Robin boundary conditions rather than Dirichlet boundary conditions. That is, for $\\alpha>0$, $\\Omega \\subseteq \\mathbb{R}^n$ a bounded Lipschitz domain with $n\\geq 2$, and $0\\leq f\\in L^2(\\Omega)$, they consider the problems\n\\[\n\\begin{array}{rclccccrclcc}\n-\\Delta u & = & f & \\text{in} & \\Omega, & & & -\\Delta v & = & f^{\\#} & \\text{in} & \\Omega^{\\#},\\\\\n\\frac{\\partial u}{\\partial \\nu}+\\alpha u & = & 0 & \\text{on} & \\partial \\Omega, & & &\\frac{\\partial v}{\\partial \\nu}+\\alpha v & = & 0 & \\text{on} & \\partial \\Omega^{\\#},\n\\end{array}\n\\]\nwith $\\frac{\\partial}{\\partial \\nu}$ the outer normal derivative and $\\#$ the symmetric decreasing rearrangement. The authors show that Talenti's conclusion $u^{\\#}\\leq v$ in $\\Omega^{\\#}$ fails in general, but that $u$ and $v$ compare via their Lorentz norms. In dimension $n=2$, they show that the $L^1$- and $L^2$-norms of $u$ are dominated by those of $v$, and when $f=1$, that $u^{\\#} \\leq v$ in $\\Omega^{\\#}$. These results are extended in the subsequent work of Alvino, Chiacchio, Nitsch, and Trombetti \\cite{ACNT}. In related work \\cite{AGM}, Amato, Gentile, and Masiello, generalize results of \\cite{ANT} to a nonlinear setting, replacing the Laplacian with the $p$-Laplace operator.\n\nIn addition to the the results of \\cite{ACNT}, \\cite{ANT}, and \\cite{AGM}, in \\cite{Langford4} the first author studies Poisson problems of the form\n\\[\n\\begin{array}{rclccccrclcc}\n-\\Delta u & = & f & \\text{in} & A, & & & -\\Delta v & = & f^{\\#} & \\text{in} & A,\\\\\n\\frac{\\partial u}{\\partial \\nu}+\\alpha u & = & 0 & \\text{on} & \\partial A, & & &\\frac{\\partial v}{\\partial \\nu}+\\alpha v & = & 0 & \\text{on} & \\partial A,\n\\end{array}\n\\]\nwhere $A\\subseteq \\mathbb{R}^n$ is a spherical shell (the region between two concentric spheres), $\\alpha>0$, $f\\in L^2(A)$ and $\\#$ is the {\\it cap symmetrization}. (To cap symmetrize a function $f:A\\to \\mathbb{R}$, one applies the spherical rearrangement (the analogue of the symmetric decreasing rearrangement on the sphere) to each of $f$'s radial slice functions). The author shows that the solutions $u$ and $v$ compare through their convex means:\n\\[\n\\int_{A}\\phi(u)\\,dx\\leq \\int_{A}\\phi(v)\\,dx\n\\]\nfor each convex function $\\phi:\\mathbb{R} \\to \\mathbb{R}$. The author obtains similar results for $\\alpha=0$ (the Neumann problem), assuming $f$, $u$, and $v$ all have zero mean. (For related work on the Neumann problem, see \\cite{Langford1}, \\cite{Langford2}, and \\cite{Langford3}).\n\nTo the best of our knowledge, references \\cite{ACNT}, \\cite{ANT}, \\cite{AGM}, and \\cite{Langford4} comprise all that has appeared in print to addresses Robin comparison principles for differential equations in the spirit of Talenti. Thus, our work adds an interesting contribution to this new direction in the study of comparison principles.\n\n\nThe rest of this note is organized as follows. In Section 2 we discuss existence and uniqueness results for the Poisson problems of Theorems \\ref{Th:RComp} and \\ref{Th:NComp}, so that our paper may be self-contained. We then discuss Robin Green's functions and relevant rearrangement inequalities needed to prove Theorems \\ref{Th:RComp} and \\ref{Th:NComp}. In Section 3, we prove our paper's main results.\n\n\\section{Background}\n\nSince the goal of our paper is to compare the solutions of one-dimensional Poisson problems with Robin and Neumann boundary conditions, we begin with two existence and uniqueness results. These results are stated on the interval $[-\\pi,\\pi]$ for convenience, but they hold for any interval.\n\n\\begin{proposition}[Robin Existence and Uniqueness]\\label{Prop:Runiq}\nLet $f\\in L^1[-\\pi,\\pi]$ and $\\alpha>0$. A unique $u\\in C^1[-\\pi,\\pi]$ exists satisfying\n\\begin{itemize}\n\\item[1.] $u'$ is absolutely continuous on $[-\\pi,\\pi]$.\n\\item[2.] $-u''=f$ a.e. on $(-\\pi,\\pi)$.\n\\item[3.] $-u'(-\\pi)+\\alpha u(-\\pi)=u'(\\pi)+\\alpha u(\\pi)=0$.\n\\end{itemize}\n\\end{proposition}\n\n\\begin{proof}\nWe first establish uniqueness. Suppose $u$ and $v$ both satisfy all the properties listed above, and let $w=u-v$. Since $w'$ is absolutely continuous, for each $x\\in[-\\pi,\\pi]$ we have\n\\[\nw'(x)=w'(x)-w'(-\\pi)+\\alpha w(-\\pi)=\\int_{-\\pi}^x(-f+f)\\,dy+\\alpha w(-\\pi)=\\alpha w(-\\pi).\n\\]\nThus, $w(x)=\\alpha w(-\\pi)x+b$ for some constant $b$. The equations $-w'(-\\pi)+\\alpha w(-\\pi)=w'(\\pi)+\\alpha w(\\pi)=0$ imply that $\\alpha w(-\\pi)=b=0$, and so $u\\equiv v$. For existence, we simply take\n\\[\nu(x)=-\\int_{-\\pi}^x\\int_{-\\pi}^tf(s)\\,ds\\,dt+cx+d,\n\\]\nwhere $c$ and $d$ are chosen to make $-u'(-\\pi)+\\alpha u(-\\pi)=u'(\\pi)+\\alpha u(\\pi)=0$.\n\\end{proof}\n\nWe also have an existence and uniqueness result for the Neumann problem. The proof is similar to that of the Robin result.\n\\begin{proposition}[Neumann Existence and Uniqueness]\\label{Prop:Nuniq}\nLet $f\\in L^1[-\\pi,\\pi]$ with $\\int_{-\\pi}^{\\pi}f\\,dx=0$. A unique $u\\in C^1[-\\pi,\\pi]$ exists satisfying\n\\begin{itemize}\n\\item[1.] $u'$ is absolutely continuous on $[-\\pi,\\pi]$.\n\\item[2.] $-u''=f$ a.e. on $(-\\pi,\\pi)$.\n\\item[3.] $u'(-\\pi)=u'(\\pi)=0$.\n\\item[4.] $\\int_{-\\pi}^{\\pi}u\\,dx=0$.\n\\end{itemize}\n\\end{proposition}\n\nThus, the solutions $u$ and $v$ in Theorems \\ref{Th:RComp} and \\ref{Th:NComp} are guaranteed to exist and be unique. For the Robin problem, we in fact prove a bit more. Namely, we show that solutions are obtained by integration against an explicitly computable Green's function.\n\n\\begin{proposition}[Green's Representation]\\label{Prop:RGreen}\nFor $\\alpha>0$, the Green's function for the Robin problem on the interval $[-\\pi,\\pi]$ equals\n\\[\nG(x,y)=-\\frac{1}{2}c_{\\alpha}xy-\\frac{1}{2}|x-y|+\\frac{1}{2c_{\\alpha}},\\qquad x,y\\in [-\\pi,\\pi],\n\\]\nwhere\n\\begin{equation}\\label{eq:calphadef}\nc_{\\alpha}=\\frac{\\alpha}{1+\\alpha \\pi}.\n\\end{equation}\nThat is,\n\\begin{itemize}\n\\item[1.] $-G_{xx}(x,y)=\\delta_x(y),$ for $x,y\\in (-\\pi,\\pi),$ \n\\item[2.] $-G_x(-\\pi,y)+\\alpha G(-\\pi,y)=G_x(\\pi,y)+\\alpha G(\\pi,y)=0$ for $y\\in (-\\pi,\\pi)$.\n\\end{itemize}\nThus, if $f\\in L^1[-\\pi,\\pi]$ and $u$ solves\n\\[\n-u''=f \\quad \\textup{in} \\quad (-\\pi,\\pi), \\qquad -u'(-\\pi)+\\alpha u(-\\pi)=u'(\\pi)+\\alpha u(\\pi)=0,\n\\]\nthen\n\\[\nu(x)=\\int_{-\\pi}^{\\pi}G(x,y)f(y)\\,dy, \\qquad x\\in [-\\pi,\\pi].\n\\]\n\\end{proposition}\n\n\\begin{proof}\nProperties 1 and 2 follow from a straightforward calculation. Define\n\\begin{equation}\\label{eq:wdefGreen}\nw(x)=\\int_{-\\pi}^{\\pi}G(x,y)f(y)\\,dy.\n\\end{equation}\nWe show that $w\\in C^1[-\\pi,\\pi]$ and that $w$ satisfies all three properties of Proposition \\ref{Prop:Runiq}.\n\nThe Dominated Convergence Theorem gives\n\\begin{align}\nw'(x)&=\\int_{-\\pi}^{\\pi}G_x(x,y)f(y)\\,dy \\label{eq:wpdefGreen}\\\\\n&=\\int_{-\\pi}^x\\left(-\\frac{1}{2}c_{\\alpha}y-\\frac{1}{2}\\right)f(y)\\,dy + \\int_{x}^{\\pi}\\left(-\\frac{1}{2}c_{\\alpha}y+\\frac{1}{2}\\right)f(y)\\,dy,\\nonumber\n\\end{align}\nand this representation shows that $w'$ is absolutely continuous on $[-\\pi,\\pi]$ with $-w''=f$ a.e. Formulas \\eqref{eq:wdefGreen} and \\eqref{eq:wpdefGreen} for $w$ and $w'$ together with property 2 of the present proposition give $-w'(-\\pi)+\\alpha w(-\\pi)=w'(\\pi)+\\alpha w(\\pi)=0$. The result now follows from uniqueness in Proposition \\ref{Prop:Runiq}.\n\n\\end{proof}\n\nTo prove our main results, we will also need several tools from symmetrization. We start with the decreasing and symmetric decreasing rearrangements.\n\n\\begin{definition}\n[Decreasing and Symmetric Decreasing Rearrangements]\\label{def:decrearr}Suppose $X\\subseteq \\mathbb{R}$ is a measurable set and $f\\in L^{1}(X)$ satisfies the finiteness condition\n\\[\n|\\{x\\in X:f(x)>t\\}|<\\infty,\\qquad t>\\underset{X}{\\textup{ess\\ inf}} \\ f.\n\\]\n\nDefine $f^{\\ast}:[0,|X|]\\rightarrow[-\\infty,+\\infty]$\nvia\n\\[\nf^{\\ast}(t)=\\begin{cases}\n\\underset{X}{\\textup{ess\\ sup}}\\ f & \\textup{if}\\ t=0,\\\\\n\\inf\\{s:|\\{x:s0,\n\\]\nand so $\\underset{[-\\pi,\\pi]}{\\min}\\ u$ cannot be achieved at $\\pi$. We conclude that $u(\\pi)\\geq 0$. An identical argument applies when the minimum occurs at $-\\pi$ from which we conclude that $u$ is nonnegative. Since the same argument applies to the function $v,$ we conclude that $v$ is also nonnegative. The theorem's remaining conclusions now follow from Proposition \\ref{cor:Lpnormsgen}.\n\\end{proof} \n\n\\subsection*{From Robin to Dirichlet} As mentioned in the introduction, our Robin comparison principle implies the corresponding Dirichlet result. The main idea behind the proof is that as $\\alpha$ tends to $+\\infty$, the Robin boundary condition converts into a Dirichlet condition.\n\n\\begin{corollary}[ODE Dirichlet Comparison Principle]\\label{Cor:DComp}\nLet $0\\leq f\\in L^1[-\\pi,\\pi]$ and suppose $u$ and $v$ solve the Poisson problems\n\\begin{align*}\n-u''&=f \\quad \\textup{in} \\quad (-\\pi,\\pi), \\qquad u(-\\pi)=u(\\pi)=0,\\\\\n-v''&=f^{\\#} \\quad \\textup{in} \\quad (-\\pi,\\pi), \\qquad v(-\\pi)=v(\\pi)=0,\n\\end{align*}\nwith $f^{\\#}$ the symmetric decreasing rearrangement of $f$. Then\n\\[\n\\int_{-\\pi}^{\\pi}\\phi(u)\\,dx\\leq \\int_{-\\pi}^{\\pi}\\phi(v)\\,dx\n\\]\nfor each increasing convex function $\\phi:\\mathbb{R}\\to \\mathbb{R}$.\n\\end{corollary}\n\n\\begin{proof}\nLet $f,u$, and $v$ be as stated, and suppose $u_k$ and $v_k$ solve the Robin problems\n\\begin{align*}\n-u_k''&=f \\quad \\textup{in} \\quad (-\\pi,\\pi), \\qquad -u_k'(-\\pi)+\\alpha_k u_k(-\\pi)=u_k'(\\pi)+\\alpha_k u_k(\\pi)=0,\\\\\n-v_k''&=f^{\\#} \\quad \\textup{in} \\quad (-\\pi,\\pi), \\qquad -v_k'(-\\pi)+\\alpha_k v_k(-\\pi)=v_k'(\\pi)+\\alpha_k v_k(\\pi)=0,\n\\end{align*}\nwhere $0<\\alpha_k\\to +\\infty$. Then by Theorem \\ref{Th:RComp} and Proposition \\ref{Prop:Majorization},\n\\begin{equation}\\label{eq:ukvkstar}\nu_k^{\\bigstar}(t)\\leq v_k^{\\bigstar}(t),\\qquad t\\in [0,2\\pi].\n\\end{equation}\nIf $G_{\\alpha_k}$ denotes the Robin Green's function of Proposition \\ref{Prop:RGreen} with parameter $\\alpha_k$, then\n\\[\nG_{\\alpha_k}(x,y)\\to G(x,y)\n\\]\nuniformly on $[-\\pi,\\pi]\\times[-\\pi,\\pi]$, where\n\\[\nG(x,y)=-\\frac{1}{2\\pi}xy-\\frac{1}{2}|x-y|+\\frac{\\pi}{2}\n\\]\nis the Dirichlet Green's function on $[-\\pi,\\pi]$. Using the Green's representation, we see $u_k\\to u$ and $v_k\\to v$ uniformly on $[-\\pi,\\pi]$. Since symmetrization decreases the $L^1$-distance, we note that for any $t\\in[0,2\\pi]$,\n\\begin{align*}\n|u_k^{\\bigstar}(t)-u^{\\bigstar}(t)|&= \\left|\\int_{-\\frac{t}{2}}^{\\frac{t}{2}}\\left(u_k^{\\#}(x)-u^{\\#}(x)\\right)\\,dx\\right|\\\\\n&\\leq \\int_{-\\pi}^{\\pi}\\left|u_k^{\\#}(x)-u^{\\#}(x)\\right|\\,dx\\\\\n&\\leq \\int_{-\\pi}^{\\pi}\\left|u_k(x)-u(x)\\right|\\,dx,\n\\end{align*}\nand this last term tends to zero by uniform convergence. Thus we have pointwise convergence of star functions:\n\\begin{equation}\\label{eq:ukstar}\nu^{\\bigstar}_k(t)\\to u^{\\bigstar}(t),\\qquad t\\in[0,2\\pi].\n\\end{equation}\nA similar argument shows\n\\begin{equation}\\label{eq:vkstar}\nv^{\\bigstar}_k(t)\\to v^{\\bigstar}(t),\\qquad t\\in[0,2\\pi].\n\\end{equation}\nCombining inequality \\eqref{eq:ukvkstar} with \\eqref{eq:ukstar} and \\eqref{eq:vkstar} gives $u^{\\bigstar}\\leq v^{\\bigstar}$ in $[0,2\\pi]$, and this inequality is equivalent to the corollary's conclusion courtesy of Proposition \\ref{Prop:Majorization}.\n\n\n\\end{proof}\n\nThe conclusion of Corollary \\ref{Cor:DComp} can be strengthened to $u^{\\#} \\leq v$ in $[-\\pi,\\pi]$, though this stronger conclusion is not needed for our paper. The argument is simple but requires additional tools developed by Baernstein. We include the proof for the sake of completeness.\n\nDefine\n\\[\nw(s)=u^{\\bigstar}(2s)-v^{\\bigstar}(2s),\\qquad s\\in[0,\\pi].\n\\]\nThen by Theorem 9.20 in \\cite{Baernstein Symmetrization in Analysis} or Theorem 5 of \\cite{Baernstein Cortona Volume}, $\\frac{d^2}{ds^2}w(s)\\geq 0$ weakly. That is,\n\\[\n\\int_{0}^{\\pi}w(s)G''(s)\\,ds \\geq 0\n\\]\nfor each $G_c^2(0,\\pi)$ nonnegative with compact support. Integrating by parts, we see\n\\[\n\\int_{0}^{\\pi}w'(s)G'(s)\\,ds \\leq 0.\n\\]\nAn easy argument gives that $w'(s)$ is increasing. But\n\\[\n\\frac{d}{ds}w(s)=\\frac{d}{ds}\\int_{-s}^s\\left(u^{\\#}(x)-v^{\\#}(x)\\right)\\,dx=2\\left(u^{\\#}(s)-v^{\\#}(s)\\right)\n\\]\nand so we see that $u^{\\#}(s)-v^{\\#}(s)$ is increasing for $s\\in[0,\\pi]$. But as $u^{\\#}(\\pi)=v^{\\#}(\\pi)=0$, we conclude $u^{\\#}(s)-v^{\\#}(s)\\leq 0$. Finally, since $v=v^{\\#}$, this last inequality implies $u^{\\#}\\leq v$ on $[-\\pi,\\pi]$.\n\n\\subsection*{An interesting example} As noted in the introduction, with Robin problems the oscillation need not increase under symmetric decreasing rearrangement. Consider the following example.\n\n\\begin{example}\\label{ex:osc}\nConsider the solutions $u$ and $v$ to the Poisson problems of Theorem \\ref{Th:RComp} with $f=\\chi_{[-\\pi,0]}$:\n\\begin{align*}\n-u''&=\\chi_{[-\\pi,0]} \\quad \\textup{in} \\quad [-\\pi,\\pi], \\qquad -u'(-\\pi)+\\alpha u(-\\pi)=u'(\\pi)+\\alpha u(\\pi)=0,\\\\\n-v''&=\\chi_{\\left[-\\frac{\\pi}{2},\\frac{\\pi}{2}\\right]} \\quad \\textup{in} \\quad [-\\pi,\\pi], \\qquad -v'(-\\pi)+\\alpha v(-\\pi)=v'(\\pi)+\\alpha v(\\pi)=0.\n\\end{align*}\n\\end{example}\nIt is straightforward to check that\n\\[\nu(x)=-u_1(x)+\\left(\\frac{\\pi}{2}+\\frac{\\pi^2}{4}c_{\\alpha}\\right)x+\\frac{\\pi}{2c_{\\alpha}}+\\frac{\\pi^2}{4},\n\\]\nwhere\n\\[\nu_1(x)=\n\\begin{cases}\n\\frac{1}{2}x^2+\\pi x+\\frac{\\pi^2}{2} & \\textup{if } -\\pi\\leq x< 0,\\\\\n\\pi x+\\frac{\\pi^2}{2} & \\textup{if } 0\\leq x\\leq \\pi,\n\\end{cases}\n\\]\nand $c_{\\alpha}$ is defined in \\eqref{eq:calphadef}. Similarly,\n\\[\nv(x)=-v_1(x)+\\frac{\\pi}{2}x+\\frac{\\pi}{2c_{\\alpha}},\n\\]\nwhere\n\\[\nv_1(x)=\n\\begin{cases}\n0 & \\textup{if }-\\pi \\leq x<-\\frac{\\pi}{2},\\\\\n\\frac{1}{2}x^2+\\frac{\\pi}{2}x+\\frac{\\pi^2}{8} & \\textup{if } -\\frac{\\pi}{2} \\leq x < \\frac{\\pi}{2},\\\\\n\\pi x & \\textup{if } \\frac{\\pi}{2}\\leq x\\leq \\pi.\n\\end{cases}\n\\]\nSince $v$ is symmetric decreasing, we have\n\\[\n\\underset{[-\\pi,\\pi]}{\\osc}\\ v=v(0)-v(\\pi)=\\frac{3\\pi^2}{8}.\n\\]\nNote, rather curiously, that this oscillation is independent of $\\alpha$. On the other hand,\n\\[\n\\underset{[-\\pi,\\pi]}{\\osc}\\ u \\geq u\\left(-\\frac{\\pi}{2}\\right)-u(\\pi)=\\frac{\\pi^2(5+2\\alpha \\pi)}{8(1+\\alpha \\pi)}.\n\\]\nNow it is easy to verify that $\\frac{3\\pi^2}{8}<\\frac{\\pi^2(5+2\\alpha \\pi)}{8(1+\\alpha \\pi)}$ so long as $\\alpha<\\frac{2}{\\pi}$, which ensures $\\underset{[-\\pi,\\pi]}{\\osc}\\ v < \\underset{[-\\pi,\\pi]}{\\osc}\\ u$.\n\nExample \\ref{ex:osc} leads to an interesting open question.\n\n\\begin{problem}[Open]\\label{Prob:1dRodRobinOpen}\nSuppose half of a rod of length $2\\pi$ with Robin boundary conditions is heated uniformly, while in the remaining half, heat is neither generated nor absorbed. Where should we place the heat sources to maximize the temperature gap?\n\\end{problem}\n\nOur intuition suggests that an optimal source in Problem \\ref{Prob:1dRodRobinOpen} is an interval of length $\\pi$. Assuming this is the case, we have the following proposition.\n\n\\begin{proposition}\\label{prop:robingapint}\nAssume the temperature gap in Problem \\ref{Prob:1dRodRobinOpen} is maximized by a source interval of length $\\pi$. Then when $0<\\alpha\\leq \\frac{2}{\\sqrt{3}\\pi}$, the temperature gap is maximized by locating the source interval at either end of the rod. As $\\alpha$ increases from $\\frac{2}{\\sqrt{3}\\pi}$ to $+\\infty$, the temperature gap is maximized by a source interval that continuously transitions from one end of the rod towards its center.\n\\end{proposition}\n\n\\begin{proof}\nDenote $I_b=\\left[b-\\frac{\\pi}{2},b+\\frac{\\pi}{2}\\right]$ for $-\\frac{\\pi}{2}\\leq b\\leq \\frac{\\pi}{2}$. We consider the gap function\n\\[\n\\textup{Gap}(\\alpha,b)=\\max_{[-\\pi,\\pi]}u-\\min_{[-\\pi,\\pi]}u,\\qquad 0<\\alpha<\\infty, \\quad -\\frac{\\pi}{2}\\leq b\\leq \\frac{\\pi}{2},\n\\]\nwhere $u$ is the solution of the Poisson problem\n\\[\n-u''=\\chi_{I_b} \\quad \\textup{in} \\quad (-\\pi,\\pi), \\qquad -u'(-\\pi)+\\alpha u(-\\pi)=u'(\\pi)+\\alpha u(\\pi)=0.\n\\]\nBy symmetry, one has\n\\[\n\\textup{Gap}(\\alpha,-b)=\\textup{Gap}(\\alpha,b),\\quad 0\\leq b\\leq \\frac{\\pi}{2},\n\\]\nand so for the remainder of the argument, we focus our attention on $\\textup{Gap}$ when $-\\frac{\\pi}{2}\\leq b\\leq 0$, i.e. on source intervals whose center lies in the left half of the rod.\n\nWith a bit of work and help from Mathematica, one computes\n\\[\n\\textup{Gap}(\\alpha,b)=-\\frac{\\pi(1+\\alpha(b+\\pi))(-3\\pi(1+\\alpha\\pi)+b(4+3\\alpha\\pi))}{8(1+\\alpha \\pi)^2}\n\\]\nand so\n\\begin{align}\n\\frac{\\partial\\textup{Gap}}{\\partial b}(\\alpha,b)&=-\\frac{\\pi(2+2\\alpha \\pi+\\alpha b(4+3\\alpha \\pi))}{4(1+\\alpha \\pi)^2},\\label{eq:gapfirstd}\\\\\n\\frac{\\partial^2\\textup{Gap}}{\\partial b^2}(\\alpha,b)&=-\\frac{\\alpha \\pi (4+3\\alpha\\pi)}{4(1+\\alpha \\pi)^2}.\\label{eq:gapconv}\n\\end{align}\nEquation \\eqref{eq:gapfirstd} implies that $\\textup{Gap}(\\alpha,b)$ is decreasing in $b$ on $\\left[-\\frac{\\pi}{2},0\\right]$ when $\\alpha \\in \\left(0,\\frac{2}{\\sqrt{3}\\pi}\\right]$. This establishes the proposition's first claim.\n\nEquation \\eqref{eq:gapconv} says that $\\textup{Gap}(\\alpha,b)$ is strictly concave in $b$ holding $\\alpha$ fixed. We also note that\n\\[\n\\frac{\\partial \\textup{Gap}}{\\partial b}\\left(\\alpha,-\\frac{\\pi}{2}\\right)=\\frac{\\pi(-4+3\\alpha^2\\pi^2)}{8(1+\\alpha\\pi)^2}>0\n\\]\nwhen $\\alpha>\\frac{2}{\\sqrt{3}\\pi}$. It also holds that \n\\[\n\\frac{\\partial \\textup{Gap}}{\\partial b}(\\alpha,0)=-\\frac{\\pi}{2(1+\\alpha \\pi)}<0,\n\\]\nand so $\\textup{Gap}(\\alpha,b)$ is maximized at some $b\\in \\left(-\\frac{\\pi}{2},0\\right)$ when $\\alpha>\\frac{2}{\\sqrt{3}\\pi}$. Returning to equation \\eqref{eq:gapfirstd}, we note that $\\textup{Gap}(\\alpha,b)$ is maximized when\n\\[\nb=b_{\\textup{crit}}=-\\frac{2(1+\\alpha \\pi)}{\\alpha(4+3\\alpha \\pi)}.\n\\]\nSince\n\\[\\frac{d b_{\\textup{crit}}}{d \\alpha}= \\frac{8+6\\alpha \\pi(2+\\alpha \\pi)}{\\alpha^2(4+3\\alpha \\pi)^2}>0,\n\\]\nwe conclude that $b_{\\textup{crit}}$ continuously increases from $-\\frac{\\pi}{2}$ towards $0$ as $\\alpha$ increases from $\\frac{2}{\\sqrt{3}\\pi}$ to $+\\infty$.\n\\end{proof}\n\n\\subsection*{The Neumann problem}\nOur approach here is driven by Fourier series. Suppose that $f\\in L^1[-\\pi,\\pi]$ has zero mean. One attempt to solve the Poisson equation $-u''=f$ in $(-\\pi,\\pi)$ might be to consider the function whose Fourier series is given by\n\\begin{equation}\\label{eq:fourier}\nu(x)=\\sum_{n\\neq 0}\\frac{1}{n^2}\\hat f(n)e^{inx},\n\\end{equation}\nwhere $\\hat f(n)=\\frac{1}{2\\pi}\\int_{-\\pi}^{\\pi}f(x)e^{-inx}\\,dx$. Formally differentiating the equation above termwise shows\n\\[\n-u''(x)=\\sum_{n\\neq 0}\\hat f(n)e^{inx}=f(x).\n\\]\nWe are thus led to consider the function $K$ whose Fourier coefficients are given by\n\\[\n\\hat K(n)=\n\\begin{cases}\n\\frac{1}{n^2} & \\textup{if }n\\neq 0,\\\\\n0 & \\textup{if }n=0.\n\\end{cases}\n\\]\nOne readily verifies that\n\\[\nK(x)=\\frac{1}{2}x^2-\\pi |x|+\\frac{1}{3}\\pi^2, \\qquad -\\pi \\leq x \\leq \\pi.\n\\]\nExtend $f$ and $K$ to all of $\\mathbb{R}$ by $2\\pi$-periodicity. The function $u$ in \\eqref{eq:fourier} has the property that\\\n\\[\n\\hat{u}(n)=\\hat{K}(n)\\hat{f}(n)=\\widehat{K\\ast f}(n)\n\\]\nwhich leads us to study the convolution\n\\begin{equation}\\label{eq:udefcont}\nu(x)=(K\\ast f)(x)=\\frac{1}{2\\pi}\\int_{-\\pi}^{\\pi}K(x-y)f(y)\\,dy.\n\\end{equation}\nWe now investigate how the $u$ defined in \\eqref{eq:udefcont} differs from the $u$ in Proposition \\ref{Prop:Nuniq}. We will see that $u$ defined in \\eqref{eq:udefcont} satisfies properties 1, 2, and 4 of Proposition \\ref{Prop:Nuniq}. Below, we continue to identify $K$ and $f$ with their $2\\pi$-periodic extensions.\n\n\\begin{proposition}\\label{prop:uconvdef}\nThe function $u=K\\ast f$ defined in \\eqref{eq:udefcont} satisfies the following properties:\n\\begin{enumerate}\n\\item[1.] $u$ is continuously differentiable on $\\mathbb{R}$ with $u'=K'\\ast f$.\n\\item[2.] $u'$ is absolutely continuous on $[-\\pi,\\pi]$ and\n\\[\nu'(x)-u'(-\\pi)=-\\int_{-\\pi}^xf(y)\\,dy,\\qquad -\\pi \\leq x \\leq \\pi,\n\\]\nwith $-u''=f$ a.e. in $(-\\pi,\\pi)$.\n\\item[3.] $u'(-\\pi)=u'(\\pi)=-\\frac{1}{2\\pi}\\int_{-\\pi}^{\\pi}xf(x)\\,dx$.\n\\item[4.] $\\int_{-\\pi}^{\\pi}u(x)\\,dx=0$.\n\\end{enumerate}\n\\end{proposition}\n\n\\begin{proof}\nProperty 1 is a standard fact about convolutions. For property 2, we compute\n\\begin{align*}\n(K'\\ast f)(x)&=\\frac{1}{2\\pi}\\int_{-\\pi}^{\\pi}K'(y)f(x-y)\\,dy\\\\\n&=\\frac{1}{2\\pi}\\int_{-\\pi}^{0}(y+\\pi)f(x-y)\\,dy+\\frac{1}{2\\pi}\\int_{0}^{\\pi}(y-\\pi)f(x-y)\\,dy\\\\\n&=\\frac{1}{2\\pi}\\int_{x-\\pi}^x(x-y-\\pi)f(y)\\,dy+\\frac{1}{2\\pi}\\int_{x}^{x+\\pi}(x-y+\\pi)f(y)\\,dy,\n\\end{align*}\nand from this representation it follows that $(K'\\ast f)(x)$ is absolutely continuous on $[-\\pi,\\pi]$. Its derivative equals $-f(x)$ by direct calculation, for almost every $x$. Thus property 2 holds. For property 3, one sees from the above calculation\n\\[\n(K'\\ast f)(-\\pi)=(K'\\ast f)(\\pi)=-\\frac{1}{2\\pi}\\int_{-\\pi}^{\\pi}yf(y)\\,dy.\n\\]\nTo establish property 4 we simply observe\n\\[\n\\frac{1}{2\\pi}\\int_{-\\pi}^{\\pi}u(x)\\,dx=\\hat{u}(0)=\\hat{K}(0) \\hat{f}(0)=0.\n\\]\n\\end{proof}\n\nA consequence of Proposition \\ref{prop:uconvdef} is:\n\n\\begin{proposition}\\label{prop:uisconvwf}\nLet $f$ and $u$ be as in Proposition \\ref{Prop:Nuniq}. If $\\int_{-\\pi}^{\\pi}xf(x)\\,dx=0$, then $u$ is given by convolution as in equation \\eqref{eq:udefcont}.\n\\end{proposition}\n\nBefore proving Theorem \\ref{Th:NComp}, we first prove a preliminary comparison result.\n\n\\begin{theorem}\\label{Th:Nprelim}\nLet $f\\in L^1[-\\pi,\\pi]$ where $\\int_{-\\pi}^{\\pi}f(x)\\,dx=\\int_{-\\pi}^{\\pi}xf(x)\\,dx=0$. Let $u$ and $v$ solve the Poisson problems\n\\begin{align*}\n-u''&=f \\quad \\textup{in} \\quad (-\\pi,\\pi), \\qquad u'(-\\pi)=u'(\\pi)=0,\\\\\n-v''&=f^{\\#} \\quad \\textup{in} \\quad (-\\pi,\\pi), \\qquad v'(-\\pi)=v'(\\pi)=0,\n\\end{align*}\nwhere $f^{\\#}$ is the decreasing rearrangement of $f$, and $\\int_{-\\pi}^{\\pi}u(x)\\,dx=\\int_{-\\pi}^{\\pi}v(x)\\,dx=0$. Then\n\\[\nu^{\\bigstar}(t)\\leq \\int_{-\\frac{t}{2}}^{\\frac{t}{2}}v(x)\\,dx \\leq v^{\\bigstar}(t)\n\\]\non $[0,2\\pi]$. \n\\end{theorem}\n\n\\begin{proof}\nBy Proposition \\ref{prop:uisconvwf}, we have $u=K\\ast f$ and $v=K\\ast f^{\\#}$ (because $f^{\\#}$ is even, and so $\\int_{-\\pi}^{\\pi}xf^{\\#}(x)\\,dx=0$). Fix $t\\in [0,2\\pi]$ and let $E\\subseteq [-\\pi,\\pi]$ denote a measurable subset of length $t$. Applying Theorem \\ref{th:Baernstein} we see\n\\[\n\\int_Eu\\,dx=\\int_E K\\ast f\\,dx \\leq \\int_{E^{\\#}}K\\ast f^{\\#}\\,dx=\\int_{E^{\\#}}v\\,dx.\n\\]\nTaking the $\\sup$ in the above inequality over all measurable subsets $E$ of $[-\\pi,\\pi]$ of length $t$, we obtain the desired conclusion.\n\\end{proof}\n\nWe are now prepared to prove our paper's second main result.\n\n\\begin{proof}[Proof of Theorem \\ref{Th:NComp}]\nGiven $f\\in L^1[0,\\pi]$, extend $f$ to $[-\\pi,\\pi]$ by even reflection, and denote this extension by $\\tilde f$. Observe that $\\int_{-\\pi}^{\\pi}x\\tilde f(x)\\,dx=0$. Clearly,\n\\begin{equation}\\label{eq:sdrftild}\n(\\tilde f)^{\\#}(x)=f^{\\ast}(x),\\qquad 0\\leq x\\leq \\pi.\n\\end{equation}\nLet $u$ correspond to $f$ in the analogue of Proposition \\ref{Prop:Nuniq} over $[0,\\pi]$ and similarly let $v$ correspond to $f^{\\ast}$. Let $\\tilde u$ and $\\tilde v$ correspond to $\\tilde f$ and $(\\tilde f)^{\\#}$ in the $[-\\pi,\\pi]$ version of Proposition \\ref{Prop:Nuniq}. Proposition \\ref{prop:uisconvwf} gives $\\tilde u = K\\ast \\tilde f$ and $\\tilde v = K \\ast (\\tilde f)^{\\#}$.\n\nWe claim that $\\tilde u$ is obtained from $u$ by even reflection. We show that $\\tilde u$ is even and that $\\tilde u$ also satisfies the properties of Proposition \\ref{Prop:Nuniq} corresponding to $f$ over the interval $[0,\\pi]$. First, $\\tilde u$ is even:\n\\[\n\\tilde u(x)=\\frac{1}{2\\pi}\\int_{-\\pi}^{\\pi}K(x-y)\\tilde f(y)\\,dy=\\frac{1}{2\\pi}\\int_{-\\pi}^{\\pi}K(x+y)\\tilde f(y)\\,dy=\\frac{1}{2\\pi}\\int_{-\\pi}^{\\pi}K(-x-y)\\tilde f(y)\\,dy=\\tilde u(-x).\n\\]\nSince $\\tilde u\\in C^1[-\\pi,\\pi]$ and is even, we must have $\\tilde u'(0)=0$. Additionally, we have $\\tilde u'(\\pi)=0$ by assumption. Again, being even implies $\\int_{0}^{\\pi}\\tilde u(x)\\,dx=\\frac{1}{2}\\int_{-\\pi}^{\\pi}\\tilde u(x)\\,dx=0$. Hence by uniqueness, $\\tilde u(x)=u(x)$ on $[0,\\pi]$ and so $(\\tilde u)^{\\#}(x)=u^{\\ast}(x)$ on $[0,\\pi]$. We similarly have $\\tilde v(x)=v(x)$ on $[0,\\pi]$. By Theorem \\ref{Th:Nprelim}, we have for each $0\\leq t\\leq 2\\pi$\n\\[\n\\int_{-\\frac{t}{2}}^{\\frac{t}{2}}(\\tilde u)^{\\#}(x)\\,dx \\leq \\int_{-\\frac{t}{2}}^{\\frac{t}{2}}\\tilde v(x)\\,dx\n\\]\nwhich implies\n\\[\n\\int_0^{\\frac{t}{2}}(\\tilde u)^{\\#}(x)\\,dx\\leq \\int_0^{\\frac{t}{2}}\\tilde v(x)\\,dx\n\\]\nfinally giving\n\\[\n\\int_0^{\\frac{t}{2}}u^{\\ast}(x)\\,dx\\leq \\int_0^{\\frac{t}{2}}v(x)\\,dx.\n\\]\nThis last inequality implies $u^{\\bigstar} \\leq v^{\\bigstar}$ on $[0,\\pi]$ and the theorem's claims on convex means follows from Proposition \\ref{Prop:Majorization}. The remaining conclusions follow from Proposition \\ref{cor:Lpnormsgen} since $u$ and $v$ have zero mean by assumption.\n\\end{proof}\n\n\n", "meta": {"timestamp": "2021-01-26T02:16:14", "yymm": "2101", "arxiv_id": "2101.09600", "language": "en", "url": "https://arxiv.org/abs/2101.09600", "abstract": "We compare the solutions of two one-dimensional Poisson problems on an interval with Robin boundary conditions, one with given data, and one where the data has been symmetrized. When the Robin parameter is positive and the symmetrization is symmetric decreasing rearrangement, we prove that the solution to the symmetrized problem has larger increasing convex means. When the Robin parameter equals zero (so that we have Neumann boundary conditions) and the symmetrization is decreasing rearrangement, we similarly show that the solution to the symmetrized problem has larger convex means.", "subjects": "Analysis of PDEs (math.AP)", "title": "Extremizing Temperature Functions of Rods with Robin Boundary Conditions", "lm_name": "Qwen/Qwen-72B", "lm_label": "1. YES\n2. YES", "lm_q1_score": 0.9830850827686585, "lm_q2_score": 0.8652240704135291, "lm_q1q2_score": 0.8505888768759199}}