Q
stringlengths
27
2.03k
A
stringlengths
4
2.98k
Result
stringclasses
2 values
Example 5.8.6 As a special case of 5.8.5 we see that every Borel set \( B \subseteq X \times Y \) with \( {B}_{x} \) a dense \( {G}_{\delta } \) set admits a Borel uniformization. However, there is an \( {F}_{\sigma } \) subset \( E \) of \( \left\lbrack {0,1}\right\rbrack \times {\mathbb{N}}^{\mathbb{N}} \) with secti...
Let \( C \subseteq \left\lbrack {0,1}\right\rbrack \times {\mathbb{N}}^{\mathbb{N}} \) be a closed set with projection to the first coordinate space \( \left\lbrack {0,1}\right\rbrack \), that does not admit a Borel uniformization. Such a set exists by 5.1.7. For each \( s \in {\mathbb{N}}^{ < \mathbb{N}} \), fix a hom...
No
Theorem 5.8.7 (Blackwell and Ryll-Nardzewski [17]) Let \( X, Y \) be Polish spaces, \( P \) a transition probability on \( X \times Y \), and \( B \) a Borel subset of \( X \times Y \) such that \( P\left( {x,{B}_{x}}\right) > 0 \) for all \( x \in {\pi }_{X}\left( B\right) \) . Then \( {\pi }_{X}\left( B\right) \) is ...
Proof. Apply 5.8.4 with \( {\mathcal{I}}_{x} \) as in Example 5.8.1.
No
Theorem 5.8.8 (Blackwell and Ryll-Nardzewski) Let \( X, Y \) be Polish spaces, \( \mathcal{A} \) a countably generated sub \( \sigma \) algebra of \( {\mathcal{B}}_{X} \), and \( P \) a transition probability on \( X \times Y \) such that for every \( B \in {\mathcal{B}}_{Y}, x \rightarrow P\left( {x, B}\right) \) is \...
Proof of 5.8.8. By a slight modification of the argument contained in the proof of 3.4.24 we see that for every \( E \in \mathcal{A} \otimes {\mathcal{B}}_{Y}, x \rightarrow P\left( {x,{E}_{x}}\right) \) is \( \mathcal{A} \) -measurable. As \( {\pi }_{X}\left( B\right) = \left\{ {x \in X : P\left( {x,{B}_{x}}\right) > ...
Yes
Lemma 5.8.9 Let \( X, Y,\mathcal{A} \), and \( P \) be as above. For every \( E \in \mathcal{A}\bigotimes {\mathcal{B}}_{Y} \) and every \( \epsilon > 0 \), there is an \( F \in \mathcal{A}\bigotimes {\mathcal{B}}_{Y} \) contained in \( E \) such that \( {F}_{x} \) is compact and \( P\left( {x,{F}_{x}}\right) \geq \eps...
Proof. Let \( \mathcal{M} \) be the class of all sets in \( \mathcal{A}\bigotimes {\mathcal{B}}_{Y} \) such that the conclusion of the lemma holds for every \( P \) and every \( \epsilon > 0 \) . By 3.4.20, \( \mathcal{M} \) contains all rectangles \( A \times B \), where \( A \in \mathcal{A} \) and \( B \) Borel in \(...
Yes
Proposition 5.8.10 Let \( X, f \), and \( \mathcal{A} \) be as above. An everywhere proper conditional distribution given \( f \) exists if and only if there is an \( \mathcal{A} \) - measurable \( g : X \rightarrow X \) such that \( f\left( {g\left( x\right) }\right) = f\left( x\right) \) for all \( x \) .
Proof. Suppose an \( \mathcal{A} \) -measurable \( g : X \rightarrow X \) such that \( f \circ g \) is the identity exists. Define\n\n\[ Q\left( {x, B}\right) = \left\{ \begin{array}{ll} 1 & \text{ if }g\left( x\right) \in B \\ 0 & \text{ otherwise. } \end{array}\right. \]\n\nIt is easy to verify that \( Q \) has the d...
Yes
Proposition 5.8.13 (Feldman and Moore [41]) Every Borel equivalence relation on a Polish space \( X \) with equivalence classes countable is induced by a countable group of Borel automorphisms.
Proof. Let \( \Pi \) be a Borel equivalence relation on \( X \) with equivalence classes countable. By 5.8.11, write\n\n\[ \Pi = \mathop{\bigcup }\limits_{n}{G}_{n} \]\n\nwhere \( {\pi }_{1} \mid {G}_{n} \) is one-to-one, \( {\pi }_{1}\left( {x, y}\right) = x \) ; i.e., the \( {G}_{n} \) ’s are graphs of Borel function...
Yes
Theorem 5.9.1 (Miller [85]) Every partition \( \Pi \) of a Polish space \( X \) into \( {G}_{\delta } \) sets such that the saturation of every basic open set is simultaneously \( {F}_{\sigma } \) and \( {G}_{\delta } \) admits a section \( s : X \rightarrow X \) that is Borel measurable of class 2. In particular, such...
Proof. Let \( \left( {U}_{n}\right) \) be a countable base for the topology of \( X \) . Let \( \left( {V}_{n}\right) \) be an enumeration of \( \left\{ {{U}_{n}^{ * } : n \in \mathbb{N}}\right\} \bigcup \left\{ {{\left( {U}_{n}^{ * }\right) }^{c} : n \in \mathbb{N}}\right\} \) . Let \( {\mathcal{T}}^{\prime } \) be th...
Yes
Theorem 5.9.2 (Srivastava [114]) Every Borel measurable partition \( \mathbf{\Pi } \) of a Polish space \( X \) into \( {G}_{\delta } \) sets admits a Borel cross section.
Proof. (Kechris) For \( x \in X \) let \( \left\lbrack x\right\rbrack \) denote the member of \( \mathbf{\Pi } \) containing \( x \) . Consider the multifunction \( p : X \rightarrow X \) defined by\n\n\[ p\left( x\right) = \operatorname{cl}\left( \left\lbrack x\right\rbrack \right) \]\n\nThen \( p : X \rightarrow X \)...
Yes
Theorem 5.9.5 \( \operatorname{irr}\left( A\right) / \sim \) is standard Borel if and only if \( A \) is GCR.
Its proof makes crucial uses of 5.4.3 and 4.5.4. We refer the interested reader to [4] and [43] for a proof.
No
Theorem 5.10.1 (The reflection theorem) Let \( X \) be a Polish space and \( \Phi \subseteq \mathcal{P}\left( X\right) {\mathbf{\Pi }}_{1}^{1} \) on \( {\mathbf{\Pi }}_{1}^{1} \) . For every \( {\mathbf{\Pi }}_{1}^{1} \) set \( A \in \Phi \) there is a Borel \( B \subseteq A \) in \( \Phi \) .
Proof. Suppose there is a \( {\mathbf{\Pi }}_{1}^{1} \) set \( A \subseteq X \) in \( \Phi \) that does not contain a Borel set belonging to \( \Phi \) . We shall get a contradiction. Let \( \varphi \) be a \( {\mathbf{\Pi }}_{1}^{1} \) -norm on \( A \) and\n\n\[ C = \left\{ {\left( {x, y}\right) : y{ < }_{\varphi }^{ ...
Yes
Theorem 5.10.2 Let \( X, Y \) be Polish spaces and \( A \subseteq X \times Y \) analytic with sections \( {A}_{x} \) countable. Then every coanalytic set \( B \) containing \( A \) contains a Borel set \( E \supseteq A \) with all sections countable.
Proof. Let \( C = {B}^{c} \) . Define \( \Phi \subseteq \mathcal{P}\left( {X \times Y}\right) \) by\n\n\[ D \in \Phi \Leftrightarrow {D}^{c} \subseteq B\& \forall x\left( {\left( {D}^{c}\right) }_{x}\right. \text{is countable})\text{.}\]\n\nUsing 4.3.7 we can easily check that \( \Phi \) is \( {\mathbf{\Pi }}_{1}^{1} \...
Yes
Theorem 5.10.3 (Lusin) Every analytic set with countable sections, in the product of two Polish spaces, can be covered by countably many Borel graphs.
Proof. The result immediately follows from 5.10.2 and 5.8.11.
No
Proposition 5.10.4 (Burgess) Let \( X \) be Polish, \( E \) an analytic equivalence relation on \( X \), and \( C \subseteq X \times X \) a coanalytic set containing \( E \) . Then there is a Borel equivalence relation \( B \) such that \( E \subseteq B \subseteq C \) .
Proof of 5.10.4. Applying 5.10.5 repeatedly, by induction on \( n \) we can define a sequence of Borel sets \( \left( {B}_{n}\right) \) such that\n\n\[ E \subseteq {B}_{n} \subseteq \mathcal{E}\left( {B}_{n}\right) \subseteq {B}_{n + 1} \subseteq C \]\n\nfor all \( n \) . Take \( B = \mathop{\bigcup }\limits_{n}{B}_{n}...
Yes
Lemma 5.10.5 Let \( X \) be a Polish space, \( P \) analytic, \( C \) coanalytic, and \( \mathcal{E}\left( P\right) \subseteq C \) . Then there is a Borel set \( B \) containing \( P \) such that\n\n\[ \mathcal{E}\left( B\right) \subseteq C\text{.} \]
Proof. Define \( \Phi \subseteq \mathcal{P}\left( {X \times X}\right) \) by\n\n\[ D \in \Phi \Leftrightarrow \mathcal{E}\left( {D}^{c}\right) \subseteq C. \]\n\n\( \Phi \) is \( {\mathbf{\Pi }}_{1}^{1} \) on \( {\mathbf{\Pi }}_{1}^{1} \) . Further, \( {P}^{c} \in \Phi \) . By the reflection theorem (5.10.1), there is a...
Yes
Corollary 5.10.6 For every analytic equivalence relation \( E \) on a Polish space \( X \) there exist Borel equivalence relations \( {B}_{\alpha },\alpha < {\omega }_{1} \), such that \( E = \) \( \mathop{\bigcap }\limits_{{\alpha < {\omega }_{1}}}{B}_{\alpha }. \)
Proof. By 4.3.17, write \( E = \mathop{\bigcap }\limits_{{\alpha < {\omega }_{1}}}{C}_{\alpha },{C}_{\alpha } \) coanalytic. By 5.10.4, for each \( \alpha \) there exists a Borel equivalence relation \( {B}_{\alpha } \) such that \( E \subseteq {B}_{\alpha } \subseteq {C}_{\alpha } \) .
Yes
Theorem 5.11.4 Every countably generated sub \( \sigma \) -algebra of the Borel \( \sigma \) - algebra of a Polish space has a minimal complement.
Proof of 5.11.4. Let \( X \) be Polish and \( \mathcal{C} \) a countably generated sub \( \sigma \) -algebra of \( {\mathcal{B}}_{X} \) . \n\nCase 1. There is a cocountable atom \( A \) of \( \mathcal{C} \) .\n\nLet \( f : X \smallsetminus A \rightarrow A \) be a one-to-one map. Take\n\n\[ \mathcal{D} = \sigma \left( {...
Yes
Lemma 5.11.5 Let \( X \) be Polish and \( \mathcal{C} \) a countably generated sub \( \sigma \) -algebra of \( {\mathcal{B}}_{X} \) . Suppose \( \mathcal{D} \) is a countably generated sub \( \sigma \) -algebra of \( {\mathcal{B}}_{X} \) such that every atom \( A \) of \( \mathcal{D} \) is a partial cross section of th...
Proof. Under the hypothesis, \( \mathcal{C} \vee \mathcal{D} \) is a countably generated sub \( \sigma \) - algebra of \( {\mathcal{B}}_{X} \) with atoms singletons. Hence, by 4.5.7, \( \mathcal{C} \vee \mathcal{D} = {\mathcal{B}}_{X} \) . Let \( {\mathcal{D}}^{ * } \) be a proper countably generated sub \( \sigma \) -...
Yes
Theorem 5.12.1 (Arsenin, Kunugui [60]) Let \( B \subseteq X \times Y \) be a Borel set, \( X, Y \) Polish, such that \( {B}_{x} \) is \( \sigma \) -compact for every \( x \) . Then \( {\pi }_{X}\left( B\right) \) is Borel, and \( B \) admits a Borel uniformization.
Proof of 5.12.1. Write \( B = \mathop{\bigcup }\limits_{n}{B}_{n} \), the \( {B}_{n} \) ’s Borel with compact sections. That this can be done follows from 5.12.3. Then\n\n\[{\pi }_{X}\left( B\right) = \mathop{\bigcup }\limits_{n}{\pi }_{X}\left( {B}_{n}\right)\]\n\nSince the projection of a Borel set with compact secti...
Yes
Theorem 5.12.2 (Saint Raymond [97]) Let \( X, Y \) be Polish spaces and \( A, B \subseteq X \times Y \) analytic sets. Assume that for every \( x \), there is a \( \sigma \) -compact set \( K \) such that \( {A}_{x} \subseteq K \subseteq {B}_{x}^{c} \) . Then there exists a sequence of Borel sets \( \left( {B}_{n}\righ...
Null
No
Theorem 5.12.3 Let \( X, Y \) be Polish spaces and \( A \subseteq X \times Y \) a Borel set with sections \( {A}_{x} \) \(\sigma\) -compact. Then \( A = \mathop{\bigcup }\limits_{n}{B}_{n} \), where each \( {B}_{n} \) is Borel with \( {\left( {B}_{n}\right) }_{x} \) compact for all \( x \) and all \( n \) .
Proof. The result trivially follows from 5.12.2 by taking \( B = {A}^{c} \) .
Yes
Proposition 5.12.4 Let \( B \subseteq X \times Y \) be a Borel set with sections \( {B}_{x} \) that are \( {G}_{\delta } \) sets in \( Y \) . Then there exist Borel sets \( {B}_{n} \) with open sections such that \( B = \mathop{\bigcap }\limits_{n}{B}_{n} \) .
Proof. Let \( Z \) be a compact metric space containing (a homeomorph of) \( Y \) . Then \( B \) is Borel in \( X \times Z \) with sections \( {G}_{\delta } \) sets (2.2.7). By 5.12.3, there exist Borel sets \( {C}_{n} \) in \( X \times Z \) with sections compact such that \( \left( {X \times Z}\right) \smallsetminus B...
Yes
Corollary 5.12.5 Let \( B \subseteq X \times Y \) be a Borel set with sections \( {B}_{x} \) that are \( {F}_{\sigma } \) sets in \( Y \) . Then there exist Borel sets \( {B}_{n} \) with closed sections such that \( B = \mathop{\bigcup }\limits_{n}{B}_{n} \)
Null
No
Proposition 5.12.6 Let \( X \) be a Polish space and \( \mathcal{B} \subseteq F\left( X\right) \) hereditary. Then \( {\Omega }_{{D}_{\mathcal{B}}} = {\mathcal{B}}_{\sigma } \cap F\left( X\right) \) .
Proof. Fix a closed set \( A \subseteq X \) and a countable base \( \left( {U}_{n}\right) \) for \( X \) .\n\nLet \( {D}^{\infty }\left( A\right) = \varnothing \) . Then\n\n\[ A = \mathop{\bigcup }\limits_{{\alpha < {\left| A\right| }_{D}}}\left( {{D}^{\alpha }\left( A\right) \smallsetminus {D}^{\alpha + 1}\left( A\rig...
Yes
Proposition 5.12.7 Let \( X \) be Polish and \( D \) a derivative on \( X \) such that\n\n\[ \n\\{ \\left( {A, B}\\right) \\in F\\left( X\\right) \\times F\\left( X\\right) : A \\subseteq D\\left( B\\right) \\} \n\]\n\nis analytic. Then\n\n(i) \( {\\Omega }_{D} \) is coanalytic, and\n\n(ii) for every analytic \( \\math...
Proof. Assertion (i) follows from the following equivalence:\n\n\[ \nA \\notin {\\Omega }_{D} \\Leftrightarrow \\exists B\\left( {B \\neq \\varnothing \\& B \\subseteq A\\& B \\subseteq D\\left( B\\right) }\\right) .\n\]\n\n(The sets \( A \) and \( B \) are closed in \( X \).)\n\nSuppose (ii) is false for some analytic...
Yes
Proposition 5.12.10 Let \( X \) and \( Y \) be Polish spaces and \( A, B \) two disjoint analytic subsets of \( X \times Y \) such that \( {A}_{x} \) is closed and nowhere dense for all \( x \) . Then there is a Borel \( C \subseteq X \times Y \) such that the sections \( {C}_{x} \) are closed and nowhere dense, and su...
Null
No
Proposition 5.12.11 (i) (Hillard [48]) Let \( X \) and \( Y \) be Polish spaces and \( A, B \) disjoint analytic subsets of \( X \times Y \) . Assume that the sections \( {A}_{x} \) are meager in \( Y \) . Then there is a sequence \( \left( {C}_{n}\right) \) of Borel sets with sections nowhere dense such that\n\n\[ A \...
Null
No
Theorem 5.12.12 (A. Louveau [66]) For every \( 1 \leq \alpha < {\omega }_{1},{\mathbf{\sum }}_{\alpha }^{ * } = {\mathcal{F}}_{\alpha } \) .
Null
No
Theorem 5.13.1 (Lopez-Escobar) A subset \( A \) of \( {X}_{L} \) is invariant (with respect to the logic action) and Borel, if and only if there is a sentence \( \sigma \) of \( {L}_{{\omega }_{1}\omega } \) such that \( A = {A}_{\sigma } \) .
Proof. The sufficient part of this result is proved by induction on formulae of \( {L}_{{\omega }_{1}\omega } \) as follows:\n\nFor every formula \( \phi \left\lbrack {{v}_{0},{v}_{1},\ldots ,{v}_{k - 1}}\right\rbrack \), the set\n\n\[ \n{A}_{\phi, k} = \left\{ {\left( {x,{n}_{0},{n}_{1},\ldots ,{n}_{k - 1}}\right) : {...
No
Theorem 5.13.2 (Becker - Kechris) Suppose a Polish group \( G \) acts continuously on a Polish space \( X \) and \( A \) is an invariant Borel subset of \( X \) . Then there is a finer Polish topology on \( X \) making \( A \) clopen such that the action still remains continuous.
Null
No
Theorem 5.13.3 (Becker - Kechris) Suppose a Polish group \( G \) acts on a Polish space \( X \) and the action is Borel. Then there is a finer Polish topology on \( X \) making the action continuous.
Null
No
Theorem 5.13.4 (Burgess) Suppose \( E \) is an analytic equivalence relation on a Polish space \( X \) . Then the number of equivalence classes is \( \leq {\aleph }_{1} \) or perfectly many.
Null
No
Theorem 5.13.8 Topological Vaught conjecture holds if \( G \) is a locally compact Polish group.
Assuming 5.13.9, we prove 5.13.8 as follows: Let \( G \) be a locally compact Polish group acting continuously on a Polish space \( X \) . Write \( G = \mathop{\bigcup }\limits_{n}{K}_{n} \) , \( {K}_{n} \) compact. Then, for \( x, y \in X \) ,\n\n\[ \exists g \in G\left( {y = g \cdot x}\right) \Leftrightarrow \exists ...
Yes
Theorem 5.13.9 Let \( E \) be an analytic equivalence relation on a Polish space \( X \) with all equivalence classes \( {F}_{\sigma } \) . Then the number of equivalence classes is \( \leq {\aleph }_{0} \) or perfectly many.
Proof of 5.13.9. Let \( X \) be a Polish space and \( E \) an analytic equivalence relation on \( X \) with all its equivalence classes \( {F}_{\sigma } \) sets. Further assume that there are uncountably many \( E \) -equivalence classes. Fix a countable base \( \left( {V}_{n}\right) \) for the topology of \( X \) . Le...
No
Proposition 5.13.10 Suppose \( X \) is a Polish space and \( E \) an equivalence relation on \( X \) which is meager in \( {X}^{2} \) . Then \( E \) has perfectly many equivalence classes.
Proof. Let \( E \subseteq \mathop{\bigcup }\limits_{n}{F}_{n},{F}_{n} \) closed and nowhere dense in \( {X}^{2} \) . Without any loss of generality, we further assume that the diagonal \( \left\{ {\left( {x, y}\right) \in {X}^{2} : x = y}\right\} \) is contained in each of \( {F}_{n} \) .\n\nFor each \( s \in {2}^{ < \...
Yes
Theorem 5.13.11 (Silver's theorem) Suppose \( E \) is a coanalytic equivalence relation on a Polish space \( X \) . Then the number of equivalence classes is countable or perfectly many.
Null
No
Theorem 5.13.12 (Sami) Topological Vaught conjecture holds if \( G \) is abelian.
Proof. Assume that the number of orbits is uncountable. We shall show that there is a perfect set of inequivalent elements.\n\nLet \( E \) be the equivalence relation on \( X \) defined by\n\n\[ \n{xEy} \Leftrightarrow {G}_{x} = {G}_{y} \n\]\n\nwhere \( {G}_{x} \) is the stabilizer of \( x \) . Let \( y = g \cdot x \) ...
Yes
Lemma 5.13.14 Suppose \( \left\{ {{A}_{\alpha } : \alpha < {\omega }_{1}}\right\} \) is a family of Borel subsets of a Polish space \( X \) and \( E \) the equivalence relation on \( X \) defined by\n\n\[ \n{xEy} \Leftrightarrow \forall \alpha \left( {x \in {A}_{\alpha } \Leftrightarrow y \in {A}_{\alpha }}\right), x, ...
Proof of 5.13.14. Although the proof of the lemma is messy looking, ideawise it is quite simple. Assume that the number of \( E \) -equivalence classes is \( > {\aleph }_{1} \) . We shall then show that there are perfectly many \( E \) -equivalence classes. The following fact will be used repeatedly in the proof of the...
Yes
Example 5.13.15 Let \( L \) be a first order language whose non-logical symbols consists of exactly one binary relation symbol. So, \( {X}_{L} = {2}^{\omega \times \omega } \) . We claim that in this case the equivalence relation \( {E}_{a} \) induced by the logic action is not Borel. Suppose not. Then \( {E}_{a} \in {...
It follows that \( W{O}^{\alpha } = \{ x \in {WO} : \left| x\right| \leq \alpha \} \in {\mathbf{\sum }}_{\beta }^{0} \) for every \( \alpha < {\omega }_{1} \) . Now take any Borel set \( A \) in \( {\mathbb{N}}^{\mathbb{N}} \) which is not of additive class \( \beta \) . Since \( {WO} \) is \( {\mathbf{\Pi }}_{1}^{1} \...
Yes
Theorem 5.13.18 (Stern) Let \( E \) be an analytic equivalence relation on a Polish space \( X \) such that all but countably many equivalence classes are \( {F}_{\sigma } \) or \( {G}_{\delta } \) . The the number of equivalence classes is \( \leq {\aleph }_{0} \) or perfectly many.
Null
No
Theorem 5.13.19 (Stern) Assume analytic determinacy. Let \( E \) be an analytic equivalence relation on a Polish space \( X \) such that all but countably many equivalence classes are of bounded Borel rank. Then the number of equivalence classes is \( \leq {\aleph }_{0} \) or perfectly many.
The proof this result is beyond the scope of this book.
No
Theorem 5.14.1 (Kondô’s theorem) Let \( X, Y \) be Polish spaces. Every coanalytic set \( C \subseteq X \times Y \) admits a coanalytic uniformization.
We shall show that there is a sequence of coanalytic norms on a given co-analytic set with certain \
No
Theorem 5.14.4 Every coanalytic subset of \( {\mathbb{N}}^{\mathbb{N}} \) admits a very good \( {\mathbf{\Pi }}_{1}^{1} \) - scale.
Null
No
Corollary 5.14.5 Let \( X \) be a Polish space and \( A \subseteq X \) coanalytic. Then A admits a very good \( {\mathbf{\Pi }}_{1}^{1} \) -scale.
Proof. By 2.6.9 there is a closed set \( D \subseteq {\mathbb{N}}^{\mathbb{N}} \) and a continuous bijection \( f : D \rightarrow X \) . Now, \( {f}^{-1}\left( A\right) \cap D \) is a \( {\mathbf{\Pi }}_{1}^{1} \) subset of \( {\mathbb{N}}^{\mathbb{N}} \) and hence admits a very good \( {\mathbf{\Pi }}_{1}^{1} \) -scal...
Yes
We consider the discretization of the boundary value problem for the ordinary differential equation\n\n\[ - {u}^{\prime \prime }\left( x\right) = f\left( {x, u\left( x\right) }\right) ,\;x \in \left\lbrack {0,1}\right\rbrack \]\n\n(2.1)\n\nwith boundary condition\n\n\[ u\left( 0\right) = u\left( 1\right) = 0. \]\n\n(2....
For the approximate solution we choose an equidistant subdivision of the interval \( \left\lbrack {0,1}\right\rbrack \) by setting\n\n\[ {x}_{j} = {jh},\;j = 0,\ldots, n + 1, \]\n\nwhere the step size is given by \( h = 1/\left( {n + 1}\right) \) with \( n \in \mathbb{N} \). At the internal grid points \( {x}_{j}, j = ...
Yes
Consider the linear integral equation\n\n\[ \varphi \left( x\right) - {\int }_{0}^{1}K\left( {x, y}\right) \varphi \left( y\right) {dy} = f\left( x\right) ,\;x \in \left\lbrack {0,1}\right\rbrack ,\]
For the numerical approximation we replace the integral by the rectangular sum\n\n\[ {\int }_{0}^{1}K\left( {x, y}\right) \varphi \left( y\right) {dy} \approx \frac{1}{n}\mathop{\sum }\limits_{{k = 1}}^{n}K\left( {x,{x}_{k}}\right) \varphi \left( {x}_{k}\right) \]\n\nwith equidistant grid points \( {x}_{k} = k/n, k = 1...
Yes
Consider some (physical) quantity \( u \) depending on time \( t \) and a parameter vector \( a = {\left( {a}_{1},\ldots ,{a}_{n}\right) }^{T} \in {\mathbb{R}}^{n} \) in terms of a known function\n\n\[ u\left( t\right) = f\left( {t;a}\right) \]\n\nIn order to determine the values of the parameter \( a \) (representing ...
The necessary conditions for a minimum,\n\n\[ \frac{\partial g}{\partial {a}_{j}} = 0,\;j = 1,\ldots, n \]\n\nlead to the normal equations\n\n\[ \mathop{\sum }\limits_{{k = 1}}^{m}\left\lbrack {u\left( {t}_{k}\right) - f\left( {{t}_{k};a}\right) }\right\rbrack \frac{\partial f\left( {{t}_{k};a}\right) }{\partial {a}_{j...
Yes
We consider the system\n\n\[ \n{x}_{1} + {200}{x}_{2} = {100} \]\n\n\[ \n{x}_{1} + \;{x}_{2} = 1 \]\n\nwith the exact solution \( {x}_{1} = {100}/{199} = {0.502}\ldots ,{x}_{2} = {99}/{199} = {0.497}\ldots \) .
For the following computations we use two-decimal-digit floating-point arithmetic. Column pivoting leads to \( {a}_{11} \) as pivot element, and the elimination yields\n\n\[ \n{x}_{1} + {200}{x}_{2} = {100} \]\n\n\[ \n- {200}{x}_{2} = - {99} \]\n\nsince \( {199} = {200} \) in two-digit floating-point representation. Fr...
Yes
Theorem 2.7 Gaussian elimination for the simultaneous solution of an \( n \times n \) system for \( r \) different right-hand sides requires a total of\n\n\[ \frac{{n}^{3}}{3} + r{n}^{2} - \frac{n}{3} \]\n\nmultiplications.
Null
No
Theorem 2.9 For a nonsingular matrix \( A \), Gaussian elimination (without reordering rows and columns) yields an LR decomposition.
Proof. In the first elimination step we multiply the first equation by \( {a}_{j1}/{a}_{11} \) and subtract the result from the \( j \) th equation; i.e., the matrix \( {A}_{1} = A \) is multiplied from the left by the lower triangular matrix\n\n\[ \n{L}_{1} = \left( \begin{array}{rrrr} 1 & & & \\ - \frac{{a}_{21}}{{a}...
Yes
Theorem 2.13 To each \( n \times n \) matrix a QR decomposition can be obtained through \( n - 1 \) Householder transformations.
Null
No
Example 3.2 Some examples of norms on \( {\mathbb{R}}^{n} \) and \( {\mathbb{C}}^{n} \) are given by\n\n\[ \n\parallel x{\parallel }_{1} \mathrel{\text{:=}} \mathop{\sum }\limits_{{j = 1}}^{n}\left| {x}_{j}\right| ,\;\parallel x{\parallel }_{2} \mathrel{\text{:=}} {\left( \mathop{\sum }\limits_{{j = 1}}^{n}{\left| {x}_...
Null
No
Theorem 3.5 The limit of a convergent sequence is uniquely determined.
Proof. Assume that \( {x}_{n} \rightarrow x \) and \( {x}_{n} \rightarrow y \) for \( n \rightarrow \infty \) . Then from the triangle inequality we obtain that\n\n\[ \parallel x - y\parallel = \begin{Vmatrix}{x - {x}_{n} + {x}_{n} - y}\end{Vmatrix} \leq \begin{Vmatrix}{x - {x}_{n}}\end{Vmatrix} + \begin{Vmatrix}{{x}_{...
Yes
Theorem 3.7 Two norms \( \parallel \cdot {\parallel }_{a} \) and \( \parallel \cdot {\parallel }_{b} \) on a linear space \( X \) are equivalent if and only if there exist positive numbers \( c \) and \( C \) such that\n\n\[ c\parallel x{\parallel }_{a} \leq \parallel x{\parallel }_{b} \leq C\parallel x{\parallel }_{a}...
Proof. Provided that the conditions are satisfied, from \( {\begin{Vmatrix}{x}_{n} - x\end{Vmatrix}}_{a} \rightarrow 0 \) , \( n \rightarrow \infty \), it follows that \( {\begin{Vmatrix}{x}_{n} - x\end{Vmatrix}}_{b} \rightarrow 0, n \rightarrow \infty \), and vice versa.\n\nConversely, let the two norms be equivalent ...
Yes
Theorem 3.8 On a finite-dimensional linear space all norms are equivalent.
Proof. In a linear space \( X \) with finite dimension \( n \) and basis \( {u}_{1},\ldots ,{u}_{n} \) every element can be expressed in the form\n\n\[ x = \mathop{\sum }\limits_{{j = 1}}^{n}{\alpha }_{j}{u}_{j} \]\n\nAs in Example 3.2,\n\n\[ \parallel x{\parallel }_{\infty } \mathrel{\text{:=}} \mathop{\max }\limits_{...
Yes
Theorem 3.11 Any bounded sequence in a finite-dimensional normed space \( X \) contains a convergent subsequence.
Proof. Let \( {u}_{1},\ldots ,{u}_{n} \) be a basis of \( X \) and let \( \left( {x}_{\nu }\right) \) be a bounded sequence. Then writing\n\n\[ \n{x}_{\nu } = \mathop{\sum }\limits_{{j = 1}}^{n}{\alpha }_{j\nu }{u}_{j} \n\] \n\nand using the norm (3.2), as in the proof of Theorem 3.8 we deduce that each of the sequence...
Yes
Theorem 3.14 For a scalar product we have the Cauchy-Schwarz inequality\n\n\[ \n{\left| \left( x, y\right) \right| }^{2} \leq \left( {x, x}\right) \left( {y, y}\right) \n\]\n\nfor all \( x, y \in X \), with equality if and only if \( x \) and \( y \) are linearly dependent.
Proof. The inequality is trivial for \( x = 0 \) . For \( x \neq 0 \) it follows from\n\n\[ \n\left( {{\alpha x} + {\beta y},{\alpha x} + {\beta y}}\right) = {\left| \alpha \right| }^{2}\left( {x, x}\right) + 2\operatorname{Re}\{ \alpha \bar{\beta }\left( {x, y}\right) \} + {\left| \beta \right| }^{2}\left( {y, y}\righ...
Yes
Theorem 3.14 For a scalar product we have the Cauchy-Schwarz inequality\n\n\\[ \n{\\left| \\left( x, y\\right) \\right| }^{2} \\leq \\left( {x, x}\\right) \\left( {y, y}\\right) \n\\]\n\nfor all \\( x, y \\in X \\), with equality if and only if \\( x \\) and \\( y \\) are linearly dependent.
Proof. The inequality is trivial for \\( x = 0 \\) . For \\( x \\neq 0 \\) it follows from\n\n\\[ \n\\left( {{\\alpha x} + {\\beta y},{\\alpha x} + {\\beta y}}\\right) = {\\left| \\alpha \\right| }^{2}\\left( {x, x}\\right) + 2\\operatorname{Re}\\{ \\alpha \\bar{\\beta }\\left( {x, y}\\right) \\} + {\\left| \\beta \\ri...
Yes
Theorem 3.15 A scalar product \( \left( {\cdot , \cdot }\right) \) on a linear space \( X \) defines a norm by\n\n\[ \parallel x\parallel \mathrel{\text{:=}} {\left( x, x\right) }^{1/2} \]\n\nfor all \( x \in X \) ; i.e., a pre-Hilbert space is always a normed space.
Proof. We leave it as an exercise for the reader to verify the norm axioms. The triangle inequality follows by\n\n\[ \parallel x + y{\parallel }^{2} = \left( {x + y, x + y}\right) \leq \parallel x{\parallel }^{2} + 2\parallel x\parallel \parallel y\parallel + \parallel y{\parallel }^{2} = {\left( \parallel x\parallel +...
No
Theorem 3.17 The elements of an orthonormal system are linearly independent.
Proof. From\n\n\[ \mathop{\sum }\limits_{{k = 1}}^{n}{\alpha }_{k}{q}_{k} = 0 \]\n\nfor the orthonormal system \( \left\{ {{q}_{1},\ldots ,{q}_{n}}\right\} \), by taking the scalar product with \( {q}_{j} \), we immediately have that \( {\alpha }_{j} = 0 \) for \( j = 1,\ldots, n \) .
Yes
Theorem 3.18 Let \( \left\{ {{u}_{0},{u}_{1},\ldots }\right\} \) be a finite or countable number of linearly independent elements of a pre-Hilbert space. Then there exists a uniquely determined orthogonal system \( \left\{ {{q}_{0},{q}_{1},\ldots }\right\} \) of the form\n\n\[ \n{q}_{n} = {u}_{n} + {r}_{n},\;n = 0,1,\l...
Proof. Assume that we have constructed orthogonal elements of the form (3.3) with the property (3.4) up to \( {q}_{n - 1} \) . By (3.4), the \( \left\{ {{q}_{0},\ldots ,{q}_{n - 1}}\right\} \) are linearly independent, and therefore \( \begin{Vmatrix}{q}_{k}\end{Vmatrix} \neq 0 \) for \( k = 0,1,\ldots, n - 1 \) . Henc...
Yes
Theorem 3.21 A linear operator is continuous if it is continuous at one element.
Proof. Let \( A : X \rightarrow Y \) be continuous at \( {x}_{0} \in X \) . Then for every \( x \in X \) and every sequence \( \left( {x}_{n}\right) \) with \( {x}_{n} \rightarrow x, n \rightarrow ∞ \), we have\n\n\[ A{x}_{n} = A\left( {{x}_{n} - x + {x}_{0}}\right) + A\left( {x - {x}_{0}}\right) \rightarrow A\left( {x...
Yes
Theorem 3.23 A linear operator \( A : X \rightarrow Y \) is bounded if and only if\n\n\[ \parallel A\parallel \mathrel{\text{:=}} \mathop{\sup }\limits_{{\parallel x\parallel = 1}}\parallel {Ax}\parallel < \infty . \]\n\nThe number \( \parallel A\parallel \) is the smallest bound for \( A \) and is called the norm of \...
Proof. Assume that \( A \) is bounded with the bound \( C \) . Then\n\n\[ \mathop{\sup }\limits_{{\parallel x\parallel = 1}}\parallel {Ax}\parallel \leq C \]\n\nand, in particular, \( \parallel A\parallel \) is less than or equal to any bound for \( A \) . Conversely, if \( \parallel A\parallel < \infty \), then using ...
Yes
Theorem 3.24 A linear operator is continuous if and only if it is bounded.
Proof. Let \( A : X \rightarrow Y \) be bounded and let \( \left( {x}_{n}\right) \) be a sequence in \( X \) with \( {x}_{n} \rightarrow 0, n \rightarrow \infty \) . Then from \( \begin{Vmatrix}{A{x}_{n}}\end{Vmatrix} \leq C\begin{Vmatrix}{x}_{n}\end{Vmatrix} \) it follows that \( A{x}_{n} \rightarrow 0 \) , \( n \righ...
Yes
Theorem 3.27 To each matrix \( A \) there exists a unitary matrix \( Q \) such that \( {Q}^{ * }{AQ} \) is an upper triangular matrix.
Proof. Assume that it has been shown that for each \( \left( {n - 1}\right) \times \left( {n - 1}\right) \) matrix \( {A}_{n - 1} \) there exists a unitary \( \left( {n - 1}\right) \times \left( {n - 1}\right) \) matrix \( {Q}_{n - 1} \) such that \( {Q}_{n - 1}^{ * }{A}_{n - 1}{Q}_{n - 1} \) is an upper triangular mat...
Yes
Lemma 3.28 For an \( n \times n \) matrix \( A \) and its adjoint \( {A}^{ * } \) we have that\n\n\[ \left( {{Ax}, y}\right) = \left( {x,{A}^{ * }y}\right) \]\n\nfor all \( x, y \in {\mathbb{C}}^{n} \), where \( \left( {\cdot , \cdot }\right) \) denotes the Euclidean scalar product.
Proof. Simple calculations yield\n\n\[ \left( {{Ax}, y}\right) = \mathop{\sum }\limits_{{j = 1}}^{n}{\left( Ax\right) }_{j}{\bar{y}}_{j} = \mathop{\sum }\limits_{{j = 1}}^{n}\mathop{\sum }\limits_{{k = 1}}^{n}{a}_{jk}{x}_{k}{\bar{y}}_{j} \]\n\n\[ = \mathop{\sum }\limits_{{k = 1}}^{n}\mathop{\sum }\limits_{{j = 1}}^{n}{...
Yes
The eigenvalues of a Hermitian \( n \times n \) matrix are real, and the eigenvectors form an orthogonal basis in \( {\mathbb{C}}^{n} \) .
Proof. If \( A \) is Hermitian, i.e., if \( A = {A}^{ * } \), then the matrix \( \widetilde{A} \mathrel{\text{:=}} {Q}^{ * }{AQ} \) from Theorem 3.27 is also Hermitian, since\n\n\[ \n{\widetilde{A}}^{ * } = {\left( {Q}^{ * }AQ\right) }^{ * } = {Q}^{ * }{A}^{ * }{Q}^{* * } = {Q}^{ * }{AQ} = \widetilde{A}.\n\]\n\nTherefo...
Yes
For an \( n \times n \) matrix \( A \) we have\n\n\[ \parallel A{\parallel }_{2} = \sqrt{\rho \left( {{A}^{ * }A}\right) } \]\n\nIf \( A \) is Hermitian, then\n\n\[ \parallel A{\parallel }_{2} = \rho \left( A\right) \]
Proof. From Lemma 3.28 we have that\n\n\[ \parallel {Ax}{\parallel }_{2}^{2} = \left( {{Ax},{Ax}}\right) = \left( {x,{A}^{ * }{Ax}}\right) \]\n\nfor all \( x \in {\mathbb{C}}^{n} \) . Hence the Hermitian matrix \( {A}^{ * }A \) is positive semidefinite and therefore has \( n \) orthonormal eigenvectors\n\n\[ {A}^{ * }A...
Yes
Theorem 3.32 For each norm on \( {\mathbb{C}}^{n} \) and each \( n \times n \) matrix \( A \) we have that\n\n\[ \rho \left( A\right) \leq \parallel A\parallel \]\n\nConversely, to each matrix \( A \) and each \( \varepsilon > 0 \) there exists a norm on \( {\mathbb{C}}^{n} \) such that\n\n\[ \parallel A\parallel \leq ...
Proof. Let \( \lambda \) be an eigenvalue of \( A \) with eigenvector \( u \) . We may assume that \( \parallel u\parallel = 1 \) . Then the first part of the theorem follows from\n\n\[ \parallel A\parallel = \mathop{\sup }\limits_{{\parallel x\parallel = 1}}\parallel {Ax}\parallel \geq \parallel {Au}\parallel = \paral...
Yes
Theorem 3.34 Every convergent sequence is a Cauchy sequence.
Proof. Let \( {x}_{n} \rightarrow x, n \rightarrow \infty \) . Then, for \( \varepsilon > 0 \) there exists \( N\left( \varepsilon \right) \in \mathbb{N} \) such that \( \begin{Vmatrix}{{x}_{n} - x}\end{Vmatrix} < \varepsilon /2 \) for all \( n \geq N\left( \varepsilon \right) \) . Now the triangle inequality yields\n\...
Yes
Example 3.36 The linear space \( C\left\lbrack {a, b}\right\rbrack \) furnished with the maximum norm\n\n\[ \parallel f{\parallel }_{\infty } \mathrel{\text{:=}} \mathop{\max }\limits_{{x \in \left\lbrack {a, b}\right\rbrack }}\left| {f\left( x\right) }\right| \]\n\nis a Banach space.
Proof. The norm axioms (N1)-(N3) are trivially satisfied. The triangle inequality follows from\n\n\[ \parallel f + g{\parallel }_{\infty } = \mathop{\max }\limits_{{x \in \left\lbrack {a, b}\right\rbrack }}\left| {\left( {f + g}\right) \left( x\right) }\right| = \left| {\left( {f + g}\right) \left( {x}_{0}\right) }\rig...
Yes
The linear space \( C\left\lbrack {a, b}\right\rbrack \) equipped with the \( {L}_{1} \) norm\n\n\[ \parallel f{\parallel }_{1} \mathrel{\text{:=}} {\int }_{a}^{b}\left| {f\left( x\right) }\right| {dx} \]\n\nis not complete.
The norm axioms are trivially satisfied. Without loss of generality we take \( \left\lbrack {a, b}\right\rbrack = \left\lbrack {0,2}\right\rbrack \) and choose\n\n\[ {f}_{n}\left( x\right) \mathrel{\text{:=}} \left\{ \begin{array}{ll} {x}^{n}, & 0 \leq x \leq 1 \\ 1, & 1 \leq x \leq 2 \end{array}\right. \]\n\nThen for ...
Yes
Example 3.38 The linear space \( C\left\lbrack {a, b}\right\rbrack \) equipped with the \( {L}_{2} \) norm\n\n\[ \n\parallel f{\parallel }_{2} \mathrel{\text{:=}} {\left( {\int }_{a}^{b}{\left| f\left( x\right) \right| }^{2}dx\right) }^{1/2} \n\]\n\nis not complete.
Proof. The norm is generated by the scalar product\n\n\[ \n\left( {f, g}\right) \mathrel{\text{:=}} {\int }_{a}^{b}f\left( x\right) g\left( x\right) {dx}. \n\]\n\nConsidering the same sequence as in Example 3.37, it can be seen that \( C\left\lbrack {a, b}\right\rbrack \) also is not complete with respect to the \( {L}...
Yes
Theorem 3.39 Each finite-dimensional normed space is a Banach space.
Proof. Let \( X \) be finite-dimensional with basis \( {u}_{1},\ldots ,{u}_{n} \) and assume that \( \left( {x}_{\nu }\right) \) is a Cauchy sequence in \( X \) . We represent\n\n\[ \n{x}_{\nu } = \mathop{\sum }\limits_{{j = 1}}^{n}{\alpha }_{j\nu }{u}_{j} \n\] \n\nand recall from Theorem 3.8 that there exists \( C > 0...
Yes
Theorem 3.44 Each contraction operator has at most one fixed point.
Proof. Assume that \( x \) and \( y \) are two different fixed points of the contraction operator \( A \) . Then\n\n\[ 0 \neq \parallel x - y\parallel = \parallel {Ax} - {Ay}\parallel \leq q\parallel x - y\parallel \]\n\nwhence \( 1 \leq q \) follows. This is a contradiction to the fact that \( A \) is a contraction op...
Yes
Theorem 3.45 (Banach) Let \( U \) be a complete subset of a normed space \( X \) and let \( A : U \rightarrow U \) be a contraction operator. Then \( A \) has a unique fixed point.
Proof. Starting from an arbitrary element \( {x}_{0} \in U \) we define a sequence \( \left( {x}_{n}\right) \) in \( U \) by the recursion\n\n\[ \n{x}_{n + 1} \mathrel{\text{:=}} A{x}_{n},\;n = 0,1,2,\ldots \n\]\n\nThen we have\n\n\[ \n\begin{Vmatrix}{{x}_{n + 1} - {x}_{n}}\end{Vmatrix} = \begin{Vmatrix}{A{x}_{n} - A{x...
Yes
Theorem 3.46 Let \( A \) be a contraction operator with contraction constant \( q \) mapping a complete subset \( U \) of a normed space \( X \) into itself. Then the successive approximations\n\n\[ \n{x}_{n + 1} \mathrel{\text{:=}} A{x}_{n},\;n = 0,1,2,\ldots ,\n\]\n\nwith arbitrary \( {x}_{0} \in U \) converge to the...
Proof. The a priori error estimate follows from (3.12) by passing to the limit \( m \rightarrow \infty \) . The a posteriori estimate follows from the a priori estimate applied with starting element \( {x}_{0} = {x}_{n - 1} \) .
No
Example 3.47 The function \( f : \lbrack 0,\infty ) \rightarrow \lbrack 0,\infty ) \) given by\n\n\[ f\left( x\right) \mathrel{\text{:=}} x + \frac{1}{1 + x} \]\n\nas a consequence of\n\n\[ f\left( x\right) - f\left( y\right) = \frac{x + y + {xy}}{1 + x + y + {xy}}\left( {x - y}\right) \]\n\nfulfills the condition\n\n\...
Null
No
Theorem 3.48 Let \( B : X \rightarrow X \) be a bounded linear operator on a Banach space \( X \) with \( \parallel B\parallel < 1 \), and let \( I : X \rightarrow X \) denote the identity operator. Then \( I - B \) is bijective; i.e., for each \( z \in X \) the equation\n\n\[ x - {Bx} = z \]\n\nhas a unique solution \...
Proof. For fixed, but arbitrary, \( z \in X \) we define the operator \( A : X \rightarrow X \) by\n\n\[ {Ax} \mathrel{\text{:=}} {Bx} + z,\;x \in X. \]\n\nThen we have\n\n\[ \parallel {Ax} - {Ay}\parallel = \parallel B\left( {x - y}\right) \parallel \leq \parallel B\parallel \parallel x - y\parallel \]\n\nfor all \( x...
Yes
Theorem 3.50 Let \( U \) be a finite-dimensional subspace of a normed space \( X \) . Then for every element in \( X \) there exists a best approximation with respect to \( U \) .
Proof. Let \( w \in X \) and choose a minimizing sequence \( \left( {u}_{n}\right) \) for \( w \) ; i.e., \( {u}_{n} \in U \) satisfies\n\n\[\n\begin{Vmatrix}{w - {u}_{n}}\end{Vmatrix} \rightarrow d \mathrel{\text{:=}} \mathop{\inf }\limits_{{u \in U}}\parallel w - u\parallel ,\;n \rightarrow \infty .\n\]\n\nBecause of...
Yes
Theorem 3.51 Let \( U \) be a linear subspace of a pre-Hilbert space \( X \) . An element \( v \) is a best approximation to \( w \in X \) with respect to \( U \) if and only if\n\n\[ \left( {w - v, u}\right) = 0 \]\n\n(3.13)\n\nfor all \( u \in U \), i.e., if and only if \( w - v \bot U \) . To each \( w \in X \) ther...
Proof. We begin by noting the equality\n\n\[ \parallel w - u{\parallel }^{2} = \parallel w - v{\parallel }^{2} + 2\operatorname{Re}\left( {w - v, v - u}\right) + \parallel v - u{\parallel }^{2}, \]\n\n(3.14)\n\nwhich is valid for all \( u, v \in U \) . From this, sufficiency of the condition (3.13) is obvious, since \(...
Yes
Theorem 3.52 Let \( U \) be a complete linear subspace of a pre-Hilbert space \( X \) . Then to each element \( w \in X \) there exists a unique best approximation with respect to \( U \) . The operator \( P : X \rightarrow U \) mapping \( w \in X \) onto its best approximation is a bounded linear operator with the pro...
Proof. Choose a sequence \( \left( {u}_{n}\right) \) with\n\n\[ \n{\begin{Vmatrix}w - {u}_{n}\end{Vmatrix}}^{2} \leq {d}^{2} + \frac{1}{n},\;n \in \mathbb{N},\n\]\n\n(3.15)\n\nwhere \( d \mathrel{\text{:=}} \mathop{\inf }\limits_{{u \in U}}\parallel w - u\parallel \) . Then\n\n\[ \n\parallel \left( {w - {u}_{n}}\right)...
Yes
Corollary 3.53 Let \( U \) be a finite-dimensional linear subspace of a pre-Hilbert space \( X \) with basis \( {u}_{1},\ldots ,{u}_{n} \) . The linear combination\n\n\[ v = \mathop{\sum }\limits_{{k = 1}}^{n}{\alpha }_{k}{u}_{k} \]\n\nis the best approximation to \( w \in X \) with respect to \( U \) if and only if th...
Proof. The normal equations (3.16) obviously are equivalent to (3.13).
No
Corollary 3.54 Let \( U \) be a finite-dimensional linear subspace of a pre-Hilbert space \( X \) with orthonormal basis \( {u}_{1},\ldots ,{u}_{n} \) . Then the orthogonal projection operator is given by\n\n\[ \n{Pw} = \mathop{\sum }\limits_{{k = 1}}^{n}\left( {w,{u}_{k}}\right) {u}_{k},\;w \in X.\n\]
Proof. This is trivial from either the orthogonality condition of Theorem 3.51 or the normal equations of Corollary 3.53.
Yes
Theorem 4.1 Let \( B \) be an \( n \times n \) matrix. Then the successive approximations\n\n\[ \n{x}_{\nu + 1} \mathrel{\text{:=}} B{x}_{\nu } + z,\;\nu = 0,1,2,\ldots ,\n\]\n\nconverge for each \( z \in {\mathbb{C}}^{n} \) and each \( {x}_{0} \in {\mathbb{C}}^{n} \) if and only if\n\n\[ \n\rho \left( B\right) < 1\n\]...
Proof. If \( \rho \left( B\right) < 1 \), then by Theorem 3.32 there exists a norm \( \parallel \cdot \parallel \) on \( {\mathbb{C}}^{n} \) such that \( \parallel B\parallel < 1 \) . Now convergence follows from Theorem 3.48 together with the equivalence of all norms on \( {\mathbb{C}}^{n} \) according to Theorem 3.8....
Yes
Theorem 4.2 Assume that the matrix \( A = \left( {a}_{jk}\right) \) satisfies\n\n\[ \n{q}_{\infty } \mathrel{\text{:=}} \mathop{\max }\limits_{{j = 1,\ldots, n}}\mathop{\sum }\limits_{\substack{{k = 1} \\ {k \neq j} }}^{n}\left| \frac{{a}_{jk}}{{a}_{jj}}\right| < 1 \n\]\n\n(4.1)\n\nor\n\n\[ \n{q}_{1} \mathrel{\text{:=}...
Proof. The Jacobi matrix \( - {D}^{-1}\left( {{A}_{L} + {A}_{R}}\right) \) has diagonal entries zero and off-diagonal entries \( - {a}_{jk}/{a}_{jj} \) . Hence by Theorem 3.26 we have\n\n\[ \n{\begin{Vmatrix}-{D}^{-1}\left( {A}_{L} + {A}_{R}\right) \end{Vmatrix}}_{\infty } = {q}_{\infty } \n\]\n\n\[ \n{\begin{Vmatrix}-...
Yes
Theorem 4.3 Assume that the matrix \( A = \left( {a}_{jk}\right) \) fulfills the Sassenfeld criterion\n\n\[ p \mathrel{\text{:=}} \mathop{\max }\limits_{{j = 1,\ldots, n}}{p}_{j} < 1 \]\n\nwhere the numbers \( {p}_{j} \) are recursively defined by\n\n\[ {p}_{1} \mathrel{\text{:=}} \mathop{\sum }\limits_{{k = 2}}^{n}\le...
Proof. Consider the equation\n\n\[ \left( {D + {A}_{L}}\right) x = - {A}_{R}z \]\n\nfor \( z \in {\mathbb{C}}^{n} \) with \( \parallel z{\parallel }_{\infty } = 1 \), that is,\n\n\[ {x}_{j} = - \mathop{\sum }\limits_{{k = 1}}^{{j - 1}}\frac{{a}_{jk}}{{a}_{jj}}{x}_{k} - \mathop{\sum }\limits_{{k = j + 1}}^{n}\frac{{a}_{...
Yes
Corollary 4.4 Assume that the matrix \( A \) is strictly row-diagonally dominant. Then the Gauss-Seidel iterations converge.
Null
No
Example 4.5 The tridiagonal matrix\n\n\[ A = \left( \begin{array}{rrrrrr} 2 & - 1 & & & & \\ - 1 & 2 & - 1 & & & \\ & - 1 & 2 & - 1 & & \\ & \cdot & \cdot & \cdot & \cdot & \cdot \\ & & & - 1 & 2 & - 1 \\ & & & & - 1 & 2 \end{array}\right) \] from Example 2.1 is not strictly row-diagonally dominant, but it satisfies th...
Proof. Obviously, \( {q}_{\infty } = 1 \) ; i.e.,(4.1) is not fulfilled. We have the recursion\n\n\[ {p}_{1} = \frac{1}{2},\;{p}_{j} = \frac{1}{2}{p}_{j - 1} + \frac{1}{2},\;j = 2,\ldots, n - 1,\;{p}_{n} = \frac{1}{2}{p}_{n - 1}. \]\n\nFrom this, by induction, it follows that\n\n\[ {p}_{j} = 1 - \frac{1}{{2}^{j}},\;j =...
Yes
Assume that the Jacobi matrix \( B \mathrel{\text{:=}} - {D}^{-1}\left( {{A}_{L} + {A}_{R}}\right) \) has real eigenvalues and spectral radius less than one. Then the spectral radius of the iteration matrix\n\n\[ I - \omega {D}^{-1}A = \left( {1 - \omega }\right) I - \omega {D}^{-1}\left( {{A}_{L} + {A}_{R}}\right) \]\...
Proof. For \( \omega > 0 \) the equation \( {Bu} = {\lambda u} \) is equivalent to\n\n\[ \left\lbrack {\left( {1 - \omega }\right) I + {\omega B}}\right\rbrack u = \left\lbrack {1 - \omega + {\omega \lambda }}\right\rbrack u. \]\n\nHence the eigenvalues \( \lambda \) of \( B \) correspond to the eigenvalues \( 1 - \ome...
Yes
Theorem 4.11 (Kahan) A necessary condition for the SOR method to be convergent is that \( 0 < \omega < 2 \) .
Proof. Since the eigenvalues \( {\mu }_{1},\ldots ,{\mu }_{n} \) of \( B\left( \omega \right) \) are the zeros of the characteristic polynomial, they satisfy\n\n\[ \mathop{\prod }\limits_{{j = 1}}^{n}{\mu }_{j} = \det B\left( \omega \right) \]\n\n(where multiple eigenvalues are repeated according to their algebraic mul...
Yes
Theorem 4.12 (Ostrowski) If \( A \) is Hermitian and positive definite, then the SOR method converges for all \( {x}_{0} \in {\mathbb{C}}^{n} \), all \( y \in {\mathbb{C}}^{n} \), and all \( 0 < \omega < 2 \) to the unique solution of \( {Ax} = y \) .
Proof. Let \( \mu \) be an eigenvalue of \( B\left( \omega \right) \) with eigenvector \( x \) ; i.e., \[ \left\lbrack {\left( {1 - \omega }\right) D - \omega {A}_{R}}\right\rbrack x = \mu \left( {D + \omega {A}_{L}}\right) x. \] With the aid of \[ \left( {2 - \omega }\right) D - {\omega A} - \omega \left( {{A}_{R} - {...
Yes
Corollary 4.16 Under the assumptions of Theorem 4.15 the Gauss-Seidel method converges twice as fast as the Jacobi method.
Proof. From (4.8) we observe that \( \mu = {\lambda }^{2} \) for \( \omega = 1 \) ; i.e., we have\n\n\[ \rho \left\lbrack {B\left( 1\right) }\right\rbrack = {\left\{ \rho \left\lbrack -{D}^{-1}\left( {A}_{L} + {A}_{R}\right) \right\rbrack \right\} }^{2} \]\n\nfor the spectral radii of the Gauss-Seidel matrix \( B\left(...
Yes
For the tridiagonal matrix \( A \) from Example 4.5 we have\n\n\[ \frac{N\left( \mathrm{{SOR}}\right) }{N\left( \mathrm{{Jacobi}}\right) } \approx \frac{\pi }{4\left( {n + 1}\right) } \] \n\nfor the optimal relaxation parameter.
Proof. Using the trigonometric addition theorem\n\n\[ \frac{1}{2}\sin \frac{{\pi j}\left( {k - 1}\right) }{n + 1} + \frac{1}{2}\sin \frac{{\pi j}\left( {k + 1}\right) }{n + 1} = \cos \frac{\pi j}{n + 1}\sin \frac{\pi jk}{n + 1}, \] \n\nit can be seen that the Jacobi matrix\n\n\[ - {D}^{-1}\left( {{A}_{L} + {A}_{R}}\rig...
Yes
Theorem 4.18 For the spectral radius of \( T \) we have that \( \rho \left( T\right) = {0.5} \) ; i.e., the two-grid iterations converge.
Proof. We note that from (4.18) and (4.19), with \( h \) replaced by \( {2h} \), we have\n\nthat\n\[ \n{A}^{\left( 2h\right) }{v}_{j}^{\left( 2h\right) } = \frac{1}{{h}^{2}}{\sin }^{2}\left( {\pi jh}\right) {v}_{j}^{\left( 2h\right) } = \frac{4}{{h}^{2}}{c}_{j}^{2}{s}_{j}^{2}{v}_{j}^{\left( 2h\right) }, \n\]\n\nwhence\...
Yes
We consider the best approximation of a given continuous function \( f : \left\lbrack {0,1}\right\rbrack \rightarrow \mathbb{R} \) by a polynomial \[ p\left( x\right) = \mathop{\sum }\limits_{{k = 0}}^{n}{\alpha }_{k}{x}^{k} \] of degree \( n \) in the least squares sense, i.e., with respect to the \( {L}_{2} \) norm.
Using the monomials \( x \mapsto {x}^{k}, k = 0,1,\ldots, n \), as a basis of the subspace \( {P}_{n} \subset C\left\lbrack {0,1}\right\rbrack \) of polynomials of degree less than or equal to \( n \) (see Theorem 8.2), from Corollary 3.53 and the integrals \[ {\int }_{0}^{1}{x}^{j}{x}^{k}{dx} = \frac{1}{j + k + 1} \] ...
Yes
Theorem 5.3 Let \( X \) and \( Y \) be Banach spaces, let \( A : X \rightarrow Y \) be a bounded linear operator with a bounded inverse \( {A}^{-1} : Y \rightarrow X \) and let \( {A}^{\delta } : X \rightarrow Y \) be a bounded linear operator such that \( \begin{Vmatrix}{A}^{-1}\end{Vmatrix}\begin{Vmatrix}{{A}^{\delta...
Proof. Writing \( {A}^{\delta } = A\left\lbrack {I + {A}^{-1}\left( {{A}^{\delta } - A}\right) }\right\rbrack \), by Theorem 3.48 we observe that the inverse operator \( {\left\lbrack {A}^{\delta }\right\rbrack }^{-1} = {\left\lbrack I + {A}^{-1}\left( {A}^{\delta } - A\right) \right\rbrack }^{-1}{A}^{-1} \) exists and...
Yes
Theorem 5.4 Let \( A \) be an \( m \times n \) matrix of rank \( r \) . Then there exist nonnegative numbers\n\n\[ \n{\mu }_{1} \geq {\mu }_{2} \geq \cdots \geq {\mu }_{r} > {\mu }_{r + 1} = \cdots = {\mu }_{n} = 0 \n\]\n\nand orthonormal vectors \( {u}_{1},\ldots ,{u}_{n} \in {\mathbb{C}}^{n} \) and \( {v}_{1},\ldots ...
Proof. The Hermitian and semipositive definite matrix \( {A}^{ * }A \) of rank \( r \) has \( n \) orthonormal eigenvectors \( {u}_{1},\ldots ,{u}_{n} \) with nonnegative eigenvalues\n\n\[ \n{A}^{ * }A{u}_{j} = {\mu }_{j}^{2}{u}_{j},\;j = 1,\ldots, n \n\]\n\n(5.7)\n\nwhich we may assume to be ordered according to \( {\...
Yes
Theorem 5.5 Let \( A \) be an \( m \times n \) matrix of rank \( r \) with singular system \( \left( {{\mu }_{j},{u}_{j},{v}_{j}}\right) \). The linear system\n\n\[ \n{Ax} = y \n\]\n\n(5.9)\n\nis solvable if and only if\n\n\[ \n\left( {y, z}\right) = 0 \n\]\n\n\( \left( {5.10}\right) \)\n\nfor all \( z \in {\mathbb{C}}...
Proof. Let \( x \) be a solution of (5.9) and let \( {A}^{ * }z = 0 \). Then\n\n\[ \n\left( {y, z}\right) = \left( {{Ax}, z}\right) = \left( {x,{A}^{ * }z}\right) = 0. \n\]\n\nThis implies the necessity of condition (5.10) for the solvability of (5.9).\n\nConversely, assume that (5.10) is satisfied. In terms of the ort...
Yes
Theorem 5.7 Let \( A \) be an \( m \times n \) matrix of rank \( r \) with singular system \( \left( {{\mu }_{j},{u}_{j},{v}_{j}}\right) \) and let \( \alpha > 0 \) . Then for each \( y \in {\mathbb{C}}^{m} \) the linear system\n\n\[ \n\alpha {x}_{\alpha } + {A}^{ * }A{x}_{\alpha } = {A}^{ * }y \n\]\n\n(5.18)\n\nis uni...
Proof. For \( \alpha > 0 \) the matrix \( {\alpha I} + {A}^{ * }A \) is positive definite and therefore nonsingular. Since\n\n\[ \n\alpha {u}_{j} + {A}^{ * }A{u}_{j} = \left( {\alpha + {\mu }_{j}^{2}}\right) {u}_{j} \n\]\n\na singular system for the matrix \( {\alpha I} + {A}^{ * }A \) is given by \( \left( {\alpha + {...
Yes
Corollary 5.8 Under the assumptions of Theorem 5.7 we have convergence:\n\n\[ \mathop{\lim }\limits_{{\alpha \rightarrow 0}}{\left( \alpha I + {A}^{ * }A\right) }^{-1}{A}^{ * }y = {A}^{ \dagger }y. \]
Proof. This is obvious from (5.13) and (5.19).
Yes