chunk_uid
stringlengths
40
40
chunk_type
stringclasses
2 values
chunk_index
int64
0
6.71k
total_chunks
int64
1
6.71k
section_title
stringlengths
1
157
embed_text
stringlengths
1
83.3k
spans
dict
paper_doi
stringlengths
0
63
paper_id_arxiv
stringlengths
9
16
title
stringlengths
7
245
authors
listlengths
1
768
categories
listlengths
1
7
year
int64
2k
2.02k
language
stringclasses
2 values
discipline
stringclasses
8 values
dense_vector
listlengths
1.02k
1.02k
64c3db49db426357cb6a943951a5a9beaa72c806
subsection
41
74
A Class of Examples
In particular, \Gamma _{{U}} is a principal groupoid with unit space \Gamma _{{U}}^{(0)} = \coprod U_{i}, and is equivalent to the space X (see ).Let \Sigma _{c} be the groupoid extension equal as a topological space to G\times \Gamma _{{U}} endowed with the operations\bigl (g,(i,x,j)\bigr )\bigl (h,(j,x,k)\bigr ) = \...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.049666453152894974, 0.013981091789901257, -0.05223067104816437, -0.004132511559873819, -0.0052123828791081905, -0.05461173132061958, -0.011859506368637085, 0.015370042994618416, 0.020986899733543396, 0.02564217709004879, -0.011882402002811432, -0.032815881073474884, -0.006235017441213131,...
52e36cd9e74a2f837903b14e902b4f9765f8e562
subsection
42
74
A Class of Examples
We equip \Sigma _{c} with the Haar system \lambda =\lbrace \lambda ^{(i,x)}\rbrace given by\lambda ^{(i,x)}(f)=\sum _{j}\int _{G} f\bigl (g,(i,x,j)\bigr )\,d\mu (g).Then using Lemma REF ,f*f^{\prime }\bigl (g,&(i,x,j)\bigr )\\ &=\sum _{k}\int _{G} f\bigl (h,(i,x,k)\bigr ) f^{\prime }\bigl ( g-h-c_{iki}(x)+c_{kij}(x),(k...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.06774049252271652, 0.021130762994289398, -0.04635036736726761, -0.0129988519474864, 0.013036994263529778, -0.06114953011274338, -0.00905495136976242, 0.03539593517780304, 0.017652196809649467, 0.004893640521913767, -0.016843583434820175, -0.00021836392988916487, -0.030300140380859375, 0...
2fff3a549be4c793a269cf40b750c23ad523fa78
subsection
43
74
A Class of Examples
Then \operatorname{Ind}((i,x),\tau ) is equivalent to the representation L on \ell ^{2}(I(x)) where L(f) is given by multiplication by the matrix A=(a_{jk}) witha_{jk}=\tau {(c_{ijk}(x))} \int _{G}f(g,(j,x,k))\tau (g)\,d\mu (g).We have (\Sigma _{c})_{(i,x)}=\lbrace \,(g,(j,x,i):j\in I(x)\,\rbrace . Then \operatorname{I...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.012279386632144451, 0.03971530869603157, -0.03858581557869911, -0.014080464839935303, -0.0028332644142210484, -0.014362838119268417, -0.0249708853662014, 0.039928995072841644, 0.0019880549516528845, 0.022940855473279953, -0.007604977581650019, -0.022360848262906075, 0.0012620905181393027,...
a4e6a0d9c16eb4baa0f61d0a445027bb6d3efd2a
subsection
44
74
A Class of Examples
ButU(&f_{1}*f_{2})(j) \\ &= \int _{G} f_{1}*f_{2}(g-c_{iji}(x), (j,x,i))\tau (g)\,d\mu (g) \\ &= \sum _{k} \int _{G}\int _{G} f_{1}(h,(j,x,k)) f_{2}(-h+g-c_{iji}(x) -c_{jki}(x),(k,x,i)) \tau (g)\,d\mu (h)\,d\mu (g) \\ {which, since c_{iji}(x)+c_{jki}(x) = c_{ijk}(x)+c_{iki}(x), is} &=\sum _{k} \int _{G}\int _{G} f_{1}(...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.0402727909386158, 0.010464823804795742, -0.056625984609127045, -0.01073178369551897, -0.012440325692296028, -0.010701273567974567, -0.003005204489454627, 0.011280957609415054, -0.0032683503814041615, 0.03832016885280609, -0.008519831113517284, -0.01410310436040163, -0.009442748501896858, ...
6d7f458bae20ad2776f32fa33120f7db001841fe
subsection
45
74
A Class of Examples
Given a Čech 2-cocycle c \in Z^2(U, {G}), where {U}= \lbrace U_i\rbrace _{i \in I} is a locally finite open cover of X, the cover \lbrace \widehat{G}\times U_i\rbrace _{i \in I} of \widehat{G}\times X is locally finite and supports a normalized 2-cocycle \nu ^{c} such that\nu ^{c}_{ijk}(\tau ,x)=\overline{\tau \bigl (c...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.06910911947488785, 0.008180761709809303, -0.04551311582326889, -0.006997908931225538, 0.008104448206722736, -0.02087925747036934, -0.004338398110121489, 0.010065694339573383, 0.012912171892821789, 0.015598392114043236, 0.019063005223870277, 0.008852316066622734, -0.005425859242677689, 0...
a0391f7158e1f97e676dbe4e746dea5f64b1917b
subsection
46
74
The Raeburn–Taylor Algebra
For our computation of the Dixmier-Douady class of C^{*}(\Sigma _{c}), we want a slight modification of the Raeburn–Taylor Algebra based on their original construction in and reproduced in *Proposition 5.40. Specifically, given a normalized 2-cocycle \nu =\lbrace \nu _{ijk}\rbrace \in Z^{2}({U},{S}) defined on a locall...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.056940242648124695, 0.022626575082540512, -0.02874475158751011, 0.02110084518790245, 0.02284017764031887, -0.05086783692240715, -0.008658517152071, 0.038845084607601166, 0.02558649145066738, 0.044764917343854904, 0.020536325871944427, 0.011694720014929771, -0.013281479477882385, -0.0103...
052e33dcba32d48a1ce79ebb975486e869e865e9
subsection
47
74
The Raeburn–Taylor Algebra
We define a multiplication on A_{1}(\nu ) by twisting the usual matrix multiplication with \nu :(f_{ij} )(g_{lk}) = (h_{ij})whereh_{ik}(x)=\sum _{j} \overline{\nu _{ijk}(x)} f_{ij}(x)g_{jk}(x).To see that the sum in (REF ) is meaningful, observe that it is always a finite sum:h_{ik}(x) ={\left\lbrace \begin{array}{ll} ...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.05010664463043213, -0.0004322251188568771, -0.00611075246706605, -0.0017155498499050736, 0.03091232106089592, -0.029111899435520172, -0.011206860654056072, 0.008712208829820156, 0.024290431290864944, 0.02792179025709629, -0.024305690079927444, 0.024366719648241997, -0.0226578451693058, ...
5c7260bdec0374f11590ec8348ddab8fd61f51c6
subsection
48
74
The Raeburn–Taylor Algebra
To see that it is anti-multiplicative, we require Lemma REF (e):(f*g)^{*}_{ij}(x) &= \nu _{iji}(x) \overline{(f*g)_{ji}(x)}\\ &= \nu _{iji}(x) \sum _{k}\nu _{jki}(x) \overline{f_{jk}(x)} \overline{g_{ki}(x)} \\ {which, using Lemma~\ref {lem-norm-coc}(e), is} &= \sum _{k}\overline{\nu _{ikj}(x)} \nu _{iki}(x) \overline{...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.05391572415828705, 0.023113276809453964, -0.040276601910591125, 0.039849426597356796, 0.01315854862332344, -0.04952191188931465, 0.0016514932503923774, 0.049888063222169876, 0.007807405665516853, 0.05971311405301094, -0.02215213142335415, 0.004840057343244553, -0.027415549382567406, -0....
677773b40463170016a6bc22077b3b1f51347a14
subsection
49
74
The Raeburn–Taylor Algebra
The groupoid is the blow-up \Gamma _{{U}} associated to the cover {U} of X corresponding to \nu and the cocycle \varphi _{\nu } in Z^{2}(\Gamma _{{U}},\mathbf {T}) is given by\varphi _{\nu }\bigl ((i,x,j),(j,x,k)\bigr )=\overline{\nu _{ijk}(x)}.(The complex conjugate in (REF ) is missing from the formula in .) Note th...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.06672380864620209, 0.02582266367971897, -0.05384300276637077, -0.03354504331946373, -0.012163512408733368, -0.023457111790776253, 0.009736915118992329, 0.046639516949653625, 0.01825289987027645, 0.013651520945131779, -0.027165686711668968, -0.017016708850860596, -0.0029035236220806837, ...
85be91b85a184eaf1b65d5378a53bc3b1bb70fc7
subsection
50
74
The Dixmier–Douady Class of
Given a locally compact space X, a locally compact abelian group G and a 2-cocycle c \in H^2(X, {G}), let \nu ^{c} be as in (REF ). We can form the associated Raeburn–Taylor twisted groupoid (\Gamma _{,\varphi _{\nu ^{c}}). We want to verify that C^{*}(\Gamma _{,\varphi _{\nu ^{c}}) is a continuous-trace C^{*}-algebra ...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.029535064473748207, 0.017513561993837357, -0.046590954065322876, 0.011159555986523628, 0.010625605471432209, -0.05870399251580238, 0.004115229472517967, 0.024714259430766106, 0.027368754148483276, 0.03389820083975792, 0.003106445074081421, -0.002894771983847022, 0.0022940782364457846, 0...
60b844e6bd2d4048d2307fc162bc3bf4e2d7ef16
subsection
51
74
The Dixmier–Douady Class of
\end{}Since each summand is in C_{0}(\widehat{G}\times X), it will suffice to see that h_{ik} is continuous on \widehat{G}\times X. Suppose that (\tau _{n},x_{n})\rightarrow (\tau _{0},x_{0}). It is enough to show that h_{ik}(\tau _{n},x_{n})\rightarrow h_{ik}(\tau _{0},x_{0}). For this, it suffices to consider each su...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.04593610391020775, 0.01098040770739317, -0.03366612643003464, -0.005039999261498451, 0.008614927530288696, -0.022952791303396225, -0.017351943999528885, 0.031193818897008896, 0.021228279918432236, 0.0326894149184227, -0.025257226079702377, -0.04343327134847641, -0.012338651344180107, 0....
06ef29c12d62dc65c3b40f219707b9ca4d9bc106
subsection
52
74
The Dixmier–Douady Class of
Just as in the proof of \cite {rw:morita}*{Proposition~5.40}, A(c) is complete with respect to the norm \begin{align}\Vert f\Vert =\sup _{(\tau ,x)}\Vert f\Vert _{(\tau ,x)}, \end{align} and has spectrum identified (topologically) with GX. Moreover, we have the following analogue of the Raeburn--Taylor result. \begin{}...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.04716998711228371, 0.03151784837245941, -0.03273829072713852, -0.009092274121940136, 0.01284512598067522, -0.06578169018030167, -0.0030263084918260574, 0.0015388979809358716, 0.043325603008270264, 0.05000750720500946, 0.02590382844209671, 0.0008214130066335201, -0.013302790932357311, 0....
e35004bd4ab76373aef344c01fb1ed7a6299a95c
subsection
53
74
The Dixmier–Douady Class of
Then as above, we get p(n,i)\in A(\nu ^{c}) such that\pi _{(i,(\tau ,x))}(p(n,i))is a rank-one projection for all (\tau ,x)\in F_{n,i}. Similarly, let \phi _{(n,i)(m,j)} \in C_{c}^{+}(W_{(n,i)(m,j)}) be identically one on F_{(n,i)(m,j)}.
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.024093693122267723, 0.024139469489455223, -0.021484432741999626, 0.007045764941722155, 0.0037632088642567396, -0.01797490194439888, -0.014877359382808208, 0.031829919666051865, 0.020843561738729477, 0.041473500430583954, 0.010513333603739738, -0.030654989182949066, -0.011489898897707462, ...
f02055d7f36cf7b1a5275dd6ffd829df6b42d7a7
subsection
54
74
The Dixmier–Douady Class of
Then we get v((n,i),(m,j))\in A(\nu ^{c}) withv((n,i),(m,j))_{rs}(\tau ,x)= {\left\lbrace \begin{array}{ll} \phi _{(n,i)(m,j)}(\tau ,x)&\text{if $r=i$ and $s=j$, and} \\ 0 &\text{otherwise.} \end{array}\right.}Then check that\pi _{(i,(\tau ,x))} \bigl (v((n,i),(m,j))v((n,i),(m,j))^{*}\bigr ) = \pi _{(i,(\tau ,x))} \big...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.03600570186972618, 0.04238298535346985, -0.017789257690310478, -0.030147148296236992, -0.007727495394647121, 0.0363718643784523, -0.007101972587406635, 0.029140209779143333, 0.03475465625524521, 0.03615827113389969, 0.010969534516334534, -0.0030360741075128317, -0.015172742307186127, 0....
5196a6cce2559b9646521a99cdb631d80acb2fb0
subsection
55
74
The Dixmier–Douady Class of
With these modifications in place, we can prove Theorem REF . [Proof of Theorem REF ] By Lemma , it suffices to produce an isomorphism \Phi :C^{*}(\Sigma _{c})\rightarrow A(\nu ^{c}) that intertwines each \operatorname{Ind}((i,x),\tau ) with \pi _{(i, (\tau ,x))}. We use the Fourier Transform. If f\in C_{c}(\Sigma _{c}...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.018356362357735634, 0.01052095741033554, -0.019012492150068283, -0.01206209883093834, -0.011703516356647015, -0.016006503254175186, -0.008453386835753918, 0.014862092211842537, 0.028961243107914925, 0.029205385595560074, -0.0003826626343652606, -0.030212467536330223, 0.016479527577757835,...
10255b6b8aeca5bba30ab96eaf4439e1aed61750
subsection
56
74
The Dixmier–Douady Class of
Since finite sums of such functions are dense in the inductive limit topology, we deduce that \Phi (f)\in A(\nu ^{c}) for all f.Note\Phi (f^{*}(i,(\tau ,x),j)) &= \int _{G}\tau (g) \overline{f(-g-c_{iji}(x),(j,x,i))} \,d\mu (g) \\ &= \overline{\int _{G} \tau (g) f(g-c_{iji}(x),(j,x,i))\,d\mu (g)} \\ &= \overline{\tau (...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.04313027113676071, 0.008561953902244568, 0.014208875596523285, -0.027318891137838364, -0.015010127797722816, -0.04816671460866928, 0.012255346402525902, 0.02232823334634304, 0.007070098537951708, 0.03519405797123909, -0.0514632947742939, 0.014498853124678135, 0.015529034659266472, 0.016...
59c7a125b6de02412ec29ec939ecd73bc553d131
subsection
57
74
The Dixmier–Douady Class of
It follows from Lemma REF , that \operatorname{Ind}((i,x),\tau )(f) is equivalent to multiplication by the matrix\bigl [ \tau (c_{ijk})(x)\Phi (f) \bigr ].Since \tau (c_{ijk}(x))=\overline{\nu ^{c}_{ijk}(\tau ,x)}, we see that\operatorname{Ind}((i,x),\tau )(f)=\pi _{(i,(\tau ,x))}(\Phi (f)).This shows immediately that ...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.04531041905283928, 0.017895327880978584, -0.024592725560069084, -0.00350507372058928, 0.011243697255849838, -0.019954891875386238, -0.002576362807303667, 0.01194547489285469, 0.04344918578863144, 0.03524143621325493, -0.011693749576807022, -0.024592725560069084, -0.003487910609692335, 0...
c22509bbb8cfa7db8b1b008a900bc466dc74ed07
subsection
58
74
Groupoids Associated to Local Homeomorphisms
In this section we extend our results from Section  to a more general setting. Let \psi :Y\rightarrow X be a local homeomorphism and form the principal groupoidR(\psi )=\lbrace \,(x,y)\in Y\times Y:\psi (x)=\psi (y)\,\rbrace .Let G be a locally compact abelian group. Given a unit-space-preserving groupoid extension\beg...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.0517265684902668, 0.03744454309344292, 0.00017904983542393893, 0.01306896936148405, 0.033538345247507095, -0.05694499984383583, -0.0023784302175045013, -0.024749407544732094, 0.030501890927553177, 0.02758750319480896, -0.009124627336859703, 0.02758750319480896, -0.019714077934622765, 0....
481d3c9df6db51408c5df539f85393412c4dde4c
subsection
59
74
Groupoids Associated to Local Homeomorphisms
Equivalently, all locally trivial principle G-bundles over the double-overlaps W_{ij} are trivial. A special case where these assumptions automatically hold is when G is a Lie group and X admits good covers in the sense that every open over of X admits a refinement in which all nontrivial overlaps are contractible. Thi...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.06970366835594177, 0.03436354175209999, -0.012062335386872292, -0.002685605315491557, 0.016968142241239548, -0.06378312408924103, 0.05008043348789215, 0.0035649121273308992, 0.013588246889412403, 0.008651921525597572, -0.02430777996778488, -0.014618237502872944, 0.023804228752851486, 0....
8c78ddc1f73a6a1c4942934f5697db872c273a45
subsection
60
74
Groupoids Associated to Local Homeomorphisms
12 or *Equation 1.3, the C_{c}(\Sigma ({U}))-valued inner product on C_{c}(\Sigma _{U}) is given by\langle f_{1}\mathrel {,}f_{2} \rangle _{{\hspace{-1.66656pt}}\copy \scriptscriptstyle {\Sigma (U)}}(\gamma )=\int _{\Sigma } \overline{f_{1}(\sigma ^{-1})} f_{2}(\sigma ^{-1}\gamma ) \,d\lambda ^{r(\gamma )}(\sigma ).Th...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.028973683714866638, 0.02140604518353939, -0.027081774547696114, -0.024488026276230812, 0.024060821160674095, -0.03997422382235527, -0.018659723922610283, 0.015715057030320168, 0.00838390737771988, 0.03799077123403549, 0.004462771117687225, 0.006366123910993338, 0.010230044834315777, 0.0...
2f654ff0b354c45085817e1c86fa382f93236958
subsection
61
74
Groupoids Associated to Local Homeomorphisms
Since the representations are irreducible, W must be a unitary and the representations must be equivalent.We also will need to examine the case of blowing up the unit space \Sigma ^{(0)} with respect to a locally finite cover {U}=\lbrace U_{i}\rbrace . This gives us the equivalent groupoid\Sigma ^{\prime }=\lbrace \,(i...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.03961551561951637, 0.05130483955144882, -0.003998176194727421, -0.008354967460036278, -0.007004438433796167, -0.030642511323094368, 0.026857977733016014, 0.024263741448521614, 0.034335482865571976, 0.04776447266340256, -0.0056806146167218685, -0.00157370965462178, 0.00144399784039706, 0...
9d592e9e3e27c73990ccc15401500171dd85b29f
subsection
62
74
Groupoids Associated to Local Homeomorphisms
This is verified in the next lemma which is analogous to Lemma REF .Lemma 6.2 With {U} and \Sigma ^{\prime } as above, we have \mathop {Z\mathord {\mathop {\text{--}}}}\!\operatorname{Ind}\nolimits (\operatorname{Ind}^{\Sigma ^{\prime }}((i,y),\tau ) equivalent to \operatorname{Ind}^{\Sigma }(y,\tau ) for all y\in U_{i...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.046604663133621216, 0.027773572131991386, -0.0019199245143681765, -0.03845571354031563, 0.01909051649272442, -0.02900964766740799, -0.02060127630829811, -0.001472609699703753, 0.031436022371053696, 0.04248441010713577, -0.024446846917271614, 0.0019208781886845827, -0.028307678177952766, ...
03d78b314024790fba61cf48a3776982c063a260
subsection
63
74
Groupoids Associated to Local Homeomorphisms
Suppose that \psi :Y\rightarrow X is a local homeomorphism, and let \Sigma be a groupoid extension as in (REF ). Then C^{*}(\Sigma ) has continuous trace with spectrum identified with \widehat{G}\times X via (\tau ,x)\mapsto \operatorname{Ind}((i,y),\tau ) for any y\in \Phi ^{-1}(x). Furthermore, there is a locally fin...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.06445252150297165, 0.01789839193224907, -0.019271669909358025, -0.00919333379715681, 0.016738733276724815, -0.05715888738632202, 0.016708217561244965, 0.019363220781087875, 0.03652919828891754, 0.02801487222313881, 0.0062026395462453365, 0.034881263971328735, 0.01142109651118517, 0.0174...
8c287d7dbb1aaa67e0ca9fdc7edb07ae01be74f5
subsection
64
74
Groupoids Associated to Local Homeomorphisms
If we letR^{\prime } = \lbrace \,(i,(x,y),j):\text{$\psi (x)=\psi (y)$, $x\in V_{i}$ and $y\in V_{j}$}\,\rbrace ,then we obtain a generalised twist\begin{}[column sep=3cm] G \times \coprod V_{j} [r,"\iota ^{\prime }", hook] [dr,shift left, bend right = 15] [dr,shift right, bend right = 15]&\Sigma ^{\prime } [r,"\pi ^{\...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.036063097417354584, 0.022074155509471893, -0.008382687345147133, 0.018489202484488487, 0.0074406834319233894, -0.01514070387929678, -0.02031981572508812, 0.012943967245519161, 0.01217358373105526, 0.021204613149166107, -0.04616198688745499, 0.003462911583483219, -0.030739061534404755, 0...
b9eb61407446ca5dcb22f0eddc77f5f8d1cb044f
subsection
65
74
Groupoids Associated to Local Homeomorphisms
There is a groupoid homomorphism \tau :R^{\prime }\rightarrow \Gamma _{W} such that\tau (i,(x,y),j)=(i,\psi (x),j)and\tau ^{-1}(i,w,j)=(i,(x,y),j)\quad \text{if $x\in V_{i}$ and $y\in V_{j}$ satisfy $\psi (x) = w = \psi (y)$.}So, defining \tilde{\varphi } := \varphi \circ (\tau ^{-1} \times \tau ^{-1}) \in Z^{2}(\Gamma...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.04272150248289108, 0.01118387933820486, -0.011870474554598331, -0.009269040077924728, -0.0216506477445364, -0.021467555314302444, 0.02462589554488659, 0.026105890050530434, 0.025388779118657112, 0.04800065979361534, 0.018187154084444046, -0.018370246514678, 0.00073713663732633, -0.00532...
60b2371920a72ae976ba681d8b676171aaaf9413
subsection
66
74
Extensions and Cocycles
Let G be a locally compact Hausdorff abelian group and let \Gamma be a locally compact Hausdorff groupoid with Haar system \lbrace \lambda ^{u}\rbrace _{u\in \Gamma ^{(0)}}. Following , we define an extension (or twist) by G over \Gamma to be a central groupoid extension \Gamma ^{(0)}\times G\xrightarrow{}\Sigma \xri...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.044590141624212265, 0.013192394748330116, -0.033083993941545486, 0.006012497004121542, 0.018754716962575912, -0.06787712126970291, 0.026277968659996986, 0.010804181918501854, 0.005764519330114126, 0.027407219633460045, -0.022432412952184677, 0.013177135027945042, -0.030413467437028885, ...
5ec763da3f0a273ffac4587f486f114717661f7f
subsection
67
74
Extensions and Cocycles
Hence there is a subsequence \lbrace \sigma _{n_{k}}\rbrace _{k\ge 1} such that \sigma _{k_{n}}\rightarrow \sigma \in K and g_{k_{n}}\sigma _{k_{n}}\rightarrow \sigma ^{\prime }\in K. It follows that \pi (\sigma )=\pi (\sigma ^{\prime }). Hence there exists g\in G such that \sigma ^{\prime }=g\sigma . Therefore\iota (r...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.048951826989650726, 0.04644929990172386, -0.03756839036941528, 0.03137311711907387, 0.02903844229876995, -0.02685636281967163, 0.040986474603414536, 0.0509050153195858, -0.0012054844992235303, 0.04654085636138916, -0.03041177988052368, 0.01739559881389141, -0.04196306690573692, 0.011818...
0e81446dbf6892668c44fd585b6d10d1a55a6123
subsection
68
74
Extensions and Cocycles
The collection T_{\Gamma }(G) of proper isomorphism classes of twists by G forms an abelian group under * with neutral element [\Gamma \times G].Remark 7.2 Let \Sigma be an extension by G over a principal groupoid \Gamma ; that is, the map \gamma \mapsto (r(\gamma ),s(\gamma )) is injective; equivalently, I(\Gamma ) = ...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.02641375921666622, -0.02237005904316902, -0.028534794226288795, 0.010956140235066414, 0.02420116774737835, -0.054170332849025726, 0.019943837076425552, 0.037446197122335434, 0.002990812761709094, 0.026291685178875923, -0.00037695112405344844, 0.004848625976592302, 0.0038663120940327644, ...
f258d7e37e0e43d02dfee676cbaa796fcc65b3aa
subsection
69
74
Extensions and Cocycles
So \psi (\gamma _{0},\gamma _{1}) := \varphi (\gamma _{0}, \gamma _{1}) - (d^{1}f)(\gamma _{0}, \gamma _{1}) defines a continuous normalized 2-cocycle cohomologous to \varphi .Notation 7.4 Recall from *Lemma I.1.14 that given a normalized 2-cocycle \varphi on \Gamma , there is an extension \Sigma ({\Gamma }, {\varphi ...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.053944095969200134, -0.03368454799056053, -0.01486666314303875, -0.00041071159648709, 0.015896420925855637, -0.04771977663040161, 0.012158779427409172, 0.026316052302718163, 0.01249440386891365, 0.034782957285642624, -0.015156520530581474, 0.011106137186288834, 0.009862799197435379, 0.0...
f1cd131bb7bcb525da4887950094779905e28352
subsection
70
74
Extensions and Cocycles
By replacing \tau with the map \gamma \mapsto \tau (r(\gamma ))^{-1}\tau (\gamma ), we can assume without loss of generality that \tau (u)=u for all u\in \Sigma ^{(0)}. Then \tau (\gamma _{1})\tau (\gamma _{2})\tau (\gamma _{1} \gamma _{2})^{-1}\in \pi ^{-1}(\Gamma ^{(0)})=\iota (\Gamma ^{(0)}\times G) for all (\gamma ...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.03701533004641533, 0.0016650032484903932, -0.0322854220867157, 0.011763733811676502, 0.011000845581293106, -0.05099143832921982, 0.039395540952682495, 0.08147644251585007, 0.005927640479058027, 0.038327496498823166, 0.01583755575120449, -0.008071355521678925, -0.003339542541652918, -0.0...
3b26754bce7906f4930d0b4be369cc23ad7a8042
subsection
71
74
Extensions and Cocycles
The map \psi : \Sigma ({\Gamma }, {\varphi }) \rightarrow \Sigma defined by \psi (g,\gamma ) := g\cdot \tau (\gamma ) = \iota (r(\gamma ),g)\tau (\gamma ) is a homeomorphism and a groupoid morphism.bottu:diff82book author=Bott, Raoul, author=Tu, Loring W., title=Differential forms in algebraic topology, series=Graduat...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.024094007909297943, 0.022384995594620705, -0.03723203018307686, -0.00988785084336996, 0.022064555436372757, -0.051453445106744766, 0.02317846566438675, 0.0075455899350345135, 0.02481118217110634, 0.06408791989088058, -0.000850214040838182, 0.027069518342614174, -0.004619671031832695, -0...
cda0c43206c323e2d770a74be98b62a576e0e45c
subsection
72
74
Extensions and Cocycles
Soc., volume=137, number=4, pages=13231332, review=MR2465655,kum:lnim85inproceedings author=Kumjian, Alexander, title=Diagonals in algebras of continuous trace, date=1985, booktitle=Operator algebras and their connections with topology and ergodic theory, series=Lecture Notes in Mathematics, volume=1132, publisher=Spri...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.04367689788341522, -0.007527473848313093, -0.04761422798037529, 0.008156988769769669, -0.0012542610056698322, 0.016298716887831688, 0.004078494384884834, -0.024997469037771225, 0.017366984859108925, 0.047705791890621185, -0.02052600495517254, -0.007966226898133755, 0.035497017204761505, ...
8c5a8e34024e1c9d5150500ecf31ef0fc2d844ad
subsection
73
74
Extensions and Cocycles
(3), pages=109134,pal:aom61article author=Palais, Richard S., title=On the existence of slices for actions of non-compact Lie groups, date=1961, journal=Ann. of Math., volume=73, pages=295323,ren:groupoidbook author=Renault, Jean, title=A groupoid approach to C^{*}-algebras, series=Lecture Notes in Mathematics, publish...
{ "cite_spans": [] }
1801.00832
The Dixmier-Douady Classes of Certain Groupoid $C^*$-Algebras with Continuous Trace
[ "Marius Ionescu", "Alex Kumjian", "Aidan Sims", "Dana P. Williams" ]
[ "math.OA" ]
2,018
en
Mathematics
[ -0.03820602223277092, -0.005130500998347998, -0.054745689034461975, 0.0326521098613739, 0.008704686537384987, 0.013114247471094131, 0.01670750603079796, -0.025816529989242554, -0.015242737717926502, 0.07757165282964706, -0.018050208687782288, -0.003921304829418659, -0.009009846486151218, -...
3fd04cfd0fc73048dd6267b513fe9acee04172b9
abstract
0
33
Abstract
We derive an algorithm to compute satisfiability bounds for arbitrary {\omega}-regular properties in an Interval-valued Markov Chain (IMC) interpreted in the adversarial sense. IMCs generalize regular Markov Chains by assigning a range of possible values to the transition probabilities between states. In particular, we...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ 0.0021093899849802256, -0.032468587160110474, -0.021711841225624084, 0.02154400572180748, -0.02375638484954834, 0.0008038950036279857, 0.025480516254901886, -0.03304838389158249, 0.018645023927092552, 0.03451313078403473, -0.0483672060072422, 0.010306639596819878, -0.015234907157719135, 0....
a47ea5d4ad40e9761284145bafbc31bdd298100f
subsection
1
33
INTRODUCTION
Markov Chains have been extensively used as an intuitive yet powerful mathematical tool for modeling systems evolving through time in a stochastic fashion. They allow us to answer critical questions about the behavior of the underlying systems, often specified in terms of symbolic temporal logics, and derive appropriat...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1109/tac.2014.2298143", "end": 343, "openalex_id": "https://openalex.org/W2059470663", "raw": "X. Ding, S. L. Smith, C. Belta, and D. Rus, “Optimal control of Markov decision processes with linear temporal logic constraints,” IEEE Transa...
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.0005186585476621985, -0.0049196286126971245, -0.016764873638749123, 0.009907903149724007, -0.001679919776506722, -0.034689102321863174, 0.030326269567012787, -0.02086837962269783, -0.007371815852820873, 0.026634640991687775, -0.05455067381262779, 0.008001070469617844, -0.00337700103409588...
4cb037c3ff621008eb91ee2f9027d4757bd9b28a
subsection
2
33
INTRODUCTION
Constructing the Cartesian product of a Markov Chain with a DRA enables to compute the probability that the stochastic evolution of the Markov Chain's state fulfills the property encoded in the DRA. In particular, it was shown that this probability is equal to that of reaching special sets of states called accepting Bo...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ 0.010886525735259056, -0.01916394755244255, -0.013999140821397305, 0.024107513949275017, -0.025618046522140503, -0.009627747349441051, 0.021330570802092552, -0.01953013800084591, 0.01621916890144348, 0.051236093044281006, -0.06457151472568512, 0.01949962042272091, -0.009627747349441051, 0....
75428939e91f7dde354409cb65137cff7a765d30
subsection
3
33
PRELIMINARIES
An Interval-Valued Markov Chain (IMC) is a 5-tuple \mathcal {I} = (Q, {, \widehat{T}, \Pi , L) where: \begin{} \item Q is a finite set of states, \item {: Q \times Q \rightarrow [0, 1] maps pairs of states to a lower transition bound so that {_{Q_{j} \rightarrow Q_{\ell }} := {(Q_{j}, Q_{\ell }) denotes the lower boun...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.014078516513109207, -0.010812605731189251, -0.021747302263975143, -0.007394082378596067, 0.022434057667851448, 0.005738235544413328, 0.015459660440683365, 0.037115395069122314, -0.02611202374100685, 0.04499020799994469, -0.04163273051381111, 0.0033593857660889626, -0.014742380939424038, ...
5d16cfe234b118f35041cb84589e507dbf9ebb93
subsection
4
33
PRELIMINARIES
\\ }A Markov Chain \mathcal {M} is said to be \textit {induced} by IMC \mathcal {I} if for all Q_j,Q_\ell \in Q, \begin{align} {(Q_j,Q_\ell ) \le T(Q_j, Q_\ell ) \le \widehat{T}(Q_j,Q_\ell ) \; \;. } \end{align}An IMC \mathcal {I}_{2} with transition functions {_{2} and \widehat{T}_{2} is said to be \textit {induced} b...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ 0.03509312868118286, 0.016554800793528557, -0.0365884006023407, -0.011458668857812881, -0.0091623580083251, -0.01014649122953415, 0.0016449902905151248, 0.02407693676650524, 0.003063019597902894, 0.028578011319041252, -0.020979586988687515, 0.0010547010460868478, -0.024305805563926697, 0.0...
a1c5f85ac5dedb01084ef2ae229a2cf1c07ab878
subsection
5
33
PRELIMINARIES
An element (E_{i}, F_{i}) \in Acc, with E_{i}, F_{i} \subset S, is called a \textit {Rabin Pair}. \end{} }The probability of satisfying \omega -regular property \phi starting from initial state Q_i in IMC \mathcal {I} under adversary \mathcal {\nu } is denoted by \mathcal {P}_{\mathcal {I}[\mathcal {\nu }]}(Q_i \models...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.009362775832414627, 0.022021980956196785, -0.03787841647863388, 0.015612256713211536, 0.008721803314983845, -0.031087161973118782, 0.030324099585413933, 0.002560074208304286, 0.00925594661384821, 0.020083801820874214, -0.05912207439541817, 0.01593274250626564, -0.03668804094195366, 0.01...
4526109e16d7316f93a973b4bfd9c063978e1384
subsection
6
33
PRELIMINARIES
Then, we construct the product \mathcal {I} \otimes \mathcal {A}, which is itself an IMC.\\ }\begin{} Let \mathcal {I} = (Q, {, \widehat{T}, \Pi , L) be an Interval-valued Markov Chain and \mathcal {A} = (S, 2^{\Pi }, \delta , s_0, Acc) be a Deterministic Rabin Automaton.
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ 0.03930147737264633, 0.0331987626850605, 0.007487267255783081, 0.03203924745321274, -0.005728922784328461, -0.009314266964793205, -0.0035090604797005653, 0.002566954120993614, 0.0013912280555814505, 0.06227819621562958, -0.02117641642689705, 0.010145762003958225, -0.029399823397397995, 0.0...
60a2c45cefee8a51341ea29d06c9588237713b2c
subsection
7
33
PRELIMINARIES
The \textit {product} \mathcal {I} \otimes \mathcal {A} = (Q \times S, {, \widehat{T^{\prime }}, Acc^{\prime }, L^{\prime }) is an Interval-valued Markov Chain where: \begin{} \item Q \times S is a set of states,\\ \item {_{ \left<Q_{j},s\right> \rightarrow \left<Q_{\ell },s^{\prime }\right>} = {\left\lbrace \begin{arr...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ 0.009757948108017445, 0.01276391465216875, -0.03067001700401306, -0.006618467625230551, 0.01240533497184515, 0.00760265439748764, 0.013793878257274628, 0.0238341111689806, -0.03887920454144478, 0.04016093537211418, -0.03793316334486008, 0.020965473726391792, -0.0029411171562969685, 0.02665...
9390a1867f62d4794f5336e96231fa40ab1ecb1b
subsection
8
33
PRELIMINARIES
\; F_{i} \in L^{\prime }(\left<Q_{j},s_{\ell } \right>) \; \Bigg ) \\ & \wedge \Bigg ( \; \forall \left<Q_{j},s_{\ell } \right> \in B \; . \; E_{i} \notin L^{\prime }(\left<Q_{j},s_{\ell } \right>) \; \Bigg ). \end{align} \end{}\mbox{}\\ }In words, every state in a BSCC B is reachable from any state in B, and every sta...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ 0.008262557908892632, -0.0031585399992763996, -0.03643764927983284, -0.022384434938430786, -0.00894919689744711, 0.008163375779986382, 0.03115815669298172, 0.001775725046172738, -0.004871323239058256, 0.0132521353662014, -0.039947137236595154, 0.027358753606677055, 0.01560959592461586, 0.0...
740fe8ccceaf86b24b0d9c2c3b8f0328d5fa9b9c
subsection
9
33
PRELIMINARIES
Indeed, for any two states Q_j and Q_{\ell } in \mathcal {I} and some states s, s^{\prime }, s^{\prime \prime } and s^{\prime \prime \prime } in \mathcal {A}, we allow T_{ \left<Q_{j},s\right> \rightarrow \left<Q_{\ell },s^{\prime }\right>} and T_{ \left<Q_{j},s^{\prime \prime }\right> \rightarrow \left<Q_{\ell },s^{\p...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 991, "openalex_id": "https://openalex.org/W1498432697", "raw": "C. Baier, J.-P. Katoen, and K. G. Larsen, Principles of model checking. MIT press, 2008.", "source_ref_id": "6c85a3916ca55ee529dab7740534e04894c88056", "s...
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.013258092105388641, 0.005583960562944412, -0.035700730979442596, -0.010916185565292835, -0.02128312923014164, 0.005297896917909384, 0.009512566961348057, 0.019787969067692757, 0.003531931433826685, 0.0464719794690609, -0.01572968065738678, 0.01737740822136402, 0.016919706016778946, -0.0...
8063a486984f783f469ab13a5dd294ec1ab7d235
subsection
10
33
PRELIMINARIES
However, the set of accepting and non-accepting states may not be fixed in product IMCs and varies as a function of the assumed values for each transition. Specifically, U^A and U^N are determined by transitions that can be turned ``on" or ``off", i.e. those whose lower bound is zero and upper bound non-zero, as seen i...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.010452776215970516, -0.017136448994278908, -0.04416107013821602, 0.003868289990350604, -0.015259526669979095, -0.013664905913174152, -0.007107124663889408, 0.03811829909682274, -0.0024834880605340004, 0.027436628937721252, -0.050783704966306686, -0.010468035005033016, -0.00234805955551564...
b27178b01bbcc9dad8c1bdd627d0035e5237248a
subsection
11
33
BOUNDING THE SATISFIABILITY OF
In , the authors discussed an algorithm for computing the probability bounds of reaching any fixed set of states in an IMC. We remarked in the previous section that, in general, the set of accepting and non-accepting states in a product IMC may depend on the assumed transition values. This is however not always the cas...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 123, "openalex_id": "", "raw": "K. Chatterjee, K. Sen, and T. Henzinger, “Model-checking \\omega -regular properties of interval Markov chains,” Foundations of Software Science and Computational Structures, pp. 302–317, 2008.", ...
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ 0.018552714958786964, -0.019025687128305435, -0.015409738756716251, 0.026257583871483803, -0.02255009301006794, -0.013876393437385559, 0.02596769854426384, -0.015791168436408043, 0.00979510135948658, 0.05532248690724373, -0.04943322017788887, 0.00421097781509161, 0.004626735579222441, 0.02...
f9812ff6e08ee34d866432ba228a26d2ee2e4295
subsection
12
33
BOUNDING THE SATISFIABILITY OF
Specifically, we prove that all product IMCs induce a worst-case NASIMCs containing the largest set of non-accepting states and in which the probability of reaching an accepting BSCC is minimized from any initial state. Then, we show that the converse best-case ASIMCs are always induced by product IMCs with one Rabin p...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.019113657996058464, -0.028525570407509804, -0.012371250428259373, 0.057844363152980804, -0.022271301597356796, 0.008786485530436039, 0.026283185929059982, -0.04024088382720947, -0.002797260181978345, 0.061230823397636414, -0.05473248288035393, 0.016078049317002296, 0.007131392601877451, ...
d2630e1b159e741e271d947a16cb5175a1be3cbd
subsection
13
33
Lower Bound Computation
A key observation is that any infinite sequence of states in a Markov Chain eventually reaches a BSCC.Lemma 1 For any infinite sequence of states \pi = q_{0}q_{1}q_{2}\ldots in a Markov Chain, there exists an index i \ge 0 such that q_i belongs to a BSCC.The following corollary relies on the fact that a BSCC is either ...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 257, "openalex_id": "https://openalex.org/W1498432697", "raw": "C. Baier, J.-P. Katoen, and K. G. Larsen, Principles of model checking. MIT press, 2008.", "source_ref_id": "6c85a3916ca55ee529dab7740534e04894c88056", "s...
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.01591600850224495, 0.010811593383550644, -0.041872989386320114, -0.006538839545100927, -0.017365694046020508, -0.007843555882573128, 0.004471132066100836, -0.00952976755797863, 0.00237481202930212, 0.045016516000032425, -0.017213094979524612, -0.003772994503378868, 0.0013228370808064938, ...
bb70b6767563824e9cd94f3ead95528cb0bdf238
subsection
14
33
Lower Bound Computation
Let {, \widehat{T}, {_{1}, \widehat{T}_{1}, {_{2}, \widehat{T}_{2} and {_{3}, \widehat{T}_{3} be the transition bounds functions in the product IMCs \mathcal {I} \otimes \mathcal {A}, (\mathcal {I} \otimes \mathcal {A})_{1}, (\mathcal {I} \otimes \mathcal {A})_{2} and (\mathcal {I} \otimes \mathcal {A})_{3} respectivel...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.0009377425303682685, 0.03597269952297211, -0.06566008925437927, -0.00009981689800042659, -0.008962663821876049, -0.0028356341645121574, -0.016765901818871498, 0.007765098940581083, -0.03438611701130867, 0.023356320336461067, -0.008657551370561123, -0.01007632166147232, -0.0256446599960327...
765885c4f09e938d3190e6fc23ffafb4ba1ee6e1
subsection
15
33
Lower Bound Computation
Set {_{3}(Q_{i}, Q_{j}) = {_{2}(Q_{i}, Q_{j}) and \widehat{T}_{3}(Q_{i}, Q_{j}) = \widehat{T}_{2}(Q_{i}, Q_{j}) for all Q_{i} \in U^N_2 \setminus D and for all Q_{j} \in Q \times S. Set {_{3}(Q_{i}, Q_{j}) > 0 for all Q_{i}, Q_{j} \in D such that {_{1}(Q_{i}, Q_{j}) > 0 and {_{2}(Q_{i}, Q_{j}) > 0. Set \widehat{T}_{3}(...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ 0.003944157622754574, 0.02520599029958248, -0.052120860666036606, -0.013289598748087883, -0.014075378887355328, -0.01710406504571438, -0.01762283220887184, 0.00245270156301558, -0.023008856922388077, 0.03259079530835152, -0.009078429080545902, 0.009719259105622768, 0.0013741613365709782, 0...
e22f68890a3e14675629f7fd9a28dfef0334c318
subsection
16
33
Lower Bound Computation
There exists a non-empty set of NASIMCs [\mathcal {I} \otimes \mathcal {A}]^{N}_{\ell } \subseteq [\mathcal {I} \otimes \mathcal {A}]^{N} such that, for all (\mathcal {I} \otimes \mathcal {A})_{\ell } \in [\mathcal {I} \otimes \mathcal {A}]^{N}_{\ell } with transition functions {_{\ell } and \widehat{T}_{\ell }, {_{\el...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.00800932664424181, 0.02370760776102543, -0.031137213110923767, 0.024027980864048004, -0.020107224583625793, -0.039146538823843, 0.029886232689023018, 0.02938278764486313, -0.012441154569387436, 0.04128235951066017, -0.019252896308898926, -0.010846917517483234, -0.0027079153805971146, 0....
83f9ed9359b515a1ba3086b6bd0d8af4c988a43a
subsection
17
33
Lower Bound Computation
For any NASIMC (\mathcal {I} \otimes \mathcal {A})_2 with non-accepting states U^N_2 induced by \mathcal {I} \otimes \mathcal {A}, there exists a NASIMC (\mathcal {I} \otimes \mathcal {A})_1 with non-accepting states U^N_1 induced by \mathcal {I} \otimes \mathcal {A} such that, for any initial state \left<Q_i, s_0\righ...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ 0.006866820622235537, 0.022965699434280396, -0.0499294139444828, -0.009735625237226486, -0.0318010076880455, -0.02908479981124401, 0.007568762172013521, 0.0318925678730011, -0.001486857421696186, 0.03412047028541565, -0.03918665647506714, -0.02314881607890129, -0.017197569832205772, 0.0042...
9f1821cc684e8dc03673e3950f4d9c0bcc5b1f5b
subsection
18
33
Lower Bound Computation
Since U^N_2 \subseteq U^N_1, if \left<Q_i,s_0\right> \notin U^N_1, it must be true that \mathcal {\widehat{P}}_{(\mathcal {I} \otimes \mathcal {A})_1}(\left<Q_i,s_0\right> \models \Diamond U^N_1) \ge \mathcal {\widehat{P}}_{(\mathcal {I} \otimes \mathcal {A})_2}(\left<Q_i,s_0\right> \models \Diamond U^N_2) \end{} }\mbo...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.03616683557629585, 0.03866952285170555, -0.06342167407274246, -0.011017913930118084, -0.03900524601340294, -0.03644152358174324, -0.02823149785399437, 0.014100488275289536, -0.026354484260082245, 0.02994064800441265, -0.024645334109663963, -0.011964051052927971, -0.026766512542963028, 0...
46ca0b9fe2bbd37d2b200d135503b38bb83e9dd7
subsection
19
33
Lower Bound Computation
If \left<Q_i, s_0\right> \notin U^{N}_{\ell }, the inequality follows from the fact that, by the definition of [\mathcal {I} \otimes \mathcal {A}]^{N}_{l}, for any Markov Chain induced by (\mathcal {I} \otimes \mathcal {A})^{^{\prime }}, there exists a Markov Chain induced by (\mathcal {I} \otimes \mathcal {A})_{\ell }...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.024783305823802948, 0.024493353441357613, -0.05206935852766037, -0.02516482211649418, -0.017320845276117325, 0.002258577151224017, -0.004574381746351719, 0.038731545209884644, 0.015886344015598297, 0.05173362419009209, -0.03867050260305405, 0.023501411080360413, -0.026370413601398468, 0...
c50dec18fe0eec399762db50e78f607e4ffffd2a
subsection
20
33
Lower Bound Computation
By Lemma 3 and Lemma 4, the maximum value for \mathcal {P}_{(\mathcal {M} \otimes \mathcal {A})_{\nu }}(\left<Q_i, s_0\right> \models \Diamond U^N) is reached for some (\mathcal {M} \otimes \mathcal {A})_{\nu } induced by the NASIMC (\mathcal {I} \otimes \mathcal {A})_{\ell }.
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.03655821830034256, 0.0385722778737545, -0.02699143998324871, -0.022764967754483223, -0.03570377081632614, -0.03942672908306122, 0.0018395452061668038, 0.005271646194159985, 0.0026930407620966434, 0.04241729900240898, 0.007087350822985172, 0.04388206824660301, -0.009772762656211853, 0.00...
844a4901b88711aa198e65462560e65f93195d75
subsection
21
33
Upper Bound Computation
One could think of a similar approach here and compute a satisfiability upper bound by maximizing the probability of transition to an accepting BSCC. However, due to the acceptance condition of Rabin Automata, the analogous version of Lemma 2 for accepting states does not always hold true, as shown in Fig. 2. We conseq...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.04673098400235176, 0.001134126796387136, -0.02164093218743801, 0.03397229686379433, -0.00839387346059084, 0.0011532037751749158, 0.011644591577351093, 0.011003605090081692, -0.012010869570076466, 0.04819609597325325, -0.04526587203145027, -0.00320111820474267, 0.007348454557359219, 0.02...
5ac6be09c250686db8b05b40ae10c6822bedbf4c
subsection
22
33
Product IMC with one Rabin pair
The theorem and lemma in this section are similar to the ones in section IV A, and are provided without proof due to space constraints.We denote by U^A_{u} the largest set of accepting states induced by \mathcal {I} \otimes \mathcal {A}. We define the set of best case ASIMCs [\mathcal {I} \otimes \mathcal {A}]^{A}_{u} ...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.004748156759887934, 0.018214616924524307, -0.035269845277071, -0.009107308462262154, -0.03012886829674244, -0.017299309372901917, 0.004317199811339378, 0.004648998845368624, 0.03475116938352585, 0.03414096310734749, -0.014362698420882225, 0.02942713163793087, -0.007242369465529919, 0.00...
13c0ee49c7520374574e29f1ac247f0df51efcdd
subsection
23
33
Product IMC with more than one Rabin pairs
We previously observed that product IMCs with more than one Rabin pair don't necessarily induce a unique largest set of accepting states. Instead, we exploit the fact that \omega -regular expressions, and consequently DRAs, are closed under complementation . The following theorem states that any \omega -regular propert...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 258, "openalex_id": "", "raw": "B. Farwer, “\\omega -automata,” in Automata logics, and infinite games. Springer, 2002, pp. 3–21.", "source_ref_id": "c9391c4d026975ac7cc8539fa7e4fb99e132eb60", "start": 138 } ] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.007958942092955112, 0.02834085375070572, -0.0220835842192173, 0.0017073953058570623, -0.03443024680018425, -0.0350712351500988, -0.001933458261191845, 0.021534165367484093, 0.010377910919487476, 0.007272168528288603, -0.06312211602926254, 0.013407344929873943, -0.04538809880614281, -0.0...
da8d4f3687615e31fb9c6c24d7a10c4be96a0696
subsection
24
33
Product IMC with more than one Rabin pairs
By theorem 1, we have that\mathcal {P}_{\mathcal {I}[\mathcal {\nu }_{\ell }]}(Q_i \models \lnot \phi ) = 1 - \mathcal {\widehat{P}}_{(\mathcal {I} \otimes \mathcal {\overline{A}})_{\ell }}(\left<Q_i, s_0\right> \models \Diamond U^N_{\ell }) \;\; .Therefore, the first inequality reduces to\mathcal {\widehat{P}}_{\mathc...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.020873332396149635, 0.04815514385700226, -0.03369028866291046, -0.042692676186561584, -0.0364062637090683, -0.02984520234167576, -0.018477780744433403, 0.041990794241428375, 0.005073379259556532, 0.02982994355261326, -0.012282918207347393, -0.009406731463968754, -0.02346723899245262, -0...
36fe26f519288b8862bea62f5a21ec5a410876df
subsection
25
33
Search Algorithm
[t!]InputInput OutputOutput Lower and upper bound probabilities of satisfying \phi in \mathcal {I}, {_{\mathcal {I}}(Q_{i} \models \phi ) and \widehat{\mathcal {P}}_{\mathcal {I}}(Q_{i} \models \phi ), for all initial states Q_{i}.} \mbox{}\\Construct a DRA A corresponding to ; \mbox{}\\ Generate the product I A; \mbo...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1016/0898-1221(81)90008-0", "end": 1900, "openalex_id": "https://openalex.org/W2075410688", "raw": "M. Sharir, “A strong-connectivity algorithm and its applications in data flow analysis,” Computers & Mathematics with Applications, vol. ...
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.030387798324227333, 0.018305903300642967, -0.05412445217370987, 0.01755841262638569, -0.0315166637301445, 0.0010659374529495835, 0.013065838254988194, -0.004793858155608177, -0.0030586114153265953, 0.05040225386619568, -0.036428745836019516, 0.03554396331310272, 0.020883984863758087, 0....
71ff533fccb0a62e39c4cd4b08f51f29f2962d20
subsection
26
33
Search Algorithm
The resulting product IMC induces a directed graph with a vertex for each state and an edge for all non-zero transitions. In this graph, all SCCs are enumerated. Then, for each SCC and if necessary, we remove the states that prevent it from being a BSCC and accepting (or non-accepting) if allowed by \mathcal {I} \otime...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.05020459368824959, 0.005058608949184418, -0.04056042805314064, 0.006702847313135862, -0.014038974419236183, -0.04373445734381676, -0.00918637216091156, -0.003557430347427726, -0.01870846003293991, 0.05945200473070145, 0.0015631711576133966, -0.004249841906130314, 0.004181172698736191, 0...
c4f3e6254485dff4dbdb9afa4f6551e81bc84b55
subsection
27
33
Search Algorithm
\item \underline{Search for U^{N}_{l}}: For all such F_{i}^{\prime }s, check whether some state in C^{j} maps to the corresponding non-accepting set E_{i}. If this is the case for all such F_{i}^{\prime }s, C^{j} is a non-accepting BSCC. Otherwise, the unmatched F_{i} states cannot belong to a non-accepting BSCC. Treat...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.029853811487555504, -0.01300380751490593, -0.03073904663324356, 0.02666391059756279, -0.00374889699742198, 0.0024000571575015783, 0.01181331742554903, -0.012034626677632332, -0.02542763203382492, 0.04694801941514015, -0.03482944518327713, 0.004586436785757542, 0.009378919377923012, 0.02...
069d32b40d9de292f2f4a6918b1bd21852565e55
subsection
28
33
Search Algorithm
Matrices \widehat{T} and { respectively contain the upper and lower probabilities of transition from state to state: }\begin{} \begin{}{c || c c c c c c} { & q_0 & q_1 & q_2 & q_3 & q_4 & q_5\\ \hline \hline q_0 & 0 & 0.2 & 0 & 0.3 & 0.2 & 0 \\ q_1 & 0 & 0.05 & 0.25 & 0 & 0.1 & 0\\ q_2 & 0 & 0 & 0 & 0 & 1 & 0\\ q_3 & 0...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1007/978-3-319-11936-6_17", "end": 1983, "openalex_id": "https://openalex.org/W141312837", "raw": "Z. Komárková and J. Křetínskỳ, “Rabinizer 3: Safraless translation of LTL to small deterministic automata,” in International Symposium on ...
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.04942931979894638, 0.02132783643901348, -0.01934456080198288, 0.031091652810573578, 0.004756046459078789, -0.025919880717992783, -0.005900243762880564, 0.024134933948516846, 0.004981071688234806, 0.041770827025175095, -0.04433382675051689, -0.008207707665860653, 0.016186576336622238, 0....
6b2f14f7ba4ee820efaa5ec20b127da294054a63
subsection
29
33
Search Algorithm
We demonstrated its application through a case study. In future works, we will seek to exploit the mechanisms unveiled in this paper and apply them to Bounded-parameter Markov Decision Processes, the controllable counterparts of IMCs, e.g. to minimize or maximize the probability of occurrence of some behavior in a syst...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.02912209928035736, -0.010777769610285759, -0.025094736367464066, 0.03615472838282585, -0.04134147986769676, 0.02117416076362133, 0.03389696404337883, 0.02208947017788887, 0.018443487584590912, 0.036673404276371, -0.06535310298204422, -0.003769931150600314, 0.007600883021950722, 0.014339...
517bd1dfbb301b56f0e9839b4b2247df1adcb3a7
subsection
30
33
CASE STUDY
We now apply the concepts developed in the previous sections to a case study. Our system of interest is an agent moving stochastically on a two-dimensional grid shown in Fig. 3. The grid is divided into 6 locations, representing 6 different states the agent can visit. We assume the system to be evolving in a discrete-t...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.013018360361456871, -0.0030104005709290504, -0.03305717185139656, 0.010065192356705666, -0.01666594296693802, -0.0033995776902884245, -0.015643399208784103, 0.008920553140342236, 0.010774868540465832, 0.02486155740916729, -0.015857065096497536, -0.0004177932278253138, -0.01259865984320640...
8827ff260d1cd11d445c74b8f3022453b407bb64
subsection
31
33
CASE STUDY
We aim to bound the probability of satisfying \omega -regular properties \phi _{1} and \phi _{2}, represented by automata \mathcal {A}_{1} and \mathcal {A}_{2} in Fig. 4, from every initial state q. In natural language, these properties respectively translate to1) “The agent visits a green state infinitely many times w...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1007/978-3-319-11936-6_17", "end": 949, "openalex_id": "https://openalex.org/W141312837", "raw": "Z. Komárková and J. Křetínskỳ, “Rabinizer 3: Safraless translation of LTL to small deterministic automata,” in International Symposium on A...
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.0645904466509819, -0.007585409563034773, 0.009264272637665272, 0.004517668392509222, -0.030936870723962784, -0.03363831341266632, 0.0009605768718756735, -0.00764264352619648, 0.008630883879959583, 0.03705709055066109, -0.041269510984420776, 0.014735841192305088, 0.022328881546854973, 0....
64af6aaeedb0c841f97ac210b6b207baf544dfe9
subsection
32
33
CONCLUSIONS
We derived an efficient automaton-based technique for bounding the probability of satisfying any \omega -regular property in an IMC interpreted as an IMDP. We demonstrated its application through a case study. In future works, we will seek to exploit the mechanisms unveiled in this paper and apply them to Bounded-param...
{ "cite_spans": [] }
1809.06352
Satisfiability Bounds for {\omega}-regular Properties in Interval-valued Markov Chains
[ "Maxence Dutreix", "Samuel Coogan" ]
[ "cs.SY" ]
2,018
en
Computer Science
[ -0.03092006780207157, -0.014506181702017784, -0.012033797800540924, 0.029348120093345642, -0.02754724770784378, -0.04599856212735176, 0.05481978505849838, 0.013689684681594372, 0.04315989837050438, 0.01703198254108429, -0.06818897277116776, 0.006905888672918081, -0.005730743054300547, 0.01...
7729bb486ed0f3e43385ecb5178683779b079e0a
abstract
0
52
Abstract
GANs excel at learning high dimensional distributions, but they can update generator parameters in directions that do not correspond to the steepest descent direction of the objective. Prominent examples of problematic update directions include those used in both Goodfellow's original GAN and the WGAN-GP. To formally d...
{ "cite_spans": [] }
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.026265796273946762, -0.023304980248212814, -0.00997367687523365, 0.0010797821450978518, 0.0021385790314525366, -0.01771911047399044, 0.01759701408445835, -0.022953955456614494, -0.008806138299405575, 0.043832287192344666, -0.02394598163664341, -0.040566232055425644, -0.008302494883537292,...
ced99e7c8cdc7b1fc6970a4b5e00a889bf752535
subsection
1
52
Introduction
5mm Generative adversarial networks (GANs) excel at learning generative models of complex distributions, such as images , , textures , , , and even texts , .GANs learn a generative model G that maps samples from multivariate random noise into a high dimensional space. The goal of GAN training is to update G such that t...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.3156/jsoft.29.5_177_2", "end": 158, "openalex_id": "https://openalex.org/W2099471712", "raw": "Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., and Bengio, Y. Generative adversarial nets. ...
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.009903641417622566, -0.018678054213523865, -0.030260277912020683, -0.03004663996398449, 0.028795333579182625, -0.029878782108426094, 0.026018040254712105, 0.0037081430200487375, -0.0024263157974928617, 0.03622687980532646, -0.041659384965896606, -0.03390737995505333, -0.040347035974264145...
042a665736d9a0e6ca8531e285856251442ff2ce
subsection
2
52
Introduction
As we see later, popular methods such as WGAN-GP are affected by this issue.Therefore we set out to answer a simple but fundamental question: Is there an adversarial divergence and corresponding method that produces unbiased estimates of the direction of steepest descent in a mini-batch setting?In this paper, under rea...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.48550/arxiv.1704.00028", "end": 77, "openalex_id": "https://openalex.org/W4295521014", "raw": "Gulrajani, I., Ahmed, F., Arjovsky, M., Dumoulin, V., and Courville, A. C. Improved Training of Wasserstein GANs. In Advances in Neural Inform...
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.03311461955308914, -0.0030215182341635227, -0.031954843550920486, 0.012063182890415192, 0.022310402244329453, -0.01634366810321808, 0.03497636318206787, -0.0020944615826010704, 0.015222042798995972, 0.049107301980257034, -0.027971934527158737, -0.01078895665705204, -0.008980623446404934, ...
4dd05efd5a4bbd5ad5f47e7d1dcaeca37b58f062
subsection
3
52
Introduction
This divergence enforces not just correct values for the critic, but also ensures that the critic's gradient, its first order information, assumes values that allow for an easy formulation of an update rule. Together, this divergence and update rule fulfill all four requirements.We hope that this gradient penalty trick...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.48550/arxiv.1710.08446", "end": 492, "openalex_id": "https://openalex.org/W2766290211", "raw": "Fedus, W., Rosca, M., Lakshminarayanan, B., Dai, A. M., Mohamed, S., and Goodfellow, I. Many paths to equilibrium: GANs do not need to decrea...
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.036112427711486816, -0.0036993869580328465, -0.014965441077947617, 0.009966785088181496, 0.00863889791071415, -0.06611962616443634, 0.0009825602173805237, 0.03135034814476967, -0.012172299437224865, 0.004735367838293314, -0.04817025363445282, -0.02521459385752678, -0.03235771134495735, ...
6ce4f3c9042ab244dec645771275b24c592631b0
subsection
4
52
Notation, Definitions and Assumptions
In an adversarial divergence is defined:Definition 1 (Adversarial Divergence) Let X be a topological space, C(X^2) the set of all continuous real valued functions over the Cartesian product of X with itself and set \mathcal {G}\subseteq C(X^2), \mathcal {G}\ne \emptyset . An adversarial divergence \tau (\cdot \Vert \c...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.48550/arxiv.1705.08991", "end": 274, "openalex_id": "https://openalex.org/W2620086128", "raw": "Liu, S., Bousquet, O., and Chaudhuri, K. Approximation and convergence properties of generative adversarial learning. In Advances in Neural I...
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.03692866116762161, -0.010147752240300179, -0.040865685790777206, 0.0032465176191180944, 0.00968232937157154, -0.03326631709933281, 0.015152959153056145, 0.01454256847500801, 0.041231919080019, 0.02386629208922386, 0.006790601182729006, -0.06140534579753876, -0.027818575501441956, 0.0154...
8c7b5090a5893eeaa7124238ae808e4d72113a22
subsection
5
52
Notation, Definitions and Assumptions
Then define\tau :\mathcal {P}(X)\times \mathcal {P}(X)\times \mathcal {F}&\rightarrow \mathbb {R}\cup \lbrace +\infty \rbrace \\ (\mathbb {P},\mathbb {Q},f)&\mapsto \tau (\mathbb {P}\Vert \mathbb {Q};f)=\mathbb {E}_{\mathbb {P}\otimes \mathbb {Q}}[m_f - r_f]and set \tau (\mathbb {P}\Vert \mathbb {Q})=\sup _{f\in \mathc...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.48550/arxiv.1704.00028", "end": 988, "openalex_id": "https://openalex.org/W4295521014", "raw": "Gulrajani, I., Ahmed, F., Arjovsky, M., Dumoulin, V., and Courville, A. C. Improved Training of Wasserstein GANs. In Advances in Neural Infor...
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.02112436294555664, -0.002026198897510767, -0.023978592827916145, -0.009890597313642502, 0.011844296008348465, -0.0023352799471467733, 0.02155173383653164, 0.06206044554710388, 0.0371507927775383, 0.04884245619177818, -0.01981172151863575, -0.031656015664339066, -0.03373182192444801, 0.0...
9ac3e2ec8eab227d593730d1b362afe03407b556
subsection
6
52
Notation, Definitions and Assumptions
\tau is called a strict adversarial divergence if for any \mathbb {P},\mathbb {P}^*\in \mathcal {P}(X),\tau (\mathbb {P}^*\Vert \mathbb {P})=\inf _{\mathbb {P}^{\prime }\in \mathcal {P}(X)}\tau (\mathbb {P}^*\Vert \mathbb {P}^{\prime })\Rightarrow \mathbb {P}^*=\mathbb {P}In order to analyze GANs that minimize a critic...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.48550/arxiv.1701.04862", "end": 1071, "openalex_id": "https://openalex.org/W2964201867", "raw": "Arjovsky, M. and Bottou, L. Towards principled methods for training generative adversarial networks. In International Conference of Learning...
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.03445923328399658, 0.022311514243483543, -0.0495065338909626, 0.005650368053466082, 0.012468199245631695, -0.01603926345705986, 0.008851353079080582, 0.022448863834142685, 0.055916134268045425, 0.07722042500972748, -0.017366966232657433, -0.024371743202209473, -0.04303589090704918, 0.02...
73dd0387d4c50bba57dbf987a65dd7ce28cd4195
subsection
7
52
Notation, Definitions and Assumptions
We say \mathbb {Q}_\theta \in \mathcal {P}(X), \theta \in \Theta satisfies assumption REF if there is a locally Lipschitz function g:\Theta \times \mathbb {R}^d\rightarrow X which is differentiable in the first argument and a distribution \mathbb {Z} with bounded support in \mathbb {R}^d such that for all \theta \in \...
{ "cite_spans": [] }
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.03223401680588722, 0.01613227091729641, 0.0009534179698675871, 0.004742765333503485, 0.004746580962091684, -0.06245340779423714, 0.021016212180256844, -0.01545309741050005, 0.01277456060051918, 0.0057615251280367374, 0.010592048987746239, 0.001210492686368525, -0.0470995157957077, 0.029...
b31ed25df7e7c577c5168f0128b6ba811a4dbcf0
subsection
8
52
Requirements Derived From Related Work
With the concept of an Adversarial Divergence now formally defined, we can investigate existing GAN methods from an Adversarial Divergence minimization standpoint. During the last few years, weaknesses in existing GAN frameworks have been highlighted and new frameworks have been proposed to mitigate or eliminate these ...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1162/neco.1992.4.6.863", "end": 649, "openalex_id": "https://openalex.org/W1994618660", "raw": "Schmidhuber, J. Learning factorial codes by predictability minimization. Neural Computation, 4(6):863–879, 1992.", "source_ref_id": "f6...
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.038726914674043655, -0.0074691977351903915, -0.02754219062626362, -0.029983602464199066, 0.011390717700123787, 0.0005459800595417619, 0.06439226865768433, 0.05365004763007164, 0.03439340367913246, 0.05416885018348694, -0.014419594779610634, -0.03384408727288246, -0.030822837725281715, 0...
610f592ebc99c0f7575676b65bb3deec12852818
subsection
9
52
Requirements Derived From Related Work
The Wasserstein distance \tau _W is the weakest divergence in the class of strict adversarial divergences , leading to the following requirement:Requirement 1 (Equivalence to \tau _W) An adversarial divergence \tau is said to fulfill Requirement REF if \tau is a strict adversarial divergence which is weaker than \tau ...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.48550/arxiv.1705.08991", "end": 326, "openalex_id": "https://openalex.org/W2620086128", "raw": "Liu, S., Bousquet, O., and Chaudhuri, K. Approximation and convergence properties of generative adversarial learning. In Advances in Neural I...
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.0027166607324033976, 0.0049869040958583355, -0.01700202375650406, -0.009721983224153519, 0.03000536561012268, 0.005910263396799564, 0.034828200936317444, 0.004300107713788748, 0.01689518801867962, 0.042428746819496155, -0.01733778975903988, -0.006654292810708284, -0.0198560431599617, 0....
c1371f26a481f0d9c8f2535ec4ebf168fd8f3821
subsection
10
52
Requirements Derived From Related Work
Although this method has impressive experimental results, it is not yet ideal. showed that an optimal critic for \tau _I has undefined gradients on the support of the generated distribution \mathbb {Q}_\theta . Thus, the update direction \nabla _\theta \mathbb {E}_{\mathbb {Q}_\theta }[f^*] is undefined; even if a dir...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.48550/arxiv.1709.08894", "end": 211, "openalex_id": "https://openalex.org/W2759973161", "raw": "Petzka, H., Fischer, A., and Lukovnicov, D. On the regularization of wasserstein GANs. In International Conference on Learning Representation...
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.016780301928520203, 0.002806886797770858, -0.025536566972732544, -0.012325894087553024, 0.008222347125411034, 0.00032583341817371547, 0.03237072750926018, 0.006311681587249041, -0.03151645511388779, 0.029899446293711662, -0.03883877024054527, 0.0024483983870595694, -0.053178299218416214, ...
8fd35c074049fe4c18bb28a7b75eddd15ef7bf4a
subsection
11
52
Correct Update Rule Requirement
In the previous section, we stated a bare minimum requirement for an update rule (namely that it is well defined). In this section, we'll go further and explore criteria for a “good” update rule. For example in Lemma REF in Section of Appendix, it is shown that there exists a target \mathbb {P} and a family of generate...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.48550/arxiv.1801.04406", "end": 1088, "openalex_id": "https://openalex.org/W2792263949", "raw": "Mescheder, L., Geiger, A., and Nowozin, S. Which training methods for gans do actually converge. arXiv preprint arXiv:1801.04406v2, 2018.", ...
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.032813627272844315, 0.01314071286469698, -0.033240966498851776, 0.018711399286985397, 0.012705741450190544, -0.04453495889902115, 0.02451101690530777, 0.0214280616492033, 0.007085454184561968, 0.043130841106176376, -0.03134845942258835, 0.013751199468970299, -0.038857437670230865, 0.026...
a4b89a9f9dc0197d2371210757a270eb228dc968
subsection
12
52
Correct Update Rule Requirement
REF together with Theorem 1 from is that for every f^*\in \operatorname{OC}_{\tau }(\mathbb {P},\mathbb {Q}_{\theta _0})\nabla _\theta \tau (\mathbb {P}\Vert \mathbb {Q}_\theta )|_{\theta _0} &=\nabla _\theta \tau (\mathbb {P}\Vert \mathbb {Q};f^*)\\ &=-\nabla _\theta (\mathbb {E}_{\mathbb {Q}_\theta }[m_2(f^*)]+\math...
{ "cite_spans": [] }
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.02614670991897583, 0.007127800490707159, -0.025811104103922844, -0.021448234096169472, 0.01266147568821907, 0.005648085847496986, 0.05479520559310913, 0.03905226290225983, 0.02068549580872059, 0.03777085989713669, -0.042469337582588196, 0.008336742408573627, -0.022394031286239624, 0.026...
b4efb823a6c4e6cc69164329b740bd774b71cd0e
subsection
13
52
Correct Update Rule Requirement
GAN learning happens in mini-batches, therefore \nabla _\theta \mathbb {E}_{\mathbb {P}\otimes \mathbb {Q}_\theta }[r_{f^*}] isn't calculated directly, but estimated based on samples which can lead to variance in the estimate.To analyze this issue, we use the notation from where \mathbf {X}_m := X_1, X_2,\ldots , X_m a...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.48550/arxiv.1705.10743", "end": 514, "openalex_id": "https://openalex.org/W2619903301", "raw": "Bellemare, M. G., Danihelka, I., Dabney, W., Mohamed, S., Lakshminarayanan, B., Hoyer, S., and Munos, R. The cramer distance as a solution to...
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.016788959503173828, -0.011607276275753975, -0.02237510494887829, 0.013980625197291374, -0.01628529094159603, -0.006658348720520735, 0.03589784726500511, 0.05995184928178787, -0.020894624292850494, 0.006906367838382721, -0.028556494042277336, -0.009127088822424412, -0.04783327504992485, ...
1d95167142813a7cdaae05954ae7f6cea4027e59
subsection
14
52
Correct Update Rule Requirement
In the same way, because the second expectation doesn't depend on the mini-batch \mathbf {X}_m sampled, \mathbb {V}_{\mathbf {X}_m\sim \mathbb {P}}[\mathbb {E}_{\mathbb {Q}_\theta }[m_2(f^*)]]=0.&\,\mathbb {V}_{\mathbf {X}_m\sim \mathbb {P}}[\nabla _\theta \mathbb {E}_{\hat{\mathbb {P}}_m\otimes \mathbb {Q}_\theta }[m_...
{ "cite_spans": [] }
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.04381673038005829, -0.00009487649367656559, -0.024562999606132507, 0.016507556661963463, -0.0074871378019452095, 0.036096930503845215, 0.014379271306097507, 0.05406910926103592, 0.03304562345147133, 0.016339734196662903, -0.023083116859197617, 0.0011985903838649392, -0.02738545462489128, ...
6a27b6c4f2b418e371b6f7733c1b3e4d7a9b1dca
subsection
15
52
Correct Update Rule Requirement
REF we see that\nabla _\theta \tau (\mathbb {P}\Vert \mathbb {Q}_\theta )|_{\theta _0}&=-\nabla _\theta (\mathbb {E}_{\mathbb {Q}_\theta }[m_2(f^*)]+\mathbb {E}_{\mathbb {P}\otimes \mathbb {Q}_\theta }[r_{f^*}])|_{\theta _0}\\ &\approx -\nabla _\theta (\mathbb {E}_{\mathbb {Q}_\theta }[m_2(f^*)]+\gamma \mathbb {E}_{\ma...
{ "cite_spans": [] }
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.021631892770528793, 0.01451533567160368, -0.003224570769816637, -0.022791288793087006, 0.014553474262356758, 0.010205738246440887, 0.02935103140771389, 0.03237156197428703, 0.022562460973858833, 0.04875566437840462, -0.026391519233584404, 0.007330130320042372, -0.015728125348687172, 0.0...
80070a5bc16b851e1e4d3f22d7a138f084089181
subsection
16
52
Penalized Wasserstein Divergence
We now attempt to find an adversarial divergence that fulfills all four requirements. We start by formulating an adversarial divergence \tau _P and a corresponding update rule than can be shown to comply with Requirements REF and REF . Subsequently in Section , \tau _P will be refined to make its update rule practical ...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.48550/arxiv.1704.00028", "end": 1430, "openalex_id": "https://openalex.org/W4295521014", "raw": "Gulrajani, I., Ahmed, F., Arjovsky, M., Dumoulin, V., and Courville, A. C. Improved Training of Wasserstein GANs. In Advances in Neural Info...
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.03607915714383125, 0.007859883829951286, -0.03479715809226036, -0.01706281676888466, 0.01408674381673336, 0.010469671338796616, 0.051188450306653976, 0.01368230301886797, 0.04181763529777527, 0.03620125353336334, -0.03223315626382828, 0.013270231895148754, -0.033698298037052155, 0.01050...
d1e997d7348c5d087faccfbe2abe3c2e07834c6d
subsection
17
52
Penalized Wasserstein Divergence
Set\tau _P(\mathbb {P}\Vert \mathbb {Q};f):=\,&\mathbb {E}_{x\sim \mathbb {P}}[f(x)]-\mathbb {E}_{x^{\prime }\sim \mathbb {Q}}[f(x^{\prime })] \\ -\lambda \,&\mathbb {E}_{x\sim \mathbb {P},x^{\prime }\sim \mathbb {Q}}\left[\frac{(f(x)-f(x^{\prime }))^2}{\Vert x-x^{\prime }\Vert }\right].Define the penalized Wasserstein...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.48550/arxiv.1704.00028", "end": 741, "openalex_id": "https://openalex.org/W4295521014", "raw": "Gulrajani, I., Ahmed, F., Arjovsky, M., Dumoulin, V., and Courville, A. C. Improved Training of Wasserstein GANs. In Advances in Neural Infor...
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ 0.0005792123265564442, 0.005777821876108646, -0.028603078797459602, -0.0037164934910833836, 0.0035753848496824503, -0.0003956759173888713, 0.03954089805483818, 0.03130320832133293, 0.020502686500549316, 0.06870841234922409, -0.021738341078162193, -0.005331614054739475, -0.024453725665807724,...
d491d12b802de6095752c5fdf5cf3d02ea35c7ea
subsection
18
52
Penalized Wasserstein Divergence
See Figure REF for a simple example. [Figure: Comparison of \tau _P update rule given different optimal critics.Consider the simple example of divergence \tau _P from Definition between Dirac measureswith update rule \frac{1}{2} \frac{d}{d\theta }\mathbb {E}_{\delta _\theta }[f](the update rule is from Lemma in Appendi...
{ "cite_spans": [ { "arxiv_id": "", "doi": "", "end": 1549, "openalex_id": "https://openalex.org/W2619459889", "raw": "Kodali, N., Abernethy, J., Hays, J., and Kira, Z. How to train your DRAGAN. arXiv preprint arXiv:1705.07215, 2017.", "source_ref_id": "f2090e04199a4d842eee0636...
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.0557311587035656, 0.01712222397327423, -0.05228229984641075, 0.0030234765727072954, 0.010979892686009407, -0.013383413664996624, 0.02763667330145836, 0.02652266062796116, 0.05090885981917381, 0.055303867906332016, -0.024859271943569183, -0.013711513951420784, -0.04388905316591263, -0.00...
468ef670f84edc2812dd5133950929d42253832f
subsection
19
52
Penalized Wasserstein Divergence
In response, we propose the First Order Penalized Wasserstein Divergence.
{ "cite_spans": [] }
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.009542603977024555, 0.03359362483024597, -0.02729291468858719, -0.014744886197149754, -0.002555373590439558, -0.023555204272270203, -0.003003517398610711, 0.039909593760967255, 0.021358344703912735, 0.05373149365186691, 0.017971521243453026, -0.02692677080631256, -0.014035483822226524, ...
912b01085ec23d1e4b6b11b4abb7ef4fdca1aabf
subsection
20
52
First Order Penalized Wasserstein Divergence
As was seen in the last section, since \tau _P only constrains the value of optimal critics on the supports of \mathbb {P} and \mathbb {Q}_\theta , the gradient \nabla _\theta \mathbb {E}_{\mathbb {Q}_\theta }[f^*] is not well defined. A natural method to refine \tau _P to achieve a well defined gradient is to enforce ...
{ "cite_spans": [] }
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.003173249773681164, -0.010053709149360657, -0.027079414576292038, -0.008291640318930149, 0.013196446932852268, 0.010053709149360657, 0.02775067836046219, 0.03441755473613739, 0.017071472480893135, 0.056020062416791916, -0.030847648158669472, 0.02273145318031311, -0.017193522304296494, -...
ed0bef52b8bb341733e6cac5430926443887f303
subsection
21
52
First Order Penalized Wasserstein Divergence
Set \mathcal {F}=C^1(X), \lambda ,\mu >0 and&\tau _F(\mathbb {P}\Vert \mathbb {Q};f):=\mathbb {E}_{x\sim \mathbb {P}}[f(x)]-\mathbb {E}_{x^{\prime }\sim \mathbb {Q}}[f(x^{\prime })]\\ &-\lambda \operatornamewithlimits{\mathbb {E}}_{x\sim \mathbb {P},x^{\prime }\sim \mathbb {Q}}\left[\frac{(f(x)-f(x^{\prime }))^2}{\Vert...
{ "cite_spans": [] }
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.005484350491315126, -0.014874108135700226, -0.029351573437452316, -0.03530121594667435, 0.0036308078560978174, -0.04668181762099266, 0.021388204768300056, 0.057147085666656494, 0.03395873308181763, 0.061632201075553894, -0.03328749164938927, -0.01974061131477356, -0.025613976642489433, ...
ea770dab99f41c23b4fd982c4fc25a4f301c0854
subsection
22
52
First Order Penalized Wasserstein Divergence
Therefore updates to \theta that reduce \tau _F(\mathbb {P}\Vert \mathbb {Q}^{\prime }_\theta ) also reduce \tau _F(\mathbb {P}\Vert \mathbb {Q}_\theta ).Conveniently, as is shown in Lemma REF in Appendix, Section , any optimal critic for the First Order Penalized Wasserstein divergence is also an optimal critic for th...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.48550/arxiv.1704.00028", "end": 1882, "openalex_id": "https://openalex.org/W4295521014", "raw": "Gulrajani, I., Ahmed, F., Arjovsky, M., Dumoulin, V., and Courville, A. C. Improved Training of Wasserstein GANs. In Advances in Neural Info...
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.00972512736916542, -0.0021662244107574224, -0.023264029994606972, -0.021418161690235138, 0.014705917797982693, -0.02805413119494915, 0.028221936896443367, 0.05842703580856323, 0.02974744699895382, 0.06022713705897331, -0.03673428297042847, -0.022089386358857155, -0.04732132330536842, 0....
45c761b6cf54e81b37eb39d9cef0b9d757ae5775
subsection
23
52
Image Generation
We begin by testing the FOGAN on the CelebA image generation task , training a generative model with the DCGAN architecture and obtaining Fréchet Inception Distance (FID) scores competitive with state of the art methods without doing a tuning parameter search. Similarly, we show competitive results on LSUN and CIFAR-1...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.1109/iccv.2015.425", "end": 262, "openalex_id": "https://openalex.org/W1834627138", "raw": "Liu, Z., Luo, P., Wang, X., and Tang, X. Deep learning face attributes in the wild. In Proceedings of the IEEE International Conference on Comput...
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.01992715895175934, -0.016692427918314934, -0.045011576265096664, 0.0005702737835235894, -0.022124333307147026, -0.00006180747732287273, 0.023360244929790497, -0.020720582455396652, -0.022048043087124825, 0.010818036273121834, -0.05001625418663025, -0.020964713767170906, -0.019683027639985...
392325d328a10352a1aa56f4f2defa678d492405
subsection
24
52
One Billion Word
Finally, we use the First Order Penalized Wasserstein Divergence to train a character level generative language model on the One Billion Word Benchmark . In this setting, a 1D CNN deterministically transforms a latent vector into a 32\times C matrix, where C is the number of possible characters. A softmax nonlinearity ...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.21437/interspeech.2014-564", "end": 153, "openalex_id": "https://openalex.org/W2611669587", "raw": "Chelba, C., Mikolov, T., Schuster, M., Ge, Q., Brants, T., Koehn, P., and Robinson, T. One billion word benchmark for measuring progress ...
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.03128591552376747, -0.03778728097677231, -0.034734997898340225, -0.03821460157632828, 0.020145077258348465, -0.016787564381957054, 0.0049065472558140755, 0.016726519912481308, -0.015192746184766293, 0.025303438305854797, -0.026066510006785393, -0.018069524317979813, 0.008645595982670784, ...
4f52057d0395ad80da6b39d14412d81cc4c2fbf8
subsection
25
52
Proof of Things
[Proof of Theorem REF ] The proof of this theorem is split into smaller lemmas that are proven individually.That \tau _P is a strict adversarial divergence which is equivalent to \tau _W is proven in Lemma REF , thus showing that \tau _P fulfills Requirement REF . \tau _P fulfills Requirement REF by design. The exist...
{ "cite_spans": [] }
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.026657938957214355, 0.042115576565265656, -0.021973343566060066, -0.011474968865513802, 0.007774596102535725, 0.00716041075065732, 0.017624452710151672, 0.0056802621111273766, 0.03640861064195633, 0.06286817789077759, -0.020279565826058388, 0.009552300907671452, -0.024536900222301483, 0...
9acc61076b5ea63d8c85453d162db1ead0f1400f
subsection
26
52
Proof of Things
Now consider \gamma \in (0,1), then&\mathbb {E}_{x\sim \mathbb {P},x^{\prime }\sim \mathbb {Q}}\left[\frac{(\gamma (f(x)-f(x^{\prime })) + (1-\gamma )(\hat{f}(x)-\hat{f}(x^{\prime })))^2}{\Vert x-x^{\prime }\Vert }\right] \\ \le &\mathbb {E}_{x\sim \mathbb {P},x^{\prime }\sim \mathbb {Q}}\left[\frac{\gamma (f(x)-f(x^{\...
{ "cite_spans": [] }
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.01820835843682289, 0.009226280264556408, -0.03580620884895325, -0.033638909459114075, -0.0017370819114148617, -0.012393282726407051, 0.03568410873413086, 0.006147037725895643, 0.021688245236873627, 0.03269262611865997, -0.010630445554852486, 0.010370980016887188, -0.02060459554195404, -...
a67c612963f7f88a232c036a731f4b538fe1454c
subsection
27
52
Proof of Things
REF and REF , then f\in \operatorname{OC}_{\tau _P}(\mathbb {P},\mathbb {Q})Since in Lemma REF it was shown that the the mapping f\mapsto \tau _P(\mathbb {P}\Vert \mathbb {Q},f) is concave, f\in \operatorname{OC}_\tau (\mathbb {P},\mathbb {Q}) if and only if f\in C^1(X) and f is a local maximum of \tau _P(\mathbb {P}\V...
{ "cite_spans": [] }
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.030316919088363647, 0.027692606672644615, -0.03101876936852932, -0.04061582311987877, 0.011283020488917828, -0.00006800374831072986, 0.014166713692247868, 0.023786650970578194, 0.040158092975616455, 0.018965238705277443, -0.009177466854453087, 0.01627989485859871, 0.006053466349840164, ...
64073234b49e03c31192008795b579aa29604892
subsection
28
52
Proof of Things
Then there exists a f\in \mathcal {F}=C^1(X) such that\forall x^{\prime }\in \Omega :\quad \mathbb {E}_{x\sim \mathbb {P}}\left[\frac{f(x)-f(x^{\prime })}{\Vert x-x^{\prime }\Vert }\right]=\frac{1}{2\lambda }and\forall x\in \operatorname{supp}(\mathbb {P}):\quad \mathbb {E}_{x^{\prime }\sim \mathbb {Q}}\left[\frac{f(x)...
{ "cite_spans": [] }
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.03955157473683357, 0.02070659212768078, -0.036865975707769394, -0.03985675796866417, -0.008888423442840576, -0.04483122378587723, 0.015083615668118, 0.05044656991958618, 0.03335638344287872, -0.0028400991577655077, 0.0021896802354604006, -0.023621080443263054, -0.013916295021772385, -0....
62fcfc0cf91c7422db767d389c0d312ca05f437f
subsection
29
52
Proof of Things
REF\forall x\in \operatorname{supp}(\mathbb {P}):\quad f(x)=\frac{\mathbb {E}_{x^{\prime }\sim \mathbb {Q}}[\frac{f(x^{\prime })}{\Vert x-x^{\prime }\Vert }]+\frac{1}{2\lambda }}{\mathbb {E}_{x^{\prime }\sim \mathbb {Q}}[\frac{1}{\Vert x-x^{\prime }\Vert }]}.Now it's clear that if the mapping T:\mathcal {F}\rightarrow ...
{ "cite_spans": [] }
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.029346106573939323, 0.00020124897127971053, -0.020662833005189896, -0.041691917926073074, -0.010232222266495228, -0.013925284147262573, 0.047399379312992096, 0.04358423128724098, 0.04593436419963837, 0.0071915509179234505, 0.006020301021635532, 0.0032619500998407602, 0.005016917362809181,...
a343c4ef5d50f2d163df5f31906b3d5f38fa3331
subsection
30
52
Proof of Things
REF and REF and \tau _P(\mathbb {P}\Vert \mathbb {Q};f^*)=\tau _P(\mathbb {P}\Vert \mathbb {Q}).Define the mapping S:\mathcal {F}\rightarrow \mathcal {F} byS(f)(x)=\frac{f(x)}{2\lambda \mathbb {E}_{\tilde{x}\sim \mathbb {P},x^{\prime }\sim \mathbb {Q}}\left[\frac{f(\tilde{x})-f(x^{\prime })}{\Vert \tilde{x}-x^{\prime }...
{ "cite_spans": [] }
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.01878903992474079, 0.03336542844772339, -0.002629091963171959, -0.019842203706502914, 0.01712534762918949, 0.007250035647302866, 0.030297517776489258, 0.022849060595035553, 0.049788665026426315, 0.0474381297826767, 0.010462946258485317, 0.013225591741502285, 0.008120040409266949, 0.0024...
350d57ee74c98418c0b07d8d1a0ad1ad80a717b5
subsection
31
52
Proof of Things
REF\mathbb {E}_{\tilde{x}\sim \mathbb {P},x^{\prime }\sim \mathbb {Q}}\left[\frac{T(f)(\tilde{x})-T(f)(x^{\prime })}{\Vert \tilde{x}-x^{\prime }\Vert }\right] &=\mathbb {E}_{\tilde{x}\sim \mathbb {P}}\left[\mathbb {E}_{x^{\prime }\sim \mathbb {Q}}\left[\frac{T(f)(\tilde{x})}{\Vert \tilde{x}-x^{\prime }\Vert }\right]\ri...
{ "cite_spans": [] }
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.027339264750480652, 0.03826276585459709, -0.011625289916992188, -0.02732400968670845, 0.007807404734194279, 0.01684294082224369, 0.01815498061478138, 0.04250401258468628, 0.03634047508239746, 0.05330546572804451, 0.02215212769806385, 0.02273186668753624, -0.01476045697927475, 0.01215163...
e329d83d9b64333ec4f63f5e18bde0030e04e035
subsection
32
52
Proof of Things
REF we get\mathbb {E}_{x\sim \mathbb {P},x^{\prime }\sim \mathbb {Q}}\left[\frac{f_1(x^{\prime })-f_2(x^{\prime })}{\Vert x-x^{\prime }\Vert }\right] =\mathbb {E}_{x\sim \mathbb {P},x^{\prime }\sim \mathbb {Q}}\left[\frac{f_1(x)-f_2(x)}{\Vert x-x^{\prime }\Vert }\right].Now since for every f\in \mathcal {F} it holds by...
{ "cite_spans": [] }
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.03911891207098961, 0.00954324472695589, -0.010504435747861862, -0.046015072613954544, -0.006491847801953554, 0.006480405107140541, 0.05458949878811836, 0.0389358289539814, 0.024258607998490334, 0.02277868054807186, -0.02012396603822708, 0.0006732145557180047, -0.012014877051115036, 0.00...
61adb1b2d99eb82466cab616095d4fbd13012a69
subsection
33
52
Proof of Things
If f_1,f_2\in S(\mathcal {F}) then\sup _{x\in \operatorname{supp}(\mathbb {P})} |T(f_1)(x)-T(f_2)(x)| &=\sup _{x\in \operatorname{supp}(\mathbb {P})}\left|\frac{ \mathbb {E}_{x^{\prime }\sim \mathbb {Q}}\left[\frac{f_1(x^{\prime })-f_2(x^{\prime })}{\Vert x-x^{\prime }\Vert }\right] }{ \mathbb {E}_{x^{\prime }\sim \mat...
{ "cite_spans": [ { "arxiv_id": "", "doi": "10.5555/1756006.1859901", "end": 1494, "openalex_id": "https://openalex.org/W2124331852", "raw": "Sriperumbudur, B. K., Gretton, A., Fukumizu, K., Schölkopf, B., and Lanckriet, G. R. Hilbert space embeddings and metrics on probability measu...
1802.04591
First Order Generative Adversarial Networks
[ "Calvin Seward", "Thomas Unterthiner", "Urs Bergmann", "Nikolay Jetchev", "Sepp Hochreiter" ]
[ "cs.LG", "stat.ML" ]
2,018
en
Computer Science
[ -0.028847144916653633, 0.03000713512301445, -0.03644813969731331, -0.0031956988386809826, -0.005853375419974327, -0.009699661284685135, 0.05516062304377556, 0.060411110520362854, 0.027855046093463898, 0.05186380818486214, -0.027091894298791885, -0.0008466216968372464, -0.01723197102546692, ...