Matches in SemOpenAlex for { <https://semopenalex.org/work/W59796744> ?p ?o ?g. }
Showing items 1 to 64 of
64
with 100 items per page.
- W59796744 abstract "AbstractW e sho w ho w to adapt the Monotone Coupling from the P ast exact sampling algorithmto sample from some symmetric subsets of nite distributiv e lattices. The metho d isapplied to generate uniform random elemen ts of all symmetry classes of alternating-signmatrices. 1 In tro duction Coupling from the Past (CFTP) is an algorithm due to Propp and Wilson [6] to sam-ple from the exact stationary distribution (rather than from an appro ximation, like theclassical MCMC metho d) of an ergo dic Mark ov chain. In its arguably most practicalversion of monotone-CFTP , the Mark ov chain's state space is a partially ordered setwith unique minim um and maxim um elemen ts, and the chain is run through a monotoneup date me chanism : at eac h time step, an appropriately-distributed incr easing functionmapping the state space to itself is randomly chosen, and arbitrarily man y copies of theMark ov chain can b e coupled using the same function b y de ning the new state of eac h ofthem to b e the image of their previous state. The key elemen t in the CFTP algorithm liesin b eing able to (bac kw ard) comp ose a n um b er of suc h random up date functions un til theresulting function is found to b e a constan t. Using only incr easing functions mak es it p os-sible to detect this coalescence b y computing the successiv e images of only the minim umand maxim um elemen ts of the state space.Th us, CFTP is particularly w ell suited to sampling from distributive lattic es , eitheruniformly or according to some w ell-b eha ved distribution. In favorable cases, this mak esit p ossible to completely eliminate initiation bias from the corresp onding MCMC sim ula-tions, in time comparable to what a w ell-tuned (for small enough bias) MCMC algorithmw ould require. It is also w orth men tioning that tigh t b ounds on the mixing time of Mark ovchains are often tric ky to obtain, so that MCMC sim ulations typically either ha ve to b e runfor an empirically determined time, or for a sometimes widely overestimated guaran teedmixing time.Man y com binatorially interesting families of ob jects fall in the monotone CFTP ona distributiv e lattice framew ork, but in some cases this fails b ecause of some imp osedsymmetry condition. As an example, consider alternating-sign matric es (ASM for short;see Section 3 for de nitions). These are in bijection with another set of matrices whic hform a distributiv e lattice, and CFTP is an ecien t w ay to generate random ASMs.The group of symmetries of the unit square acts naturally on alternating-sign matrices,and for eac h of its subgroups, one can consider the class of matrices whic h are invarian tunder its action. En umerativ e form ulae for man y symmetry classes w ere conjectured b yRobbins [9], and pro ved b y Kup erb erg and others [4, 7, 8]." @default.
- W59796744 created "2016-06-24" @default.
- W59796744 creator A5018101364 @default.
- W59796744 creator A5066537086 @default.
- W59796744 date "2008-06-16" @default.
- W59796744 modified "2023-09-27" @default.
- W59796744 title "Exact Random Generation of Symmetric and Quasi-symmetric Alternating-sign Matrices" @default.
- W59796744 cites W1649780168 @default.
- W59796744 cites W1973538730 @default.
- W59796744 cites W2005999031 @default.
- W59796744 cites W2014082748 @default.
- W59796744 cites W2033900415 @default.
- W59796744 cites W2073299436 @default.
- W59796744 cites W2096665752 @default.
- W59796744 cites W2100445436 @default.
- W59796744 cites W2126209209 @default.
- W59796744 cites W3044829600 @default.
- W59796744 hasPublicationYear "2008" @default.
- W59796744 type Work @default.
- W59796744 sameAs 59796744 @default.
- W59796744 citedByCount "0" @default.
- W59796744 crossrefType "proceedings-article" @default.
- W59796744 hasAuthorship W59796744A5018101364 @default.
- W59796744 hasAuthorship W59796744A5066537086 @default.
- W59796744 hasBestOaLocation W597967441 @default.
- W59796744 hasConcept C114614502 @default.
- W59796744 hasConcept C121332964 @default.
- W59796744 hasConcept C134306372 @default.
- W59796744 hasConcept C139676723 @default.
- W59796744 hasConcept C158693339 @default.
- W59796744 hasConcept C33923547 @default.
- W59796744 hasConcept C3746008 @default.
- W59796744 hasConcept C41008148 @default.
- W59796744 hasConcept C54848796 @default.
- W59796744 hasConcept C62520636 @default.
- W59796744 hasConceptScore W59796744C114614502 @default.
- W59796744 hasConceptScore W59796744C121332964 @default.
- W59796744 hasConceptScore W59796744C134306372 @default.
- W59796744 hasConceptScore W59796744C139676723 @default.
- W59796744 hasConceptScore W59796744C158693339 @default.
- W59796744 hasConceptScore W59796744C33923547 @default.
- W59796744 hasConceptScore W59796744C3746008 @default.
- W59796744 hasConceptScore W59796744C41008148 @default.
- W59796744 hasConceptScore W59796744C54848796 @default.
- W59796744 hasConceptScore W59796744C62520636 @default.
- W59796744 hasLocation W597967441 @default.
- W59796744 hasLocation W597967442 @default.
- W59796744 hasLocation W597967443 @default.
- W59796744 hasOpenAccess W59796744 @default.
- W59796744 hasPrimaryLocation W597967441 @default.
- W59796744 hasRelatedWork W1645658348 @default.
- W59796744 hasRelatedWork W1978042415 @default.
- W59796744 hasRelatedWork W1983332406 @default.
- W59796744 hasRelatedWork W2000439998 @default.
- W59796744 hasRelatedWork W2007468163 @default.
- W59796744 hasRelatedWork W2016005585 @default.
- W59796744 hasRelatedWork W2047697295 @default.
- W59796744 hasRelatedWork W2243881134 @default.
- W59796744 hasRelatedWork W2312841100 @default.
- W59796744 hasRelatedWork W3082260596 @default.
- W59796744 isParatext "false" @default.
- W59796744 isRetracted "false" @default.
- W59796744 magId "59796744" @default.
- W59796744 workType "article" @default.