Counting without sampling. New algorithms for enumeration problems using statistical physics TeX Export

@misc{citeulike:364118, abstract = {We propose a new type of approximate counting algorithms for the problems of enumerating the number of independent sets and proper colorings in low degree graphs with large girth. Our algorithms are not based on a commonly used Markov chain technique, but rather are inspired by developments in statistical physics in connection with correlation decay properties of Gibbs measures and its implications to uniqueness of Gibbs measures on infinite trees, reconstruction problems and local weak convergence methods. On a negative side, our algorithms provide \$\epsilon\$-approximations only to the logarithms of the size of a feasible set (also known as free energy in statistical physics). But on the positive side, our approach provides deterministic as opposed to probabilistic guarantee on approximations. Moreover, for some regular graphs we obtain explicit values for the counting problem. For example, we show that every 4-regular \$n\$-node graph with large girth has approximately \$(1.494...)^n\$ independent sets, and in every \$r\$-regular graph with \$n\$ nodes and large girth the number of \$q\geq r+1\$-proper colorings is approximately \$[q(1-{1\over q})^{r\over 2}]^n\$, for large \$n\$. In statistical physics terminology, we compute explicitly the limit of the log-partition function. We extend our results to random regular graphs. Our explicit results would be hard to derive via the Markov chain method.}, archivePrefix = {arXiv}, author = {Bandyopadhyay, Antar and Gamarnik, David}, citeulike-article-id = {364118}, citeulike-linkout-0 = {http://arxiv.org/abs/math.PR/0510471}, citeulike-linkout-1 = {http://arxiv.org/pdf/math.PR/0510471}, comment = {* proposition 1, cavity equation for independent set from marginal}, day = {21}, eprint = {math.PR/0510471}, keywords = {counting, hardcore}, month = {Oct}, posted-at = {2008-02-12 00:49:25}, priority = {2}, title = {Counting without sampling. New algorithms for enumeration problems using statistical physics}, url = {http://arxiv.org/abs/math.PR/0510471}, year = {2005} }

TY - ELEC ID - citeulike:364118 L3 - citeulike-article-id:364118 N2 - We propose a new type of approximate counting algorithms for the problems of enumerating the number of independent sets and proper colorings in low degree graphs with large girth. Our algorithms are not based on a commonly used Markov chain technique, but rather are inspired by developments in statistical physics in connection with correlation decay properties of Gibbs measures and its implications to uniqueness of Gibbs measures on infinite trees, reconstruction problems and local weak convergence methods. On a negative side, our algorithms provide $\epsilon$-approximations only to the logarithms of the size of a feasible set (also known as free energy in statistical physics). But on the positive side, our approach provides deterministic as opposed to probabilistic guarantee on approximations. Moreover, for some regular graphs we obtain explicit values for the counting problem. For example, we show that every 4-regular $n$-node graph with large girth has approximately $(1.494...)^n$ independent sets, and in every $r$-regular graph with $n$ nodes and large girth the number of $q\geq r+1$-proper colorings is approximately $[q(1-{1\over q})^{r\over 2}]^n$, for large $n$. In statistical physics terminology, we compute explicitly the limit of the log-partition function. We extend our results to random regular graphs. Our explicit results would be hard to derive via the Markov chain method. TI - Counting without sampling. New algorithms for enumeration problems using statistical physics KW - counting KW - hardcore N1 - * proposition 1, cavity equation for independent set from marginal AU - Bandyopadhyay, Antar AU - Gamarnik, David PY - 2005/Oct/21/ UR - http://arxiv.org/abs/math.PR/0510471 ER -