arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 0908.2088math.PR
收藏

密度相关马尔可夫链的一个精确逼近

Sharp approximation for density dependent Markov chains

Kamil Szczegot

摘要:

考虑一个R^d中的马尔可夫链Z^n序列(索引为n),其特征在于转移核近似(在n中)仅依赖于重新标度的状态n^{-1} Z^n。在光滑条件下,这样的族可以在短时间间隔上紧密耦合到具有二次漂移的布朗运动。这种构造用于确定重标度链退出凸多胞体的概率的渐近(以n为单位)展开中的前两项。常数项和大小为n^{-1/6}的第一次修正允许通过对相关微分方程和绝对常数的解进行尖锐表征。对于任意b < 1/4,误差小于O(n^{-b}).这些结果直接应用于随机算法在相变的分析。特别地,研究了大型随机超图的“剥离”算法,或等价于低密度奇偶校验码在二进删除信道上的迭代译码方案,以确定不规则超图系综的有限尺寸缩放行为.

英文摘要:

Consider a sequence (indexed by n) of Markov chains Z^n in R^d characterized by transition kernels that approximately (in n) depend only on the rescaled state n^{-1} Z^n. Subject to a smoothness condition, such a family can be closely coupled on short time intervals to a Brownian motion with quadratic drift. This construction is used to determine the first two terms in the asymptotic (in n) expansion of the probability that the rescaled chain exits a convex polytope. The constant term and the first correction of size n^{-1/6} admit sharp characterization by solutions to associated differential equations and an absolute constant. The error is smaller than O(n^{-b}) for any b < 1/4. These results are directly applied to the analysis of randomized algorithms at phase transitions. In particular, the `peeling' algorithm in large random hypergraphs, or equivalently the iterative decoding scheme for low-density parity-check codes over the binary erasure channel is studied to determine the finite size scaling behavior for irregular hypergraph ensembles.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑