Perfect Sampling of Random Spanning Trees

نویسنده

  • JASON S. LENDERMAN
چکیده

Let G be a weighted directed graph where the edge weights are non-negative real numbers. An arborescence or (directed) spanning tree with root r is a connected subgraph of G such that there is a unique directed path from v to r for each v ∈ V (G). The weight of a spanning tree T will be the product of the weights of the edges of T . Let Υ(G) be the probability distribution on the set of all spanning trees of G where the probability of a tree is proportional to its weight, and let Υr (G) be the probability distribution of the set of all spanning trees of G with root r where the probability of a tree with root r is proportional to its weight. We are interested in efficient algorithms for generating random samples from Υ(G) and Υr (G). The algorithms we will be discussing here are based on random walks (more generally Markov chains.) We say that G is stochastic if for each vertex the sum of the weights of all outgoing arcs is equal to 1. If G is stochastic then we can of course think of it as a Markov chain. Otherwise we can define two related stochastic graphs G and G̃. G is obtained by normalizing the weights of the outgoing arcs for each vertex so that their sum is 1. G̃ is obtained by first adding a weighted loop to each vertex so that their out-degrees are all equal and then normalizing the weights of all outgoing arcs so that the out-degree of each vertex is 1. It is easy to check that Υ(G) = Υ(G̃) and Υr (G) = Υr (G) and therefore we will use the Markov chain associated with G̃ when sampling from Υ(G), and use the Markov chain associated with G when sampling from Υr (G). (Note: we also have Υr (G̃) = Υr (G) but it is preferable to use G when sampling from Υr (G) since it produces more efficient algorithms due to smaller cover and mean hitting times.)

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

0 Trees and Matchings Richard

In this article, Temperley’s bijection between spanning trees of the square grid on the one hand, and perfect matchings (also known as dimer coverings) of the square grid on the other, is extended to the setting of general planar directed (and undirected) graphs, where edges carry nonnegative weights that induce a weighting on the set of spanning trees. We show that the weighted, directed spann...

متن کامل

Trees and Matchings

In this article, Temperley’s bijection between spanning trees of a subgraph of the square grid on the one hand, and perfect matchings (also known as dimer coverings) of an associated subgraph of the square grid on the other, is extended to the setting of general planar directed (and undirected) graphs, where edges carry nonnegative weights that induce a weighting on the set of spanning trees. W...

متن کامل

from Foundations of Markov chain Monte

The aim of this course is to address the complexity of counting and sampling problems from an algorithmic perspective. Typically we will be interested in counting the size of a collection of combinatorial structures of a graph, e.g., the number of spanning trees of a graph. We will soon see that this style of counting problem is intimately related to the sampling problem which asks for a random...

متن کامل

Counting the number of spanning trees of graphs

A spanning tree of graph G is a spanning subgraph of G that is a tree. In this paper, we focus our attention on (n,m) graphs, where m = n, n + 1, n + 2, n+3 and n + 4. We also determine some coefficients of the Laplacian characteristic polynomial of fullerene graphs.

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2005