Search arXivSearch

arXiv · 1505.02948

Expanders from Markov bases

Abstract

Diaconis and Sturmfels introduced an influential method to construct Markov chains using commutative algebra. One major point of their method is that infinite families of graphs are simultaneously proved to be connected by a single algebraic calculation. For large state spaces in the infinite families these Markov chains are not rapidly mixing and only ad hoc methods have been available to improve their mixing times. We provide a method to get rapid mixing by constructing expanders for the Diaconis-Sturmfels type Markov chains.

Explore related subjects

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Alexander Engstrom. 2015-05-12. Expanders from Markov bases. https://arxiv.org/abs/1505.02948

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related papers

Rooted Spider Embeddings and the Erd\H os-Sós Conjecture

Under a local density condition, we prove that every $k$-edge spider embeds at any prescribed center of degree at least $k$, unless all legs are even and the host graph has one of two specified structures. These structures contain complete bipartite subgraphs with prescribed neighborhoods. The proof uses path rerouting and three exchange lemmas that describe equality in neighborhood estimates. As a consequence, we recover the Erd\H os-Sós bound for all spiders.

math.CO

Generalized Goulden-Yong duals and signed minimal factorizations

In this paper, we give two combinatorial ways to study signed exceptional sequences. First, we show the equivalence between one-way reflections and relatively projective representations. Secondly, we construct generalized Goulden-Yong duals using reverse Garside element actions and folded chord diagrams. We then give two applications of the generalized Goulden-Yong duals: constructing generalized Prüfer codes and counting signed factorizations using the matrix-tree theorem.

math.CO

Explicit expressions for iterates of power series

We present several formulas for both the discrete and fractional iterates of an invertible power series $f$, using a new unifying approach based on umbral calculus. Known formulas are extended, and their proofs simplified, while new expressions are introduced. In particular, by employing $q$-calculus identities, we eliminate the requirement for $f'(0)$ to equal $1$ and the resulting general expressions for the iterative logarithm are obtained as well.

math.CO