Corpus ID: 18230857

Fast and Slim Lifted Markov Chains

  title={Fast and Slim Lifted Markov Chains},
  author={Kyomin Jung and D. Shah and Jinwoo Shin},
Metropolis-Hasting method allows for designing a reversible Markov chain P on a given graph G for a target stationary distribution π. Such a Markov chain may suffer from its slow mixing time due to reversibility. Diaconis, Holmes and Neal (1997) for the ring-like chain P , and later Chen, Lovasz and Pak (2002) for an arbitrary chain P provided an explicit construction of a non-reversible Markov chain via lifting P so that its mixing time is essentially 1/Φ(P ), whereΦ(P ) is the conductance of… Expand
3 Citations

Figures and Tables from this paper

Liftings of Tree-Structured Markov Chains - (Extended Abstract)
The lifted version of the chain gives a significant speedup over the original version whenever the error in the leaf counting estimates is o(1), and it is conjecture that it may be applicable to other Markov chains used in sampling algorithms. Expand
Adding a Single State Memory Optimally Accelerates Symmetric Linear Maps
  • A. Sarlette
  • Mathematics, Computer Science
  • IEEE Transactions on Automatic Control
  • 2016
It is proved that adding one memory slot per agent allows faster convergence towards average consensus, and that this situation cannot be improved by adding more memory slots, when the knowledge about the self-adjoint linear map to be accelerated reduces to bounds on its extreme eigenvalues. Expand
Adding a single memory per agent gives the fastest average consensus
It is proved that adding one memory slot per agent allows faster convergence, and that this situation cannot be improved by adding more memory slots. Expand


Lifting Markov chains to speed up mixing
The best mixing time achievable through lifting in Markov chain reduction is characterized in terms of multicommodity flows, and it is shown that the reduction to square root is best possible. Expand
We analyze the convergence to stationarity of a simple nonreversible Markov chain that serves as a model for several nonreversible Markov chain sampling methods that are used in practice. OurExpand
Fastest Mixing Markov Chain on a Graph
The Lagrange dual of the fastest mixing Markov chain problem is derived, which gives a sophisticated method for obtaining (arbitrarily good) bounds on the optimal mixing rate, as well as the optimality conditions. Expand
Fastest Mixing Markov Chain on a Path
This note proves that fastest mixing is obtained when each edge has a transition probability of 1/2, and considers symmetric transition probabilities, meaning those that satisfy Pi j = Pji, which is a symmetric, stochastic, tridiagonal matrix. Expand
Biased random walks, Lyapunov functions, and stochastic analysis of best fit bin packing
We study the average case performance of the Best Fit algorithm for on-line bin packing under the distribution U{j, k}, in which the item sizes are uniformly distributed in the discrete range { l/lc,Expand
Improved bounds for the unsplittable flow problem
A general method for transforming arbitrary multicommodity flow solutions into solutions that use short paths only is developed, generalizing a theorem of Leighton and Rao. Expand
A decentralized algorithm for spectral analysis
This work presents a simple decentralized algorithm for computing the top k eigenvectors of a symmetric weighted adjacency matrix, and a proof that it converges essentially in O(@t"m"i"xlog^2n) rounds of communication and computation, where @t" m" i"x is the mixing time of a random walk on the network. Expand
Bounded geometries, fractals, and low-distortion embeddings
This work considers both general doubling metrics, as well as more restricted families such as those arising from trees, from graphs excluding a fixed minor, and from snowflaked metrics, which contains many families of metrics that occur in applied settings. Expand
Multicommodity max-flow min-cut theorems and their use in designing approximation algorithms
This paper establishes max-flow min-cut theorems for several important classes of multicommodity flow problems and uses the result to design the first polynomial-time (polylog n-times-optimal) approximation algorithms for well-known NP-hard optimization problems. Expand
Gossip algorithms: design, analysis and applications
This work analyzes the averaging problem under the gossip constraint for arbitrary network, and finds that the averaging time of a gossip algorithm depends on the second largest eigenvalue of a doubly stochastic matrix characterizing the algorithm. Expand