
New Upper Bounds for Trace Reconstruction
We improve the upper bound on trace reconstruction to (O(n^1/5))....
read it

A splitting method to reduce MCMC variance
We explore whether splitting and killing methods can improve the accurac...
read it

Wastefree Sequential Monte Carlo
A standard way to move particles in a SMC sampler is to apply several st...
read it

Estimating Convergence of Markov chains with LLag Couplings
Markov chain Monte Carlo (MCMC) methods generate samples that are asympt...
read it

BiasVariance Tradeoff and Overlearning in Dynamic Decision Problems
Modern Monte Carlotype approaches to dynamic decision problems face the...
read it

Item Parameter Recovery for the TwoParameter Testlet Model with Different Estimation Methods
The testlet model is a popular statistical approach widely used by resea...
read it

Maximal couplings of the MetropolisHastings algorithm
Couplings play a central role in the analysis of Markov chain Monte Carl...
read it
Making meanestimation more efficient using an MCMC trace variance approach: DynaMITE
The MarkovChain MonteCarlo (MCMC) method has been used widely in the literature for various applications, in particular estimating the expectation πΌ_Ο[f] of a function f:Ξ©β [a,b] over a distribution Ο on Ξ© (a.k.a. meanestimation), to within Ξ΅ additive error (w.h.p.). Letting R β ba, standard varianceagnostic MCMC meanestimators run the chain for Γ(TR^2/Ξ΅^2) steps, when given as input an (often loose) upperbound T on the relaxation time Ο_ rel. When an upperbound V on the stationary variance v_Οβπ_Ο[f] is known, Γ(TR/Ξ΅+TV/Ξ΅^2) steps suffice. We introduce the DYNAmic Mcmc InterTrace variance Estimation (DynaMITE) algorithm for meanestimation. We define the intertrace variance v_T for any trace length T, and show that w.h.p., DynaMITE estimates the mean within Ξ΅ additive error within Γ(TR/Ξ΅ + Ο_ rel v_Ο rel/Ξ΅^2) steps, without a priori bounds on v_Ο, the variance of f, or the trace variance v_T. When Ο΅ is small, the dominating term is Ο_ rel v_Ο rel, thus the complexity of DynaMITE principally depends on the a priori unknown Ο_ rel and v_Ο rel. We believe in many situations v_T=o(v_Ο), and we identify two cases to demonstrate it. Furthermore, it always holds that v_Ο relβ€ 2v_Ο, thus the worstcase complexity of DynaMITE is Γ(TR/Ξ΅ +Ο_ rel v_Ο/Ξ΅^2), improving the dependence of classical methods on the loose bounds T and V.
READ FULL TEXT
Comments
There are no comments yet.