论文信息 - Componentwise adaptation for high dimensional MCMC

Componentwise adaptation for high dimensional MCMC

SummaryWe introduce a new adaptive MCMC algorithm, based on the traditional single component Metropolis-Hastings algorithm and on our earlier adaptive Metropolis algorithm (AM). In the new algorithm the adaption is performed component by component. The chain is no more Markovian, but it remains ergodic. The algorithm is demonstrated to work well in varying test cases up to 1000 dimensions.

[1] W. K. Hastings,et al. Monte Carlo Sampling Methods Using Markov Chains and Their Applications , 1970 .

[2] Heikki Haario,et al. Adaptive proposal distribution for random walk Metropolis algorithm , 1999, Comput. Stat..

[3] N. Metropolis,et al. Equation of State Calculations by Fast Computing Machines , 1953, Resonance.

[4] J. Rosenthal,et al. On adaptive Markov chain Monte Carlo algorithms , 2005 .

[5] C. Robert,et al. Controlled MCMC for Optimal Sampling , 2001 .

[6] H. Haario,et al. An adaptive Metropolis algorithm , 2001 .

[7] Sylvia Richardson,et al. Markov Chain Monte Carlo in Practice , 1997 .

[8] G. Roberts,et al. Adaptive Markov Chain Monte Carlo through Regeneration , 1998 .

[9] Anatoly Zhigljavsky,et al. Self-regenerative Markov chain Monte Carlo with adaptation , 2003 .