Computational Statistics

, Volume 20, Issue 2, pp 265–273

Componentwise adaptation for high dimensional MCMC

  • Heikki Haario
  • Eero Saksman
  • Johanna Tamminen

DOI: 10.1007/BF02789703

Cite this article as:
Haario, H., Saksman, E. & Tamminen, J. Computational Statistics (2005) 20: 265. doi:10.1007/BF02789703


We 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.


MCMCadaptive MCMCMetropolis-Hastings algorithm

Copyright information

© Springer-Verlag 2005

Authors and Affiliations

  • Heikki Haario
    • 1
  • Eero Saksman
    • 2
  • Johanna Tamminen
    • 3
  1. 1.University of Helsinki Department of Mathematics and StatisticsUniversity of HelsinkiFinland
  2. 2.University of Jyväskylä Department of Mathematics and StatisticsUniversity of JyväskyläFinland
  3. 3.Finnish Meteorological Institute Geophysical Research DivisionHelsinkiFinland