Particle MCMC discussion

The Read Paper by Christophe Andrieu, Arnaud Doucet and Roman Holenstein, Particle Markov chain Monte Carlo, was quite successful in that the Royal Statistical Society room was packed and that about ten discussants (besides the proponant and the seconder) stood up to comment on the paper.  Although I had gone through the paper many times, the explanation by Christophe and Arnaud of why the algorithm is valid MCMC was definitely helpful. Being the last one to comment meant that some of my points were moot, having been already made by previous discussants. The main issues in the discussion were the robustness of the method against degeneracy, a natural worry when dealing with particle filters, the unbiasedness of the marginal estimate, in connection with the recent Andrieu and Roberts, ways to handle the “waste” of the SMC generations and to resample.  Here are the slides of my discussion (joint with Nicolas Chopin and Pierre Jacob), that I managed to present despite having forgotten my Mac charger at home (kudos to Darren Wilkinson for remembering walking by a Mac store on his way to the meeting and directing me to it!)

Again, anyone interested in commenting on the paper can send a 400 word discussion in pdf to the journal within two weeks.

8 Responses to “Particle MCMC discussion”

  1. […] algorithm), and MC² type of algorithms, a recent extension of which being the PMCMC algorithm (Andrieu et al., 2010). Then, I will present two recent pieces of work in these direction. First, and more briefly […]

  2. […] with possibly different but related invariant distributions” (p.123). I think the recent (JRSS B) Read Paper by Andrieu, Doucet and Hollenstein would have fitted the goal quite naturally, although it may have come to the knowledge of the […]

  3. […] like real number representation on the convergence but I cannot find it. Of course, the recent particle MCMC JRSS B discussion paper by Christophe,  Arnaud Doucet and Roman Hollenstein is a way to bypass the […]

  4. […] completed a massive work on SMC², a sequential Monte Carlo method that builds on the particle MCMC discussion paper of Andrieu, Doucet and Hollenstein. The major advance is to incorporate sequential requirements within the pMCMC scheme for processing […]

  5. […] MCMC in action In connection with our discussion of the Read Paper by Christophe Andrieu, Arnaud Doucet and Roman Holenstein, Particle Markov chain […]

  6. Are the comments available online somewhere for authors willing to contribute within 2 weeks ? So as to avoid the moot factor.

    • This would be a great idea, but there is no depository of available comments. In a near future, I presume comments will be part of an electronic journal where submissions will be available as soon as submitted (since there is essentially no screening). As it happens, two discussants had similar if independent slides on Wednesday on the expected number of ancestors…

  7. […] to Alan Turing In the Eurostar back from London, I read a paper in Le Monde about Gordon Brown’s posthumous apology to Alan Turing for the […]

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: