## Archive for July, 2010

## Vancouver skyline [1]

Posted in pictures, Travel with tags JSM 2010, Vancouver on July 31, 2010 by xi'an## Dragontail

Posted in Mountains, Travel, University life with tags Garmond, Greenland, Mammut, Squamish, The Economist, University of Victoria, Yosemite on July 31, 2010 by xi'an**B**efore leaving for Vancouver, I just bough this pair of Garmont dragontail approach shoes. They will hopefully prove useful in the approach walk to the Smoke Bluffs in Squamish today, where I plan to go climbing with Julien (once again with a brand new 10mm Mammut rope), weather and shape (15 hours of travelling) permitting… And later on the trails in Yosemite (no climbing there, I am afraid!).

**A**ctually, the flight to Vancouver was packed, which means I could not even open my macbook… So I looked instead at the translation of ** Introducing Monte Carlo Methods with R **into French, read the Economist from start till end (with an interesting discussion on the impact of fast-speed trains on freight transportation!), slept, watched the incredible landscape of Greenland with its immense icefields and deep red striated rock formations, and started an Icelandic crime novel. Plus talked with my neighbour of arithmetic Monte Carlo as I saw him proving a theorem about commutative algebra: he happened to be a professor in computer science at the University of Victoria. The Bose earphones did marvel to cancel the engine noise in the plane, if not the screams of the kids in the next row…

## IMIS & AMIS

Posted in R, Statistics, University life with tags adaptive importance sampling, adaptivity, AMIS, defensive mxture, IMIS, incremental mixture, mixture estimation, population Monte Carlo, SIR on July 30, 2010 by xi'an**A** most interesting paper by Adrian Raftery and Le Bao appeared in the Early View section of * Biometrics*. It aims at better predictions for HIV prevalence—in the original UNAIDS implementation, a naïve SIR procedure was used, based on the prior as importance function, which sometimes resulted in terrible degeneracy—, but its methodological input is about

*incremental mixture importance sampling*(IMIS), thus relates to the general topic of adaptive Monte Carlo methods I am interested in. (And to some extent to our recent AMIS paper.) Actually, a less elaborate (and less related) version of the IMIS algorithm first appeared in a 2006 paper by Steele, Raftery and Edmond in

**in the setting of finite mixture likelihoods and I somehow managed to miss it…**

*JCGS***R**aftery and Bao propose to replace SIR with an iterative importance sampling technique developed in 2003 by Steele et al. that has some similarities with population Monte Carlo (PMC). (A negligible misrepresentation of PMC in the current paper is that our method does *not* use “the prior as importance function’”.) In its current format, the IMIS algorithm starts from a first guess (e.g., the prior distribution) and builds a sequence of Gaussian (or Gaussian mixture) approximations whose parameters are estimated from the current population, while all simulation are *merged together* at each step, using a mixture stabilising weight

where the weights depend on the number of simulations at step *r*. This pattern also appears in our *adaptive multiple importance sampling* (AMIS) algorithm developed in this arXiv paper with Jean-Marie Cornuet, Jean-Michel Marin and Antonietta Mira, and in the original paper by Owen and Zhou (2000, JASA) that inspired us. Raftery and Bo extend the methodology to an IMIS with optimisation at the initial stage, while AMIS incorporates the natural population Monte Carlo stepwise optimisation developed in Douc et al. (2008, *Annals of Statistics*) that brings the proposal kernel closer to the target after each iteration. The application of the simulations to conduct model choice found in the current paper and in Steele et al. can also be paralleled with the one using population Monte Carlo we conducted for cosmological data in MNRAS.

**I**nterestingly, Raftery and Bo (and also Steele et al.) refer to the defensive mixture paper of Hesterberg (1995, Technometrics), which has been very influential in my research on importance sampling, and (less directly) to Owen and Zhou (2000, JASA), who did propose the deterministic mixture scheme that inspired AMIS. Besides the foundational papers of Oh and Berger (1991, JASA) and West (1993, *J. Royal Statistical Society Series B*), they also mention a paper by Raghavan and Cox (1998, *J. Statistical Simulation & Computation*) I was not aware of, which introduces as well a mixture of importance proposals as a variance stabilising technique.

## Le Monde puzzle [29]

Posted in Books with tags block matrix, Le Monde, mathematical puzzle on July 29, 2010 by xi'an**A**lthough I am not receiving ** Le Monde** magazine by mail any longer (maybe a consequence of highly critical posts?!), I bought it at a newsagent last Saturday and found a fairly easy mathematical puzzle. Or is it that easy? The question is to decide about the unicity of the solution to the equation

when the matrix is *(p,p)*. When *p=9,10,11*, the matrix is invertible, so there is no problem. But, looking at *p=12*, it is no longer invertible! When running a quick check like

for (p in 3:30){ A=diag(p) A[(1:p)+p*(((1:p)+1)%%(p))]=1 print(p) b=try(solve(A))}

we actually spot that *p=4,8,12,16,20,24,28,…*, i.e. all multiples of 4, lead to a non-invertible matrix. It is easy to see why for the *(4,4)* matrix

since the first and second lines are repeated. The *(8,8)* matrix is equivalent to the bloc matrix

hence with the same determinant as … In the general case, the determinant of the matrix is a circular determinant. Because most entries (but two) are zero, the determinant can easily be written as

where *ι* is the basic complex number. Therefore, if, and only if, *p* is a multiple of 4, *j* in the above product can take the value *n/4*, leading to a zero in the product… I can see why this makes a good puzzle, because this case of the multiples of 4 is quite counterintuitive!

## A summer of books

Posted in Books, R, Statistics, University life with tags Bayesian Core, Ben Nevis, Edinburgh, ICMS, Introducing Monte Carlo Methods with R, mixtures, Scotland, translation on July 28, 2010 by xi'an**T**he summer started with a research in pair session in CiRM on the R edition of * Bayesian Core*, but I am also involved two other book projects. The first one was mentioned in a previous post, namely the translation of

**into French. I have now recovered all translated chapters, involving not less than six translators! (My son completed his two last chapters while in CiRM, another benefit of the fortnight there!) So I need to get over those chapters to ensure some minimal homogeneity in the style and the notations. Not an immense amount of work given the near perfect productions of Robin Ryder, Julyan Arbel and Pierre Jacob, but still needs to be done (a perfect opportunity for the long flight to Vancouver!) The second book is an edited volume following the exciting meeting on mixtures last March in Edinburgh. I have so far received twelve of the fifteen chapters from the contributors and hope against all odds to pack the volume for Vancouver, in order to discuss with the Wiley representative there. (I also hope it will be possible to include a picture I took during my trip to Ben Nevis as the cover picture..!)**

*Introducing Monte Carlo Methods with R*