Category: MCMC

• Gibbs sampler in Matlab using mexme

  Darren Wilkinson has a nice post up comparing different programming languages (C, Java, scala, Python, and R) for writing Gibbs samplers. Unsurprisingly, C is fastest, although it is certainly not the easiest language to program in. In particular, I/O is a bitch. Others have suggested an interesting solution: write the core of the Gibbs sampler…

  July 16, 2011

• Gibbs sampling made easy – JAGS, RKWard, CODA

  I’ve used OpenBUGS for Gibbs sampling in the past, and while it’s a powerful piece of software, I hate OpenBUGS’ GUI, especially since it runs only on Windows. JAGS is an alternative Gibbs sampling program that uses an OpenBUGS-like syntax and runs on the command line. You can call JAGS in R through the rjags…

  July 13, 2011

• Programming for multi-core environments

  CPUs with multiple cores are currently the norm. Getting optimal performance out of these systems is challenging. I recently read Parallel Programming in C with MPI and OpenMP by Michael Quinn, a book that, while released in 2004, remains relevant and actual. Dr. Quinn introduces two technologies which are available in C (and in Fortran…

  June 28, 2011

• Matlab code for 2D Gaussian surface fitting

  Updated 10/21/2011 I have some code on Matlab Central to automatically fit a 1D Gaussian to a curve and a 2D Gaussian or Gabor to a surface. The 2D Gaussian code can optionally  fit a tilted Gaussian. In its basic form curve/surface fitting is straightforward (a call to lsqcurvefit will do the trick), but the…

  May 19, 2011

• Hierarchical Bayes and WinBUGS for tracking receptive fields

  I’ve been reading Bayesian Modeling Using WinBUGS: An introduction. It’s a really nice intro to fitting complex Bayesian models with WinBUGS (aka OpenBUGS), which is a program that can sample from the posterior distribution of a Bayesian model’s parameters using MCMC methods (the GS in BUGS stands for Gibbs sampling). MCMC methods are very general,…

  April 18, 2011