Description

imcmc is a C++ implemented MCMC library
Version: 0.1.1
Author: Youhua Xu
E-mail: yhxu@nao.cas.cn (Suggestions are welcome!)

This library aims to provide an easy to use interfece to MCMC simulations. Currently only the Affine-Invariant Ensemble Sampler algorithm has been implemented, in near future I will add other commonly used algorithms / samplers into this library, for example the classical Metropolis-Hastings and Hamiltonian Monte Carlo. Besides, I would also like to add an interface to MultiNest.

Dependence

1) OpenMPI
2) GSL (use GSL's random number generator only. Since C++11, the C++ standard library provides a random-number library, so the GSL dependence will be removed later.)

Updates

@ Oct-31-2015: Bug-fixed inside update_a_walker(***)
@ Oct-28-2015: 1) Added likelihood error handling option, i.e., whether stop when an error happends inside likelihood function; and now the sampler supports writing derived parameters into chain files, by design, these derived parameters are to be calculated inside the likelihood functions, so that's up to the users. 'gaussian.cpp' in examples shows how to add derived parameters.
@ Oct-19-2015: 1) Some minor improvements
@ Oct-18-2015: 1) Updated the output format, the same as CosmoMC . 2) keyword 'chain_size' is replaced by sample_step.

Reference

1) Ensemble samplers with affine invariance
2) emcee: The MCMC Hammer