GitHub - alexkaiser/mc_stretch: Fast, affine-invariant, GPU parallel MCMC.

Branches Tags

Name		Name	Last commit message	Last commit date
Latest commit History 39 Commits
.gitignore		.gitignore
Kernel_Ranluxcl_Init.cl		Kernel_Ranluxcl_Init.cl
README		README
cl-helper.c		cl-helper.c
cl-helper.h		cl-helper.h
constants.h		constants.h
data_struct.h		data_struct.h
license.txt		license.txt
makefile		makefile
noisy_data.txt		noisy_data.txt
pdf.h		pdf.h
plot_hist_from_files.m		plot_hist_from_files.m
ranluxcl.cl		ranluxcl.cl
readme_mc_stretch.pdf		readme_mc_stretch.pdf
stretch_move.cl		stretch_move.cl
stretch_move_main.c		stretch_move_main.c
stretch_move_sampler.c		stretch_move_sampler.c
stretch_move_sampler.h		stretch_move_sampler.h
stretch_move_util.c		stretch_move_util.c
stretch_move_util.h		stretch_move_util.h
timing.h		timing.h

Repository files navigation

MC Stretch: GPU parallel, affine invariant Markov chain Monte Carlo. 

Much more detail in readme_mc_stretch.pdf.

Please email adkaiser at the domain gmail.com for help. 

__________________________________________________


Follow these steps to get started quickly with the included examples. 


 - Place generic C code to evaluate the log of the probability density function (pdf) you would like to sample in ``pdf.h.'' 

 - Pick one of the included examples in "stretch_move_main.c" to run. Check that parameters in the example match your problem, which usually means setting the correct dimension.

 - Type "make" on the command line to build the code with the included makefile. 

 - Type "./stretch_move_main" to run the executable. The code will prompt to select an OpenCL implementation and hardware, then run the sampler.