C library implementing Geometric Parallel Population MCMC.
See http://www.hector.ac.uk/cse/distributedcse/reports/UniMCMC/.
Dependencies:
Required:
BLAS, CBLAS, LAPACK and LAPACKE (available from netlib.org)
MPI (either openMPI (http://www.openmpi.org/ ) or MPICH (http://www.mpich.org/ ))
sundials-2.5.0 or greater (http://computation.llnl.gov/casc/sundials/main.html)
C compiler supporting C99 and openmp
GNU make 3.81 or greater
Either:
MATLAB (to read and write data stored in MATLAB files)
HDF5 (to read and write data stored in HDF5 files (http://www.hdfgroup.org/ ))
Optional:
CUnit (http://cunit.sourceforge.net/. Only needed to compile tests.)
To compile:
Copy (or symbolically link) make.inc.example to make.inc and edit the paths within to point to where the
libraries are installed on your system. Type 'make' to compile 'libgmcmc.so' which contains the parallel
geometric population MCMC algorithm and associated models and prior distributions. Type 'make hdf5' to
compile a library defining functions to read/write data stored in HDF5 files. Type 'make matlab' to compile
a library defining functions to read/write data stored in MATLAB files. Type 'make examples' to compile the
example programs and'make test' to compile the unit tests. To run any of the example programs you will need
to update your library path. To remove the compiled object files, libraries and executables type 'make clean'.
Directory structure:
data/ - datasets for example programs
examples/ - source files for example programs
gmcmc/ - header files
pbs/ - Portable Batch System scripts for running the examples on Imperial's cluster
sge/ - Sun Grid Engine scripts for running the examples on UCL's stats cluster
src/ - source files for the libraries
test/ - source files for the CUnit tests