SAPHRON (Statistical Applied PHysics through Randomized On-the-fly Numerics) is a modern, open-source, light-weight C++11 based Monte Carlo physics package. It is designed to be easy to use, extendable and extremely versatile. As the name suggests, one of the primary features is the inclusion of various Density-of-States (DOS) sampling algorithms. SAPHRON is currently in early development, and as a result it currently lacks many basic features; this README is also likely to be out of date. In addition to the C++ application, a number of helper MATLAB scripts are included to assist with data analysis.
The documentation can be found here
- A very simple API that is user and developer friendly
- Lattice and off-lattice simulations.
- Anisotropic potentials including Gay-Berne.
- Arbitrary ensembles defined by collections of moves.
- OpenMP parallelization for energy and neighbor list calculation.
- MPI large scale multi-walker density of states (DOS) flat-histogram sampling along many order parameters.
- Much more!
Check out the documentation for a more detailed list of features
SAPHRON requires some linear alegra routines to get going. It's easy enough to install the required packages. On any Debian-based distro simply type:
$ sudo apt-get install libblas3 liblapack3 libarmadillo-dev
BLAS and LAPACK are the default linear algebra providers, though SAPHRON supports the Intel MKL and searches for it. If found it will use it instead of LAPACK, so you don't need to do anything. Armadillo provides C++ headers for interfacing with the libraries.
To get SAPHRON to compile it is crucial that you perform the following step
By default, Armadillo assumes you want to link against its shared library. SAPHRON does not do this, but instead links directly to BLAS and LAPACK or MKL as mentioned above. To fix this, use your editor of choice (here we use vim) to modify the Armadillo config header
$ vim /usr/include/armadillo_bits/config.hpp
Simply comment out the line that says
#define ARMA_USE_WRAPPER and you're good to go!
SAPHRON uses OpenMP to parallelize energy and neighbor list calculations. It relies on compilers that support the
-fopenmp flag. Any recent C++11 compatible compiler should support this, including GCC 4.9.x at minimum (required for regex support) and Clang 3.4+.
Boost and MPI
SAPHRON uses Boost MPI to provide support for multi-walker density of states simulations. The requirement is for Boost >= 1.55 with the MPI and serialization modules. A requisite underlying MPI library is also required. On recent Debian based systems using OpenMPI, the requirement can be installed via:
$ sudo apt-get install libopenmpi-dev openmpi-bin libboost-all-dev
To run a multi-walker simulation, the number of threads per processes is specified by the OpenMP
environment variable, and the number of walkers (processes) by
mpirun. It is necessary to map by nodes to ensure proper
distribution of processors to each walker. See the documentation for more information.
If you wish to link SAPHRON against the static BLAS/LAPACK libraries you can install those instead. However, you will also need to install
libquadmath. SAPHRON will look for and link against them as well.
The first step is to clone the repository locally.
$ git clone https://github.com/hsidky/SAPHRON.git
SAPHRON uses a CMake build system. To build on a UNIX-style platform enter the following commands
$ cd SAPHRON $ mkdir build && cd build $ cmake .. $ make $ sudo make install
This will install a
saphron binary into
/usr/local/bin and a static library in
SAPHRON uses the Google C++ Testing Framework for unit testing. It is downloaded as part of the build process. To run the unit tests after building, type
$ make test
SAPHRON utilizes the following packages:
Sitmo Random Number Generator (http://www.sitmo.com/article/multi-threaded-random-number-generation-in-c11/)
Feel free to fork this project on GitHub. Any pull-requests, feature requests or other form of contributions are welcome.