Installation

There are currently two ways to install this software library: by compiling the source code directly on your computer or by using Docker containers.

Source Code Compilation

Install dependencies [1]

  1. CMake *

  2. FFTW3 *

  3. GSL *

  4. OpenMP *

  5. Eigen3 *

  6. adol-c - Not as common, so probably will need to install from source. Use the appropriate openmp flag if that’s desired

  7. HDF5 - Not as common. Might be available on certain systems, but will need to compile with gcc (not clang). This is a headache, but things won’t compile properly if HDF5 is not compiled with the same compiler as GWAT

  8. BayesShip - Another package from me. Download from git and follow install instructions

* means the library is very common and is probably available through system package managers (apt/yum/etc) or through HPC infrastructure

Install library

  1. Download the source code from github.

  2. Make a directory called ``build’’ in the root directory of gw_analysis_tools.

  3. Move into ``build/’’ and run:

cmake ..

  1. If you need to modify compile settings (like turning on debugger flags or changing the install prefix), run:

ccmake ..

Save it and rerun

cmake ..

  1. Finally, run

make 
make install

Docker

Several public images with this software already installed are maintained on DockerHub:

  1. scottperkins/gwat

    • An image with the “master” branch of GWAT installed. This should be the default option for new users

    • For an example minimally working example using this image, see here

  2. scottperkins/parityviolation

    • An image with the parityViolation branch of GWAT installed

  3. scottperkins/gwat.jupyter

    • An image with mcmc_dev branch of GWAT installed and automatically launches a jupyter server on port 8989.

    • Useful for doing post-MCMC analysis, as there are python wrappers that unpack the data installed on this image

    • For usage notes on this, see jupyter notes