Skip to content

Latest commit

 

History

History
135 lines (98 loc) · 4.51 KB

INSTALL.md

File metadata and controls

135 lines (98 loc) · 4.51 KB
Raw

am installation guide

Mark A. Caprio Department of Physics, University of Notre Dame Patrick J. Fasano Physics Division, Argonne National Laboratory

  • 09/07/2020 (mac): Created.
  • 05/31/2023 (mac): Expand and update instructions.
  • 08/17/2023 (pjf): Switch to CMake instructions.
  • 09/30/2023 (mac): Provide explanation of C++ vs. Python installation.
  • 06/18/2024 (mac/keo): Add cloning directions and triage user between Python and C++ installation.

This package serves double duty as a Python module and a C++ library. Depending on your intended use of the am package, you might need to follow the installation instructions in either Section 1 or Section 2 below, or both.

(1) If you will be using am as a python module, see Section 1. This includes the case where you need am as a supporting module for some other Python package, such as mcscript-ncci or mfdnres.

(2) If you wish to build and/or install am as a C++ library, see Section 2. However, as a C++ library, am is often simply used as a dependency of a larger project (such as shell). In that case, you may not need to install it separately at all. The only time you are likely to need to build and/or install am separately is if (a) you are doing development work on am and want to be able to compile it on its own or (b) you want to install am as a standalone library so that you can link to it from other projects without including it as a submodule.

0. Retrieving and installing source

Change to the directory where you want the repository to be installed, e.g.,

% cd ~/code

Clone the am repository.

% git clone https://github.com/nd-nuclear-theory/am.git

1. Installation as python module

For use as a Python module, install using distutils/pip:

% python3 -m pip install --user .

However, this installation requires compiling C++ code. If you are on an HPC cluster, make sure you have loaded the appropriate modules for whatever compiler you wish to use.

Moreover, the code requires the GSL library. If GSL is not installed systemwide (e.g., in /usr/local), compilation will fail, with an error such as fatal error: gsl/gsl_sf_coupling.h: No such file or directory. If you are on an HPC cluster, you may need to load an appropriate module). Then you will need to set CFLAGS and/or LDFLAGS so that distutils/pip can find GSL, wherever it happens to be installed, e.g.:

% env CFLAGS="-I$GSL_DIR/include -L$GSL_DIR/lib" pip install -v --user .

If installation was successful, you should then be able to import the am module, from within your Python interpreter, without error:

>>> import am

You should also be able to run the test code without generating any errors:

% python3 am_test.py

2. Standalone compilation and installation as C++ library

First, some supporting modules (gsl and fmt) must be installed or otherwise be made visible to CMake. For fmt:

% git clone https://github.com/fmtlib/fmt.git
% cd fmt
% cmake -B build/ .
% cmake --build build/
% cd ..

Then, back in the am code directory, configure the project using CMake using:

% cmake -B build/ .

However, if you have not installed the dependencies, but only built them, you will need to tell CMake where to find the built-but-not-installed libraries:

% cmake -B build . \
    -Dfmt_DIR=~/code/fmt/build

To compile the library itself:

% cmake --build build/

To compile the test codes:

% cmake --build build/ -- tests

These test codes can then be run:

% ./build/am_test
% ./build/halfint_test

To install the library (here with prefix ~/install):

% cmake --install build/ --prefix ~/install