Installation
Contents
Installation¶
To install alpro, clone the repository using:
git clone https://github.com/jhmatthews/alpro.git
or download a release, then run:
python setup.py install
Depending on your system, you may need to run sudo python setup.py install or python setup.py install --user. You will need a number of standard python modules and a working installation of numba.
I have tested with Python 3.7 and later. Module requirements are relatively light, essentially matplotlib,numba,numpy,scipy. Full details can be found in the requirements.txt file that ships with the repository. A minimal running environment can be loaded by starting a clean virtual environment with, e.g.,:
python -m venv /path/to/virtual/env/
and then running:
pip install -r requirements.txt
before installing alpro.
Running ALPro¶
To check ALPro has installed, try importing it as a module:
python -c "import alpro"
A basic test of the code can be run using:
python -m unittest alpro.test
if all goes well you should see output like
Ran 4 tests in 4.447s
OK
Parallelisation¶
alpro includes a parallelisation routine. This is only imported in the __init__.py file if the mpi4py module is found, otherwise it is assumed the user does not wish to use this routine. If mpi4py was present at the time of installation, the parallel routine can be tested with
mpirun -n 4 python -m unittest alpro.parallel
and the user can check the speedup by changing the integer after -n. If this test produces an ImportError, try installing mpi4py and then reinstalling alpro from scratch. Instructions on using the parallel routines are given in Examples.