Getting started¶
Requirements¶
Tipsi requires python3, some dependencies and a fortran compiler. The easiest way to get python3 is to download a version of anaconda at:
- Packages required by TiPSi:
- numpy
- scipy
- Optional:
- h5py (for binary file input/output)
- matplotlib (for plotting)
You can get a fortran compiler by running:
sudo apt-get install gfortran
or by manually downloading the gnu fortran compiler from
Alternatively, tipsi also works with the intel fortran compiler.
Installing with setup.py¶
With git, you can get the tipsi files by using:
git clone https://gitlab.science.ru.nl/tcm/tipsi
Alternatively, the tipsi files can be downloaded manually from:
Now, if you are using the gfortran compiler, you can proceed to running setup.py. If you are using the intel compiler, you need to rename setup.cfg.intel to setup.cfg. If needed, you can link additional libraries in the setup.cfg file.
Then, go to the main directory and run:
python3 setup.py install
This will install tipsi in your default python package location.
To install it in a different location, if you don’t have sudo rights, use:
python3 setup.py install --home=<dir>
This directory needs to be added manually to your python path.
Manual installation¶
Alternatively, one can compile the files in the fortran subdirectory by hand.
For gfortran:
f2py --fcompiler=gnu95 --f90flags='-march=native -O3 -fopenmp' -lgomp -c tbpm_module.f90 tbpm_f2py.f90 -m tbpm_f2py
For intel:
f2py --fcompiler=intelem --f90flags="-qopenmp -O3 -march=native -heap-arrays" -liomp5 -lifcoremt -c tbpm_module.f90 tbpm_f2py.f90 -m tbpm_f2py
For some versions of python3/numpy you need to use the command f2py3
instead of f2py
.
A selection of compilation scripts can be found in the tipsi/fortran/
folder.