Assume that all the requirements are fulfilled:

• Python 2.7-3.5
• NumPy 1.6.1 or later (base N-dimensional array package)
• ASE 3.15.0 or later (atomic simulation environment)
• a C-compiler
• LibXC 2.0.1 or later
• BLAS and LAPACK libraries

Optional, but highly recommended:

• SciPy 0.7 or later (library for scientific computing, requirered for some features)
• an MPI library (required for parallel calculations)
• FFTW (for increased performance)
• BLACS and ScaLAPACK

LibXC compilation:

svn co http://www.tddft.org/svn/libxc/trunk/ libxc
cd libxc
autoreconf -i
./configure --enable-shared --prefix=/home/USER/xc
make -j N
make install

The LibXC compilation might not work, and GPAW would complain, so configure as follows:

./configure CFLAGS="-O2 -fPIC" --prefix=/home/USER/xc

After compiling LibXC add these lines to your .bashrc:

export C_INCLUDE_PATH=/home/USER/xc/include
export LIBRARY_PATH=/home/USER/xc/lib
export LD_LIBRARY_PATH=/home/USER/xc/lib

Let’s install ASE using pip, because it is easy.

pip install --upgrade --user ase

Get the GPAW source code and remove in libxc.c in c/xc/ line xc_mgga_x_tb09_set_params(self->functional[0], c);. Them compile GPAW with python setup.py install --user. You might want to add the .local/bin to the path.

Use either Python or Python3, and be consistent with that.

The official guideline also recommends adding these lines to your .bashrc:

export PYTHONPATH=/home/USER/gpaw:$PYTHONPATH
export PATH=/home/USER/tools:$PATH

Don’t forget to get setups. E.g. execute gpaw install-data DIR. After that run the tests.

Despite python etc we still heavily rely on the tables. Sometimes it is need to transpose a vertical data-set to a horizontal representation. That is easy. In LibreOffice Calc use special paste (Shift+Ctrl+v) and tick transpose and numbers. One can do similar trick in Excel. That is it.