A python frontend is available in the neci_guga python library, which
is based on the neci build target (i.e. plain neci, without complex
or multi-replica support) and is built by executing
make neci_guga_pylib
in the neci build directory. This will create a python3 library
neci_guga.<build-specifier>.so, which is installed in the
python/ subdirectory of the neci build directory.
To use it, load the python subdirectory of the neci build directory
into the library path of python, either by
export PYTHONPATH=<neci_build>/python:$PYTHONPATH
or by adding
import sys
sys.path.append('<neci_build>/python')
to the calling python module or script.
The neci_guga python module can then be loaded in calling python code
with import neci_guga and provides the following functionality:
neci_guga.init_guga(fcidump_path, S)
Takes the desired total spin S and initializes the GUGA functionality
of neci by reading in an existing FCIDUMP (fcidump_path) file.
neci_guga.clear_guga()
Clears all memory and deletes all objects initialized by init_guga,
returns 0 on success, 1 else.
neci_guga.csf_matel(D_i, D_j)
Returns the matrix element between D_i and D_j, passed as an array
of the size of the number of electrons in the DefineDet format.
neci_guga.run_neci(perm)
Reads a neci input file neci.inp and an FCIDUMP file in the current
directory, using an orbital permutation perm to re-order the orbitals
used for the calculation (with respect to the FCIDUMP file). Then, a neci
calculation with the specified input is run and the weight of the leading
CSF is returned.
The permutation is given by specifying the new position for each orbital,
i.e. a permutation
python
perm = [3, 4, 2, 1]
would put orbital 1 in the third position, orbital 2 in the fourth position,
orbital 3 in the second position and orbital 4 in the first position.