PSiesta: Siesta as a Python library

This repository contains Python bindings for the Siesta as a subroutine functionality in Siesta. Currently you can mpi-execute siesta controlled from Python. It's still based on fdf-files, but you can run a calculation, get energy, forces and stress, then alter the structure in Python, and hot-rerun the calculation. Example:

from mpi4py import MPI
import numpy as np
import sisl as si
from psiesta import FSiesta  # the fsiesta name comes from the siesta module that its built on

rank = MPI.COMM_WORLD.Get_rank()

# Read geometry on all nodes: You can also do it on rank 0 only and broadcast with mpi4py
geometry = si.get_sile("graphene.fdf").read_geometry()
my_run = FSiesta("graphene")  # Start a siesta session with given label.fdf. Uses comm_world by default
energy, forces, stress = my_run.run(locations=geometry.xyz, cell=geometry.sc.cell)
H = si.get_sile("graphene/graphene.fdf").read_hamiltonian()
print(f"Energy: {energy}. Max force: {np.max(np.linalg.norm(forces, axis=1))}")

# Mutate the geometry and rerun
geometry.xyz[0, 0] += 0.05
new_results = my_run.run(geometry.xyz)
H2 = si.get_sile("graphene/graphene.fdf").read_hamiltonian()

# Now  make a file for later comparison of results / usage in e-ph coupling calculation?
dH = H2-H
if rank == 0:
  dH.write("deltaH-no-correction.nc")  # for later calculations

my_run.quit()  # gracefully exit siesta

You could execute the above script with eg. mpirun -n 4 python3 myscript.py. Siesta will then run inside the Python processes. Relevant properties, other than those returned directly, can be read from the output files in the calculation directory in-between runs. It is recommended to use sisl for this.

So, although these bindings provide Python-integration, Siesta is still a file-based calculator. The Python control primarily makes it easier to do calculations where many custom displacements are needed.

An important limitation is that you can only create the one FSiesta object once on each mpi process. This is a limitation from fsiesta/siesta itself, further stemming from the fact that Siesta was never originally designed to be used as an 'object calculator' but rather as a monolithic program.

Building

You must have a properly set up arch.make for Siesta. In your Siesta Obj-dir, you must have at least compiled Siesta. You can then run OBJ=/my/custom/siesta/Obj/ python3 setup.py install [--user] [--prefix=<prefix>] (or use build instead of install) to build Psiesta. It makes use of Siesta's own Makefile (which includes your arch.make) in combination with --dry-run to extract the compilation and link arguments. It should work for both intel and gnu compilers, but be aware that LTO can complicate things, and ensure that any external libraries that are used (eg. flook) are compiled with -fPIC.

Behaviour

See also the SiestaSubroutine readme. In summary, the fsiesta module that this is based on copies all fdf and psf-files from your working directory <cwd> into <cwd>/<systemlabel> where <systemlabel> is the label you give. In that folder it will then start reading from <systemlabel>.fdf and putting any output files like a regular Siesta calculation, except with the structure provided from Python.