Skip to content

Latest commit

 

History

History
113 lines (83 loc) · 3.81 KB

README.md

File metadata and controls

113 lines (83 loc) · 3.81 KB
Raw

EQeq

Charge equilibration method for crystal structures.

Modified version, which allows to specify additional parameters:

  • lambda (default: 1.2) The dielectric screening parameter. corresponds to eps_eff = 1.67
  • hI0 (default: -2.0) The electron affinity of hydrogen
  • chargePrecision (default: 3) Number of digits to use for point charges
  • method (default: "ewald", alternative: "nonperiodic") Method to compute the Coulombic interaction
  • mR (default: 2) Number of "expansion" unit cells to consider in periodic calculation ("real space"). 2 => 5x5x5
  • mK (default: 2) Number of "expansion" unit cells to consider in periodic calculation ("frequency space"). 2 => 5x5x5
  • eta (default: 50) Ewald splitting parameter
  • ionizationdata (default: ionizationdata.dat) File with ionization potentials and electron affinities. Default data are
    EA: experimental, T.Andersen et al., 1999
    IP: experimental, C.E.Moore, 1970
  • chargecenters (default: chargecenters.dat) File with common oxidation states (lowered, if missing ionizationdata)

Usage

To run the HKUST-1 example:

cd examples/HKUST1
./run.sh

Summary

The source code in this program demonstrates the charge equilibration method described in the accompanying paper. The purpose of the source code provided is to be minimalistic and do "just the job" described. In practice, you may wish to add various features to the source code to fit the particular needs of your project.

Major highlights of program:

  • Obtains charges for atoms in periodic systems without iteration
  • Can use non-neutral charge centers for more accurate point charges
  • Designed for speed (but without significant code optimizations)

Features not implemented but that you may want to consider adding:

  • Spherical cut-offs (for both real-space and reciprocal-space sums)
  • An iterative loop that guesses the appropriate charge center (so the user does not have to guess)
  • Ewald parameter auto-optimization
  • Various code optimizations

Running the program:

Program expects two input files ionization.dat and chargecenters.dat. Please look at source code to see what the other optional inputs are for (should be mostly self-explanatory). Compile with something like:

g++ main.cpp -O3 -o eqeq

and run with

./eqeq my_file.cif

Python bindings

To facilitate automation and scaling, this version of EQeq can be operated via Python. To enable, you must build EQeq as a shared library:

g++ -c -fPIC main.cpp -O3 -o eqeq.o
g++ -shared -Wl,-soname,libeqeq.so -O3 -o libeqeq.so eqeq.o
sudo cp libeqeq.so /usr/lib

(for Macs, replace -soname with -install_name) (if you don't have sudo access, mkdir ~/lib; cp libeqeq.so ~/lib, then change the path at the top of eqeq.py)

From the command line, you can run eqeq.py:

python eqeq.py --help
python eqeq.py IRMOF-1.cif --output-type mol --method ewald

This includes extensive help documentation on running EQeq, and an easy interface to change individual parameters.

To use globally with Python scripts, you must put the EQeq directory on your PYTHONPATH:

export PYTHONPATH:/path/above/EQeq:$PYTHONPATH

Then, you can call EQeq from Python with

import EQeq
EQeq.run("IRMOF-1.cif")

The input takes both filenames and actual data, and outputs either files or strings. This change allows for streaming data, which enables EQeq to be used as part of a broader code pipeline.

import EQeq
# Load a file. In practice, this can come from any source, such as a database
with open("IRMOF-1.cif") as in_file:
    data = in_file.read()
charges = EQeq.run(data, output_type="list", method="ewald")