This repository gathers the NNAIMQ code along with some instructions and readme files.
NNAIMQ is a Python interfaced model designed to predict QTAIM charges of C, H, O and N atoms with high accuracy within simple gas-phase organic and biological structures. Altogether, NNAIMQ comprises a total of 4 Artificial Neural Networks (ANN) fitted to quantum chemical data.
-The ./code/ folder gathers all the required files to execute the NNAIMQ code.
-The ./examples/ folder contains an input example for the execution of NNAIMQ.
Requirements:
The execution of NNAIMQ requires a Linux box and the following libraries and modules:
- Python version 3.7.3 or newer.
- keras, matplotlib, numpy, pandas, seaborn and tensorflow.
- x86-64 or ARM-based (M1) processors.
It should be noticed that some of the aforementioned modules or libraries are not used explicitly but they may be suitable to run some test on the performance of this first version of the NNAIMQ code.
Execution of NNAIMQ:
The execution of NNAIMQ is straightforward:
- NNAIMQ must be executed in the same directory where all the code files are contained.
1.5) Be sure that the SSFC file that your computer is executing is the one that corresponds with the architecture processor.
- An input file with any desired name (i.e input) is required. This file contains a list of the geometries whose atomic charge is to be computed. As an example, a simple input file could be:
methane.xyz
water.xyz
toluene.xyz
- The files containing the structures (i.e methane.xyz) must be in the execution folder. The geometries of the structures to be computed are given in standard XYZ Cartesian coordinates, for instance:
5
methane
C -4.36105 1.86636 0.00000
H -3.29105 1.86636 0.00000
H -4.71772 1.07300 -0.62311
H -4.71772 2.80267 -0.37552
H -4.71772 1.72342 0.99863
-
NNAIMQ is executed from the command line as:
python NNAIMQ_v1.py input -
If executed correctly, NNAIMQ generates different output files:
-name.charge: file containing the QTAIM charges.
-name.acsf: file containing the featurization vector. -
After the execution, the .acsf files are not required anymore and can be removed from the current directory.
If you have any technical questions or need support, contact us:
[email protected] [email protected]
[email protected]
[email protected]
Shield:
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
