tad-mctc 0.0.2

Torch Autodiff Utility

This library is a collection of utility functions that are used in PyTorch (re-)implementations of projects from the Grimme group. In particular, the tad-mctc library provides:

• autograd functions (Jacobian, Hessian)

• batch utility (packing, masks, …)

• atomic data (radii, EN, example molecules, …)

• io (reading coordinate files)

• coordination numbers

• safeops (autograd-safe implementations of common functions)

• typing (base class for tensor-like behavior of arbitrary classes)

• units

The name is inspired by the Fortran pendant “modular computation tool chain library” (mctc-lib).

Installation

pip

tad-mctc can easily be installed with pip.

pip install tad-mctc

From source

This project is hosted on GitHub at tad-mctc/tad-mctc. Obtain the source by cloning the repository with

git clone https://github.com/tad-mctc/tad-mctc
cd tad-mctc

We recommend using a conda environment to install the package. You can setup the environment manager using a mambaforge installer. Install the required dependencies from the conda-forge channel.

mamba env create -n torch -f environment.yaml
mamba activate torch

Install this project with pip in the environment

pip install .

The following dependencies are required

• numpy

• torch

• pytest (tests only)

Development

For development, additionally install the following tools in your environment.

mamba install black covdefaults mypy pre-commit pylint pytest pytest-cov pytest-xdist tox
pip install pytest-random-order

With pip, add the option -e for installing in development mode, and add [dev] for the development dependencies

pip install -e .[dev]

The pre-commit hooks are initialized by running the following command in the root of the repository.

pre-commit install

For testing all Python environments, simply run tox.

tox

Note that this randomizes the order of tests but skips “large” tests. To modify this behavior, tox has to skip the optional posargs.

tox -- test

Examples

The following example shows how to calculate the DFT-D4 dispersion energy for a single structure.

import torch
import tad_mctc as mctc

numbers = mctc.convert.symbol_to_number(symbols="C C C C N C S H H H H H".split())

# coordinates in Bohr
positions = torch.tensor(
    [
        [-2.56745685564671, -0.02509985979910, 0.00000000000000],
        [-1.39177582455797, +2.27696188880014, 0.00000000000000],
        [+1.27784995624894, +2.45107479759386, 0.00000000000000],
        [+2.62801937615793, +0.25927727028120, 0.00000000000000],
        [+1.41097033661123, -1.99890996077412, 0.00000000000000],
        [-1.17186102298849, -2.34220576284180, 0.00000000000000],
        [-2.39505990368378, -5.22635838332362, 0.00000000000000],
        [+2.41961980455457, -3.62158019253045, 0.00000000000000],
        [-2.51744374846065, +3.98181713686746, 0.00000000000000],
        [+2.24269048384775, +4.24389473203647, 0.00000000000000],
        [+4.66488984573956, +0.17907568006409, 0.00000000000000],
        [-4.60044244782237, -0.17794734637413, 0.00000000000000],
    ]
)

# calculate EEQ coordination number
cn = mctc.ncoord.cn_eeq(numbers, positions)
torch.set_printoptions(precision=10)
print(cn)
# tensor([3.0519218445, 3.0177774429, 3.0132560730, 3.0197706223,
#         3.0779352188, 3.0095663071, 1.0991339684, 0.9968624115,
#         0.9943327904, 0.9947233200, 0.9945874214, 0.9945726395])

The next example shows the calculation of dispersion energies for a batch of structures.

import torch
import tad_mctc as mctc

# S22 system 4: formamide dimer
numbers = mctc.batch.pack((
    mctc.convert.symbol_to_number("C C N N H H H H H H O O".split()),
    mctc.convert.symbol_to_number("C O N H H H".split()),
))

# coordinates in Bohr
positions = mctc.batch.pack((
    torch.tensor([
        [-3.81469488143921, +0.09993441402912, 0.00000000000000],
        [+3.81469488143921, -0.09993441402912, 0.00000000000000],
        [-2.66030049324036, -2.15898251533508, 0.00000000000000],
        [+2.66030049324036, +2.15898251533508, 0.00000000000000],
        [-0.73178529739380, -2.28237795829773, 0.00000000000000],
        [-5.89039325714111, -0.02589114569128, 0.00000000000000],
        [-3.71254944801331, -3.73605775833130, 0.00000000000000],
        [+3.71254944801331, +3.73605775833130, 0.00000000000000],
        [+0.73178529739380, +2.28237795829773, 0.00000000000000],
        [+5.89039325714111, +0.02589114569128, 0.00000000000000],
        [-2.74426102638245, +2.16115570068359, 0.00000000000000],
        [+2.74426102638245, -2.16115570068359, 0.00000000000000],
    ]),
    torch.tensor([
        [-0.55569743203406, +1.09030425468557, 0.00000000000000],
        [+0.51473634678469, +3.15152550263611, 0.00000000000000],
        [+0.59869690244446, -1.16861263789477, 0.00000000000000],
        [-0.45355203669134, -2.74568780438064, 0.00000000000000],
        [+2.52721209544999, -1.29200800956867, 0.00000000000000],
        [-2.63139587595376, +0.96447869452240, 0.00000000000000],
    ]),
))

# calculate coordination number
cn = mctc.ncoord.cn_d4(numbers, positions)
torch.set_printoptions(precision=10)
print(cn)
# tensor([[2.6886456013, 2.6886456013, 2.6314170361, 2.6314167976,
#          0.8594539165, 0.9231414795, 0.8605306745, 0.8605306745,
#          0.8594539165, 0.9231414795, 0.8568341732, 0.8568341732],
#         [2.6886456013, 0.8568335176, 2.6314167976, 0.8605306745,
#          0.8594532013, 0.9231414795, 0.0000000000, 0.0000000000,
#          0.0000000000, 0.0000000000, 0.0000000000, 0.0000000000]])

Contributing

This is a volunteer open source projects and contributions are always welcome. Please, take a moment to read the contributing guidelines.

License

This project is free software: you can redistribute it and/or modify it under the terms of the Lesser GNU General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version.

This project is distributed in the hope that it will be useful, but without any warranty; without even the implied warranty of merchantability or fitness for a particular purpose. See the Lesser GNU General Public License for more details.

Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in this project by you, as defined in the Lesser GNU General Public license, shall be licensed as above, without any additional terms or conditions.