RxnNet is a reaction network based python package for property prediction in solids.

Installation

Install from source:

git clone https://github.com/rasmusfr/rxnnet.git
cd rxnnet
pip install .

Usage

Reaction network generation

Using the default database, RxnNet can generate a reaction network (i.e. a series of balanced reactions). At the moment, reactions with up to 4 compounds are supported.

from rxnnet.generate_reactions import GenerateRN

target_id = 'mp-23193'  # target compound materials project id (mp-23193 is KCl)

reaction_network_gen = GenerateRN(target_id)
rn_data = reaction_network_gen.balanced_reactions(save=True)

print(rn_data[:50].to_markdown())

Predictions from the reaction network

From the generated reaction network we can obtain a prediction for the formation enthalpy.

import pandas as pd

from rxnnet.evaluate_reactions import EvaluateRN

target_id = 'mp-23193'

preferred_method = 'e_r2scan'  # preferred method; r^2SCAN electronic energy
fallback_method = 'e_gga_gga_u'  # fallback method; GGA/GGA+U electronic energy
reference_method = 'hf_ref'  # reference method; NBS enthalpy of formation
mode = 'ssw+cf'  # calculation mode; structural similarity weighting + chemistry filter

reaction_network_eval = EvaluateRN(target_id=target_id, preferred_method=preferred_method, reference_method=reference_method,
                                   rn=rf'user_reactions/{target_id}.pkl.gz', mode=mode, fallback_method=fallback_method)

results = reaction_network_eval.rn_evaluate()
print(pd.DataFrame(results)[:50].to_markdown())

Example notebooks

Notebook	Description
Intro	Introduction to reaction generation and prediction

Datasets

Datasets and results related to RxnNet can be found here.

License

This repository is licensed under the MIT license

Citing RxnNet

Please cite this paper if you use RxnNet:

@article{Fromsejer2024,
   author = {Rasmus Fromsejer and Bjørn Maribo-Mogensen and Georgios M. Kontogeorgis and Xiaodong Liang},
   doi = {10.1038/s41524-024-01404-5},
   issue = {1},
   journal = {npj Computational Materials},
   pages = {244},
   title = {Accurate formation enthalpies of solids using reaction networks},
   volume = {10},
   year = {2024},
}

Acknowledgement

The author wishes to thank the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (Grant Agreement no. 832460), ERC Advanced Grant project “New Paradigm in Electrolyte Thermodynamics” and the Department of Chemical Engineering at the Technical University of Denmark for funding this research.