Stochiometrically-Informed Symbolic Regression (SISR)

Description

---

A stoichiometrically-informed method to fit the rate constants in a reaction mechanism through differential optimization and couple that fitting method with a genetic optimization approach that searches a symbolic space of possible reaction mechanisms to find the mechanism that best matches a time-series dataset of concentrations. It returns a symbolic chemical reaction mechanism, the rate constant for each reaction in the mechanism, and the kinetic equations that describe the chemical process under examination.

Contents

---

This repository includes the codes and results for the paper : Stoichiometrically-informed symbolic regression for extracting chemical reaction mechanisms from data (Manuel Palma Banos,Joel D. Kress,Rigoberto Hernandez,Galen T. Craven,Journal Info, Year).

The repository is organized as follows:

• Example input datasets and scripts to run the SISR code: examples/
• Extact results included in our paper: results/
• Paper and associated figures: docs/
• The python3 based source code: src/

How to use

---

• Note: The ./install.sh script will create a conda virtual environment. If you have a different directory structure, please modify the script accordingly.

• Note: The ./install.sh script includes a unittest verification step. If you want to skip this step, you can comment out the line that runs the unittests in the script.

• Note: The ./install.sh script will install the required packages in the virtual environment. It will not install them globally.

• Note: The code is designed to run on Python 3.9 or higher. It has not been tested on Python 2.x.

• Note: The code is designed to run on Linux and MacOS. It has not been tested on Windows.

• Clone SISR

git clone https://github.com/rxhernandez/SISR

• Install SISR

cd SISR
./install.sh

If system was intalled correctly and unittests passed, you should see the following message:

----------------------------------------------------------------------
Ran 19 tests in 0.042s

OK
Installation complete. To activate this environment later, run:
  conda activate sisr-env

• Virtual Environment

conda activate sisr-env

• Check Installation

pip list | grep SISR

You should see the installed SISR package listed as follows:

SISR                          0.1.0

• Run Example

cd examples
python LV_example.py

• Uninstall

conda deactivate
./uninstall.sh

Citing

---

If you use database or codes, please consider citing the paper:

M. Palma Banos, J. D. Kress, R. Hernandez, G. T. Craven, "Stoichiometrically-informed symbolic regression for ex- tracting chemical reaction mechanisms from data," (in preparation).

and/or this site:

M. Palma Banos, J. D. Kress, R. Hernandez, G. T. Craven, SISR, URL, https://github.com/rxhernandez/SISR

Acknowledgment

---

This work was supported by the Los Alamos National Laboratory (LANL) Directed Research and Development funds (LDRD). This work was supported by the National Science Foundation through Grant No.~CHE 2102455.

License

---

SISR code is distributed under terms of the [MIT License]