The code used in the work “Inverse Design of Promising Alloys for Electrocatalytic CO2 Reduction via Generative Graph Neural Networks Combined with Bird Swarm Algorithm”
This is part of the Text-to-Materials project, aimed at the inverse design of novel CO2 reduction electrocatalysts.
- Clone this respostory
git clone https://github.com/szl666/CO2RR-inverse-design.git
cd CO2RR-inverse-design
- Install dependencies
pip install -r requirements.txt
-
To begin the inverse design process, ensure that the following model files are placed in the appropriate directories:
-
Surface generation model: Place the trained CDVAE model in the slab_generation_model folder.
-
Adsorption energy prediction model: Place the model files for CO and H adsorption energy predictions in the adsorption_predictor_model folder.
Once the models are in place, run the following code to start the inverse design:
python bsa_opt.py
- To analyze the inverse design results, including the average target property value per generation during BSA optimization, as well as the structural similarity between each generation and the previous one:
from t2mat_util.analyze_utilities import t2mat_analyze
t2mat_ana = t2mat_analyze(poscars_path='opt_poscars',training_data_path='')
- To analyze and visualize the inverse design results:
from t2mat_util.visualize_utilities import t2mat_visualize
t2mat_vis = t2mat_visualize(poscars_path='opt_poscars',if_analyze=True)