A data-driven method combining symbolic regression and compressed sensing for accurate & interpretable models.
Version SISSO.3.3, July, 2023.
This code is licensed under the Apache License, Version 2.0
If you are using this code, please cite:
R. Ouyang, S. Curtarolo, E. Ahmetcik, M. Scheffler, and L. M. Ghiringhelli, Phys. Rev. Mater. 2, 083802 (2018).
(Please refer to the Refs. and the SISSO_guide.pdf for more details in using the code)
A Fortran mpi compiler is required to compile the SISSO parallel program. Below are two options for compiling the program using an IntelMPI compiler (other compilers may work as well). In the folder 'src', do:
(1) mpiifort -fp-model precise var_global.f90 libsisso.f90 DI.f90 FC.f90 SISSO.f90 -o ~/bin/SISSO
or (2) mpiifort -O2 var_global.f90 libsisso.f90 DI.f90 FC.f90 SISSO.f90 -o ~/bin/SISSO
Note:
Modules of the program:
Input Files: SISSO.in and train.dat, whose templates can be found in the folder input_templates.
Note that the input templates and the tools in the folder utilities may be modified accordingly when a new version of the code is released. Thus, users are recommended to always use the updated files, in particular the SISSO.in.
Command-line usage:
SISSO > log ! You may need to remove resource limit first by running the command 'ulimit -s unlimited'
Running on computer clusters, for example, using this command in your submission script:
mpirun -np number_of_cores SISSO >log
Primary Output Files:
More details on using this code can be found in the SISSO_guide.pdf
Created and maintained by Runhai Ouyang. Please feel free to open issues in the Github or contact Ouyang
([email protected]) in case of any problems/comments/suggestions in using the code.
SISSO++: https://gitlab.com/sissopp_developers/sissopp
MATLAB: https://github.com/NREL/SISSORegressor_MATLAB
Python interface: https://github.com/Matgenix/pysisso