ML LIB: Difference between revisions
No edit summary |
No edit summary |
||
| Line 15: | Line 15: | ||
{{sc|ML_LIB|Examples|Examples that use this tag}} | {{sc|ML_LIB|Examples|Examples that use this tag}} | ||
---- | ---- | ||
[[Category:INCAR]] | [[Category:INCAR]] | ||
[[Category:Machine-learned force fields]] | [[Category:Machine-learned force fields]] | ||
Revision as of 08:35, 24 January 2025
ML_LIB = [logical]
| Default: ML_LIB | = .TRUE. | if compiled with VASPml library (precompiler flag -Dlibvaspml)
|
| = .FALSE. | else |
Description: ML_LIB enables/disables the use of the VASPml C++ library for machine-learning features.
| Mind: Available as of VASP 6.5.0 |
Some machine learning modes of operation (ML_MODE) are implemented twice in VASP, once in the original Fortran code and again in a C++ rewrite as part of the VASPml library. However, the supported features and INCAR tags may be different. For example, the C++ does not yet implement the thermodynamic integration method. The ML_LIB tag allows to switch seamlessly between the two code paths without the need to recompile VASP.