ML CX: Difference between revisions

From VASP Wiki
m (Karsai moved page ML FF XMIX to ML XMIX)
No edit summary
Line 1: Line 1:
{{TAGDEF|ML_FF_XMIX|[integer]|0.5}}
{{TAGDEF|ML_FF_XMIX|[integer]|1.0}}


Description: The parameter determines how the  threshold  ({{TAG|ML_FF_CTIFOR}}) is updated in the machine learning force field methods.
Description: The parameter determines how the  threshold  ({{TAG|ML_FF_CTIFOR}}) is updated in the machine learning force field methods.

Revision as of 11:47, 27 August 2021

ML_FF_XMIX = [integer]
Default: ML_FF_XMIX = 1.0 

Description: The parameter determines how the threshold (ML_FF_CTIFOR) is updated in the machine learning force field methods.


If ML_FF_LCRITERIA is .TRUE., ML_FF_CTIFOR is set to the average of the Bayesian errors of the forces stored in a history (see ML_FF_LCRITERIA). The number of entries in the history are controlled by ML_FF_MHIS. To avoid too abrupt changes, the average Bayesian error is mixed with the old value of ML_FF_CTIFOR:

ML_FF_CTIFOR = (1-ML_FF_XMIX) * ML_FF_CTIFOR + ML_FF_XMIX * (average of the stored Bayesian errors).

This implies that for ML_FF_XMIX=1, the old value stored in ML_FF_CTIFOR is simply overwritten by the current average Bayesian error. This tag is particularly useful at the beginning to avoid that the value supplied in the INCAR file changes too quickly.

Related Tags and Sections

ML_FF_LMLFF, ML_FF_LCRITERIA, ML_FF_CTIFOR, ML_FF_MHIS

Examples that use this tag