ML ICRITERIA: Difference between revisions
(ML_XMIX reference added) |
|||
| Line 9: | Line 9: | ||
* {{TAG|ML_ICRITERIA}} = 0: No update of the threshold {{TAG|ML_CTIFOR}} is done. | * {{TAG|ML_ICRITERIA}} = 0: No update of the threshold {{TAG|ML_CTIFOR}} is done. | ||
* {{TAG|ML_ICRITERIA}} = 1: Update of criteria using average of the Bayesian errors of the forces from history (see description of method [[Machine learning force field calculations: Basics#Threshold for error of forces|here]]). | * {{TAG|ML_ICRITERIA}} = 1: Update of criteria using average of the Bayesian errors of the forces from history (see description of method [[Machine learning force field calculations: Basics#Threshold for error of forces|here]]). | ||
* {{TAG|ML_ICRITERIA}} = 2: Update of criteria using gliding average of Bayesian errors (probably more robust but '''not well tested'''). | * {{TAG|ML_ICRITERIA}} = 2: Update of criteria using gliding average of Bayesian errors (probably more robust but '''not well tested'''). | ||
For the options {{TAG|ML_ICRITERIA}} = 1 and {{TAG|ML_ICRITERIA}} = 2, {{TAG|ML_CTIFOR}} is typically set to the average of previous Bayesian errors. The tag {{TAG|ML_XMIX}} allows to fine tune | |||
how {{TAG|ML_CTIFOR}} is updated. | |||
== Related tags and articles == | == Related tags and articles == | ||
Revision as of 08:42, 13 September 2022
ML_ICRITERIA = [integer]
Default: ML_ICRITERIA = 1
Description: Decides whether (ML_ICRITERIA>0) or how the Bayesian error threshold (ML_CTIFOR) is updated within the machine learning force field method. ML_CTIFOR determines whether a first principles calculations is performed.
The usage of this tag in combination with the learning algorithms is described here: here.
The following options are possible for ML_ICRITERIA:
- ML_ICRITERIA = 0: No update of the threshold ML_CTIFOR is done.
- ML_ICRITERIA = 1: Update of criteria using average of the Bayesian errors of the forces from history (see description of method here).
- ML_ICRITERIA = 2: Update of criteria using gliding average of Bayesian errors (probably more robust but not well tested).
For the options ML_ICRITERIA = 1 and ML_ICRITERIA = 2, ML_CTIFOR is typically set to the average of previous Bayesian errors. The tag ML_XMIX allows to fine tune how ML_CTIFOR is updated.
Related tags and articles
ML_LMLFF, ML_CTIFOR, ML_CSLOPE, ML_CSIG, ML_MHIS, ML_XMIX