ML CDOUB: Difference between revisions

From VASP Wiki
No edit summary
No edit summary
(6 intermediate revisions by 2 users not shown)
Line 4: Line 4:
Description: This flag controls the necessity of DFT calculations in the machine learning force field method.
Description: This flag controls the necessity of DFT calculations in the machine learning force field method.
----
----
This flag is applied in the case of {{TAG|ML_FF_ISAMPLE}}=2 or 3. In that case if at any time the estimated error is {{TAG|ML_FF_CDOUB}} times larger than the threshold ({{TAG|ML_FF_CSF}} for the spilling factor and {{TAG|ML_FF_CTIFOR}} or it's updated value form {{TAG|ML_FF_LCRITERIA}}=''.TRUE.'' for the Bayesian error), the sampling is stopped and a new force field is generated (even if the counter for sampling is below the minimum amount of sampled structures {{TAG|ML_FF_NMDINT}}).
This flag is applied only if {{TAG|ML_FF_ISAMPLE}}=2 or 3. If at any time, the estimated errors are {{TAG|ML_FF_CDOUB}} times larger than the Bayesian threshold {{TAG|ML_FF_CTIFOR}} or the threshold for the spilling factor {{TAG|ML_FF_CSF}}, a first principles calculation is performed and a new force field is immediately generated (even if the counter for sampling is below the minimum amount of sampled structures {{TAG|ML_FF_NMDINT}}). For details, refer to  {{TAG|ML_FF_ISAMPLE}}.


== Related Tags and Sections ==
== Related Tags and Sections ==
{{TAG|ML_FF_LMLFF}}, {{TAG|ML_FF_ISAMPLE}}, {{TAG|ML_FF_CSF}}, {{TAG|ML_FF_CTIFOR}}, {{TAG|ML_FF_CTIFOR}}, {{TAG|ML_FF_IERR}}
{{TAG|ML_FF_ISAMPLE}}, {{TAG|ML_FF_LMLFF}}, {{TAG|ML_FF_CSF}}, {{TAG|ML_FF_CTIFOR}}, {{TAG|ML_FF_IERR}}, {{TAG|ML_FF_NMDINT}}


{{sc|ML_FF_CDOUB|Examples|Examples that use this tag}}
{{sc|ML_FF_CDOUB|Examples|Examples that use this tag}}
----
----


[[Category:INCAR]][[Category:Machine Learning]][[Category:Machine Learned Force Fields]]
[[Category:INCAR]][[Category:Machine Learning]][[Category:Machine Learned Force Fields]][[Category: Alpha]]

Revision as of 07:40, 5 October 2020

ML_FF_CDOUB = [real]
Default: ML_FF_CDOUB = 2.0 

Description: This flag controls the necessity of DFT calculations in the machine learning force field method.


This flag is applied only if ML_FF_ISAMPLE=2 or 3. If at any time, the estimated errors are ML_FF_CDOUB times larger than the Bayesian threshold ML_FF_CTIFOR or the threshold for the spilling factor ML_FF_CSF, a first principles calculation is performed and a new force field is immediately generated (even if the counter for sampling is below the minimum amount of sampled structures ML_FF_NMDINT). For details, refer to ML_FF_ISAMPLE.

Related Tags and Sections

ML_FF_ISAMPLE, ML_FF_LMLFF, ML_FF_CSF, ML_FF_CTIFOR, ML_FF_IERR, ML_FF_NMDINT

Examples that use this tag