ML IERR: Difference between revisions

Latest revision as of 14:25, 31 March 2023

Default: ML_IERR	= 0	if ML_MODE=RUN
	= 1	otherwise

Description: Calculation and output frequency of Bayesian error estimate.

Mind: This tag is only available as of VASP.6.4.0.

This tag sets the distance in units of molecular-dynamics steps at which the energies, forces, stresses, etc. are written to the corresponding files (OUTCAR, OSZICAR, ML_LOGFILE, etc.). If learning is activated (ML_MODE=TRAIN, SELECT, REFIT or REFITBAYESIAN) the only allowed option is that Bayesian error estimation is "on" at every molecular-dynamics step (ML_IERR=1 is required).

The interval can only be freely chosen via ML_IERR for ML_MODE=RUN, if an ML_FF file, that was previously trained using ML_MODE=REFITBAYESIAN, is used. The Bayesian error estimates are written to the logfile with the entry BEEF. For ML_IERR=0 the Bayesian error is never calculated (and also no corresponding log file line is written out. This is the default since for ML_MODE=RUN the Bayesian error estimation takes up a significant amount of the total calculation time.

Warning: ML_IERR>0 can never be used with the fast execution mode (force field refitted with ML_MODE=REFIT or ML_LFAST=.TRUE.).

@@ Line 1: / Line 1: @@
 {{DISPLAYTITLE:ML_IERR}}
-{{DEF|ML_IERR|0|if {{TAG|ML_ISTART}}{{=}}2|1|otherwise}}
+{{DEF|ML_IERR|0|if {{TAG|ML_MODE}}{{=}}RUN|1|otherwise}}
 Description: Calculation and output frequency of Bayesian error estimate.
+{{NB|mind|This tag is only available as of VASP.6.4.0.}}
 ----
-This tag sets the distance of molecular-dynamics steps for which the Bayesian error estimates are calculated and written to the {{TAG|ML_LOGFILE}}. This means that in every {{TAG|ML_IERR}} steps an entry corresponding to the keyword <code>BEE</code> and/or <code>BEEF</code> is written. If learning is activated ({{TAG|ML_ISTART}}=0,1 or 2) Bayesian error estimation must be "on" at every molecular-dynamics step ({{TAG|ML_IERR}}=1 is required).
-Only for {{TAG|ML_ISTART}}=2 can {{TAG|ML_IERR}} be freely chosen. For {{TAG|ML_IERR}}=0 the Bayesian error is never calculated (and also no line is written out). This is the default since for {{TAG|ML_ISTART}}=2 the Bayesian error estimation takes up a significant amount of the total calculation time.
+This tag sets the distance in units of molecular-dynamics steps at which the energies, forces, stresses, etc. are written to the corresponding files ({{TAG|OUTCAR}}, {{TAG|OSZICAR}}, {{TAG|ML_LOGFILE}}, etc.). If learning is activated ({{TAG|ML_MODE}}=''TRAIN'', ''SELECT'', ''REFIT'' or ''REFITBAYESIAN'') the only allowed option is that Bayesian error estimation is "on" at every molecular-dynamics step ({{TAG|ML_IERR}}=1 is required).
-{{TAG|ML_IERR}}>0 can be also used with the fast execution mode {{TAG|ML_LFAST}}=.TRUE., but at each step where the Bayesian error estimation is carried out, the code has to switch to the slow version. So choosing a too-narrow distance for {{TAG|ML_IERR}} will result in a huge slowdown of the code.
+The interval can only be freely chosen via {{TAG|ML_IERR}} for {{TAG|ML_MODE}}=''RUN'', if an {{TAG|ML_FF}} file, that was previously trained using {{TAG|ML_MODE}}=''REFITBAYESIAN'', is used. The Bayesian error estimates are written to the logfile with the entry ''BEEF''. For {{TAG|ML_IERR}}=0 the Bayesian error is never calculated (and also no corresponding log file line is written out. This is the default since for {{TAG|ML_MODE}}=''RUN'' the Bayesian error estimation takes up a significant amount of the total calculation time.
+{{NB|warning|{{TAG|ML_IERR}}>0 can never be used with the fast execution mode (force field refitted with {{TAG|ML_MODE}}{{=}}''REFIT'' or {{TAG|ML_LFAST}}{{=}}''.TRUE.'').}}
 == Related tags and articles ==
-{{TAG|ML_LMLFF}}, {{TAG|ML_ISTART}}, {{TAG|ML_LFAST}}, {{TAG|ML_OUTBLOCK}}, {{TAG|ML_OUTPUT_MODE}}
+{{TAG|ML_LMLFF}}, {{TAG|ML_MODE}}, {{TAG|ML_LFAST}}, {{TAG|ML_OUTBLOCK}}, {{TAG|ML_OUTPUT_MODE}}
 ----
-[[Category:INCAR tag]][[Category:Machine-learned force fields]][[Category:Development version]]
+[[Category:INCAR tag]][[Category:Machine-learned force fields]]