ML MB MIN - Revision history

Singraber at 08:03, 24 October 2025

2025-10-24T08:03:31Z

← Older revision		Revision as of 08:03, 24 October 2025
Line 2:		Line 2:
	{{TAGDEF\|ML_MB_MIN\|[integer]\|3}}		{{TAGDEF\|ML_MB_MIN\|[integer]\|3}}

	Description: {{TAG\|ML_MB_MIN}} sets the minimum number of local reference configurations required for generation of a machine-learned force field. {{NB\|mind\|Available as of {{VASP}} 6.4.3. Previous versions do not support this tag but act as if its default was {{~~TAGO~~\|ML_MB_MIN\|2}}.}}		Description: {{TAG\|ML_MB_MIN}} sets the minimum number of local reference configurations required for generation of a machine-learned force field. {{NB\|mind\|Available as of {{VASP}} 6.4.3. Previous versions do not support this tag but act as if its default was {{TAG\|ML_MB_MIN\|2}}.}}
	----		----
	Machine-learned force fields based on the kernel method require a reasonable amount of reference atomic environments in order to provide meaningful predictions. Typically, the on-the-fly algorithm in {{VASP}} collects these local reference configurations (also called kernel basis functions) during an MD simulation ({{~~TAGO~~\|ML_MODE\|train}}). However, at the start of the trajectory usually only very few local reference configurations are available. How many depends on the system size, symmetry and number of atoms (per type). This tag controls how many local reference configurations are required '''for each atom type''' before training of a force field is allowed. Setting a higher value may yield a more robust initial force field at the cost of more ''ab initio'' calculations during the first few MD steps.		Machine-learned force fields based on the kernel method require a reasonable amount of reference atomic environments in order to provide meaningful predictions. Typically, the on-the-fly algorithm in {{VASP}} collects these local reference configurations (also called kernel basis functions) during an MD simulation ({{TAG\|ML_MODE\|train}}). However, at the start of the trajectory usually only very few local reference configurations are available. How many depends on the system size, symmetry and number of atoms (per type). This tag controls how many local reference configurations are required '''for each atom type''' before training of a force field is allowed. Setting a higher value may yield a more robust initial force field at the cost of more ''ab initio'' calculations during the first few MD steps.

	If a training is scheduled in the current step, i.e., the <code>STATUS</code> line in the {{FILE\|ML_LOGFILE}} shows <code>learning</code> or <code>critical</code>, but not enough atomic environments were collected, a text message will be emitted in the <code>MSG</code> line. Here is an example:		If a training is scheduled in the current step, i.e., the <code>STATUS</code> line in the {{FILE\|ML_LOGFILE}} shows <code>learning</code> or <code>critical</code>, but not enough atomic environments were collected, a text message will be emitted in the <code>MSG</code> line. Here is an example:

Singraber at 09:28, 20 March 2024

2024-03-20T09:28:40Z

← Older revision		Revision as of 09:28, 20 March 2024
Line 46:		Line 46:

	----		----
	~~<!-- Link to categories like this:~~ [[Category:INCAR]][[Category:~~Electronic Minimization~~]]		[[Category:INCAR tag]][[Category:Machine-learned force fields]]
	~~only comment it out when you want the page to show up on the category page, i.e., not when it is in the Construction namespace.-->~~

Huebsch: Huebsch moved page Construction:ML MB MIN to ML MB MIN without leaving a redirect

2024-03-19T14:36:39Z

Huebsch moved page Construction:ML MB MIN to ML MB MIN without leaving a redirect

← Older revision	Revision as of 14:36, 19 March 2024
(No difference)

Singraber at 09:39, 19 February 2024

2024-02-19T09:39:05Z

@@ Line 4: / Line 4: @@
 Description: {{TAG|ML_MB_MIN}} sets the minimum number of local reference configurations required for generation of a machine-learned force field. {{NB|mind|Available as of {{VASP}} 6.4.3. Previous versions do not support this tag but act as if its default was {{TAGO|ML_MB_MIN|2}}.}}
 ----
-Machine-learned force fields based on the kernel method require a reasonable amount of reference atomic environments in order to provide meaningful predictions. Typically, the on-the-fly algorithm in {{VASP}} collects these local reference configurations (also called kernel basis functions) during an MD simulation ({{TAGO|ML_MODE|train}}). However, at the start of the trajectory usually only very few local reference configurations are available. How many depends on the system size, symmetry and number of atoms (per type).
+Machine-learned force fields based on the kernel method require a reasonable amount of reference atomic environments in order to provide meaningful predictions. Typically, the on-the-fly algorithm in {{VASP}} collects these local reference configurations (also called kernel basis functions) during an MD simulation ({{TAGO|ML_MODE|train}}). However, at the start of the trajectory usually only very few local reference configurations are available. How many depends on the system size, symmetry and number of atoms (per type). This tag controls how many local reference configurations are required '''for each atom type''' before training of a force field is allowed. Setting a higher value may yield a more robust initial force field at the cost of  more ''ab initio'' calculations during the first few MD steps.
 == Related tags and articles==

Singraber at 11:03, 16 February 2024

2024-02-16T11:03:34Z

← Older revision		Revision as of 11:03, 16 February 2024
Line 4:		Line 4:
	Description: {{TAG\|ML_MB_MIN}} sets the minimum number of local reference configurations required for generation of a machine-learned force field. {{NB\|mind\|Available as of {{VASP}} 6.4.3. Previous versions do not support this tag but act as if its default was {{TAGO\|ML_MB_MIN\|2}}.}}		Description: {{TAG\|ML_MB_MIN}} sets the minimum number of local reference configurations required for generation of a machine-learned force field. {{NB\|mind\|Available as of {{VASP}} 6.4.3. Previous versions do not support this tag but act as if its default was {{TAGO\|ML_MB_MIN\|2}}.}}
	----		----
	Machine-learned force fields based on the kernel method require a reasonable amount of reference atomic environments in order to provide meaningful predictions. Typically, the on-the-fly algorithm in {{VASP}} collects these local reference configurations (also called kernel basis functions) during an MD simulation ({{TAGO\|ML_MODE\|train}}). However, at the start of the trajectory usually only very few local reference configurations are available. How many depends on the system size and ~~symmetry~~.		Machine-learned force fields based on the kernel method require a reasonable amount of reference atomic environments in order to provide meaningful predictions. Typically, the on-the-fly algorithm in {{VASP}} collects these local reference configurations (also called kernel basis functions) during an MD simulation ({{TAGO\|ML_MODE\|train}}). However, at the start of the trajectory usually only very few local reference configurations are available. How many depends on the system size, symmetry and number of atoms (per type).

Singraber at 17:04, 15 February 2024

2024-02-15T17:04:00Z

← Older revision		Revision as of 17:04, 15 February 2024
Line 2:		Line 2:
	{{TAGDEF\|ML_MB_MIN\|[integer]\|3}}		{{TAGDEF\|ML_MB_MIN\|[integer]\|3}}

	Description: {{TAG\|ML_MB_MIN}} sets the minimum number of local reference configurations required for generation of a machine-learned force field. Available as of {{VASP}} 6.4.3.		Description: {{TAG\|ML_MB_MIN}} sets the minimum number of local reference configurations required for generation of a machine-learned force field. {{NB\|mind\|Available as of {{VASP}} 6.4.3. Previous versions do not support this tag but act as if its default was {{TAGO\|ML_MB_MIN\|2}}.}}
	----		----
			Machine-learned force fields based on the kernel method require a reasonable amount of reference atomic environments in order to provide meaningful predictions. Typically, the on-the-fly algorithm in {{VASP}} collects these local reference configurations (also called kernel basis functions) during an MD simulation ({{TAGO\|ML_MODE\|train}}). However, at the start of the trajectory usually only very few local reference configurations are available. How many depends on the system size and symmetry.


	== Related tags and articles==		== Related tags and articles==

Singraber at 15:53, 15 February 2024

2024-02-15T15:53:03Z

← Older revision		Revision as of 15:53, 15 February 2024
Line 1:		Line 1:
	{{DISPLAYTITLE:ML_MB_MIN}}		{{DISPLAYTITLE:ML_MB_MIN}}
	{{TAGDEF\|ML_MB_MIN\|[integer]\|2}}		{{TAGDEF\|ML_MB_MIN\|[integer]\|3}}

	Description: {{TAG\|ML_MB_MIN}} sets the minimum number of local reference configurations required for generation of a machine-learned force field. Available as of {{VASP}} 6.4.3.		Description: {{TAG\|ML_MB_MIN}} sets the minimum number of local reference configurations required for generation of a machine-learned force field. Available as of {{VASP}} 6.4.3.

Singraber at 13:23, 9 February 2024

2024-02-09T13:23:48Z

← Older revision		Revision as of 13:23, 9 February 2024
Line 1:		Line 1:
	{{~~TAGDEF\|~~ML_MB_MIN~~\|[integer]~~}}		{{DISPLAYTITLE:ML_MB_MIN}}
	{{~~DEF~~\|ML_MB_MIN\|2\|}}		{{TAGDEF\|ML_MB_MIN\|[integer]\|2}}

	Description: {{TAG\|ML_MB_MIN}} sets the minimum number of local reference configurations required for generation of a machine-learned force field.		Description: {{TAG\|ML_MB_MIN}} sets the minimum number of local reference configurations required for generation of a machine-learned force field. Available as of {{VASP}} 6.4.3.
	----		----

Singraber: Created page with "{{TAGDEF|ML_MB_MIN|[integer]}} {{DEF|ML_MB_MIN|2|}} Description: {{TAG|ML_MB_MIN}} sets the minimum number of local reference configurations required for generation of a machine-learned force field. ---- == Related tags and articles== {{TAG|ML_LMLFF}}, {{TAG|ML_MODE}}, {{TAG|ML_MB}}, {{TAG|ML_MCONF_NEW}}, {{TAG|ML_MCONF}}, {{TAG| ML_LBASIS_DISCARD}} {{sc|ML_MB_MIN|Examples|Examples that use this tag}} ----

← Older revision		Revision as of 08:03, 24 October 2025
Line 2:		Line 2:
	{{TAGDEF\|ML_MB_MIN\|[integer]\|3}}		{{TAGDEF\|ML_MB_MIN\|[integer]\|3}}

	Description: {{TAG\|ML_MB_MIN}} sets the minimum number of local reference configurations required for generation of a machine-learned force field. {{NB\|mind\|Available as of {{VASP}} 6.4.3. Previous versions do not support this tag but act as if its default was {{~~TAGO~~\|ML_MB_MIN\|2}}.}}		Description: {{TAG\|ML_MB_MIN}} sets the minimum number of local reference configurations required for generation of a machine-learned force field. {{NB\|mind\|Available as of {{VASP}} 6.4.3. Previous versions do not support this tag but act as if its default was {{TAG\|ML_MB_MIN\|2}}.}}
	----		----
	Machine-learned force fields based on the kernel method require a reasonable amount of reference atomic environments in order to provide meaningful predictions. Typically, the on-the-fly algorithm in {{VASP}} collects these local reference configurations (also called kernel basis functions) during an MD simulation ({{~~TAGO~~\|ML_MODE\|train}}). However, at the start of the trajectory usually only very few local reference configurations are available. How many depends on the system size, symmetry and number of atoms (per type). This tag controls how many local reference configurations are required '''for each atom type''' before training of a force field is allowed. Setting a higher value may yield a more robust initial force field at the cost of more ''ab initio'' calculations during the first few MD steps.		Machine-learned force fields based on the kernel method require a reasonable amount of reference atomic environments in order to provide meaningful predictions. Typically, the on-the-fly algorithm in {{VASP}} collects these local reference configurations (also called kernel basis functions) during an MD simulation ({{TAG\|ML_MODE\|train}}). However, at the start of the trajectory usually only very few local reference configurations are available. How many depends on the system size, symmetry and number of atoms (per type). This tag controls how many local reference configurations are required '''for each atom type''' before training of a force field is allowed. Setting a higher value may yield a more robust initial force field at the cost of more ''ab initio'' calculations during the first few MD steps.

	If a training is scheduled in the current step, i.e., the <code>STATUS</code> line in the {{FILE\|ML_LOGFILE}} shows <code>learning</code> or <code>critical</code>, but not enough atomic environments were collected, a text message will be emitted in the <code>MSG</code> line. Here is an example:		If a training is scheduled in the current step, i.e., the <code>STATUS</code> line in the {{FILE\|ML_LOGFILE}} shows <code>learning</code> or <code>critical</code>, but not enough atomic environments were collected, a text message will be emitted in the <code>MSG</code> line. Here is an example: