NBANDS - Revision history

Schlipf at 07:42, 31 March 2025

2025-03-31T07:42:14Z

← Older revision		Revision as of 07:42, 31 March 2025
Line 27:		Line 27:

	----		----
	[[Category:INCAR tag]][[Category:Electronic minimization]][[Category:Dielectric properties]]		[[Category:INCAR tag]][[Category:Electronic minimization]][[Category:Dielectric properties]][[Category:Parallelization]]

Schlipf: /* Many-body perturbation theory calculatioins */

2022-05-30T13:35:27Z

Many-body perturbation theory calculatioins

← Older revision		Revision as of 13:35, 30 May 2022
Line 11:		Line 11:
	{{NB\|tip\|Note that the {{TAG\|RMM-DIIS}} scheme ({{TAG\|ALGO}}{{=}}Fast) is more sensitive to the number of bands than the default Davidson algorithm ({{TAG\|ALGO}}{{=}}Normal) and can require more bands for fast convergence.\|}}		{{NB\|tip\|Note that the {{TAG\|RMM-DIIS}} scheme ({{TAG\|ALGO}}{{=}}Fast) is more sensitive to the number of bands than the default Davidson algorithm ({{TAG\|ALGO}}{{=}}Normal) and can require more bands for fast convergence.\|}}

	==== Many-body perturbation theory ~~calculatioins~~ ====		==== Many-body perturbation theory calculations ====
	In the Many-Body Perturbation Theory calculations ([[Practical guide to GW calculations\|''GW'']],[[RPA/ACFDT: Correlation energy in the Random Phase Approximation\|RPA]], and [[BSE calculations\|BSE]]), a large number of empty orbitals is usually required, which can be much higher than the number of occupied states. The convergence of the calculations with a large number of empty states can be very slow. In such cases, we recommend performing an exact diagonalization ({{TAG\|ALGO}}=Exact) of the Hamiltonian with empty bands starting from a converged charge density.		In the Many-Body Perturbation Theory calculations ([[Practical guide to GW calculations\|''GW'']],[[RPA/ACFDT: Correlation energy in the Random Phase Approximation\|RPA]], and [[BSE calculations\|BSE]]), a large number of empty orbitals is usually required, which can be much higher than the number of occupied states. The convergence of the calculations with a large number of empty states can be very slow. In such cases, we recommend performing an exact diagonalization ({{TAG\|ALGO}}=Exact) of the Hamiltonian with empty bands starting from a converged charge density.

Karsai: /* Related tags and article */

2022-04-08T14:46:48Z

Tal: /* Electronic minimization */

2022-04-08T10:43:57Z

Electronic minimization

← Older revision		Revision as of 10:43, 8 April 2022
Line 8:		Line 8:
	==== Electronic minimization ====		==== Electronic minimization ====
	In the electronic minimization calculations, empty states do not contribute to the total energy, however, empty states are required to achieve a better convergence.		In the electronic minimization calculations, empty states do not contribute to the total energy, however, empty states are required to achieve a better convergence.
	In iterative matrix-diagonalization algorithms (see {{TAG\|ALGO}}) eigenvectors close to the top of the calculated number of states converge much slower than the lowest eigenstates, thus it is important to choose a sufficiently large {{TAG\|NBANDS}}. Therefore, we recommend using the default settings for {{TAG\|NBANDS}}, i.e. ''NELECT/2 + NIONS/2'', which is a safe choice in most cases. In some cases, it is also possible to decrease it to ''NELECT/2+NIONS/4'', however, in some transition metals with open ''f'' shells a much larger number of empty bands might be required (up to ''NELECT/2+2*NIONS''). To check this parameter perform several calculations for a fixed potential ({{TAG\|ICHARG}}=12) with an increasing number of bands (e.g. starting from ''NELECT/2 + NIONS/2''). An accuracy of <math> 10^{-6}</math> should be obtained in 10-15 iterations.		In iterative matrix-diagonalization algorithms (see {{TAG\|ALGO}}) eigenvectors close to the top of the calculated number of states converge much slower than the lowest eigenstates, thus it is important to choose a sufficiently large {{TAG\|NBANDS}}. Therefore, we recommend using the default settings for {{TAG\|NBANDS}}, i.e., ''NELECT/2 + NIONS/2'', which is a safe choice in most cases. In some cases, it is also possible to decrease it to ''NELECT/2+NIONS/4'', however, in some transition metals with open ''f'' shells a much larger number of empty bands might be required (up to ''NELECT/2+2*NIONS''). To check this parameter perform several calculations for a fixed potential ({{TAG\|ICHARG}}=12) with an increasing number of bands (e.g. starting from ''NELECT/2 + NIONS/2''). An accuracy of <math> 10^{-6}</math> should be obtained in 10-15 iterations.
	{{NB\|tip\|Note that the {{TAG\|RMM-DIIS}} scheme ({{TAG\|ALGO}}{{=}}Fast) is more sensitive to the number of bands than the default Davidson algorithm ({{TAG\|ALGO}}{{=}}Normal) and can require more bands for fast convergence.\|}}		{{NB\|tip\|Note that the {{TAG\|RMM-DIIS}} scheme ({{TAG\|ALGO}}{{=}}Fast) is more sensitive to the number of bands than the default Davidson algorithm ({{TAG\|ALGO}}{{=}}Normal) and can require more bands for fast convergence.\|}}

	==== Many-body perturbation theory calculatioins ====		==== Many-body perturbation theory calculatioins ====

Tal: /* Many-Body Perturbation Theory */

2022-04-08T10:43:28Z

Many-Body Perturbation Theory

← Older revision		Revision as of 10:43, 8 April 2022
Line 11:		Line 11:
	{{NB\|tip\|Note that the {{TAG\|RMM-DIIS}} scheme ({{TAG\|ALGO}}{{=}}Fast) is more sensitive to the number of bands than the default Davidson algorithm ({{TAG\|ALGO}}{{=}}Normal) and can require more bands for fast convergence.\|}}		{{NB\|tip\|Note that the {{TAG\|RMM-DIIS}} scheme ({{TAG\|ALGO}}{{=}}Fast) is more sensitive to the number of bands than the default Davidson algorithm ({{TAG\|ALGO}}{{=}}Normal) and can require more bands for fast convergence.\|}}

	==== Many-~~Body Perturbation Theory~~ ====		==== Many-body perturbation theory calculatioins ====
	In the Many-Body Perturbation Theory calculations ([[Practical guide to GW calculations\|''GW'']],[[RPA/ACFDT: Correlation energy in the Random Phase Approximation\|RPA]], and [[BSE calculations\|BSE]]), a large number of empty orbitals is usually required, which can be much higher than the number of occupied states. The convergence of the calculations with a large number of empty states can be very slow. In such cases, we recommend performing an exact diagonalization ({{TAG\|ALGO}}=Exact) of the Hamiltonian with empty bands starting from a converged charge density.		In the Many-Body Perturbation Theory calculations ([[Practical guide to GW calculations\|''GW'']],[[RPA/ACFDT: Correlation energy in the Random Phase Approximation\|RPA]], and [[BSE calculations\|BSE]]), a large number of empty orbitals is usually required, which can be much higher than the number of occupied states. The convergence of the calculations with a large number of empty states can be very slow. In such cases, we recommend performing an exact diagonalization ({{TAG\|ALGO}}=Exact) of the Hamiltonian with empty bands starting from a converged charge density.

Tal: /* Many-Body Perturbation Theory */

2022-04-08T09:50:54Z

Many-Body Perturbation Theory

← Older revision		Revision as of 09:50, 8 April 2022
Line 12:		Line 12:

	==== Many-Body Perturbation Theory ====		==== Many-Body Perturbation Theory ====
	In the Many-Body Perturbation Theory calculations ([[Practical guide to GW calculations\|''GW'']],[[RPA/ACFDT: Correlation energy in the Random Phase Approximation\|RPA]], and [[BSE calculations\|BSE]]), a large number of empty orbitals is usually required ~~and~~ can be much higher than the number of occupied states. The convergence of the calculations with a large number of empty states can be very slow. In such cases, we recommend performing an exact diagonalization ({{TAG\|ALGO}}=Exact) of the Hamiltonian with empty bands starting from a converged charge density.		In the Many-Body Perturbation Theory calculations ([[Practical guide to GW calculations\|''GW'']],[[RPA/ACFDT: Correlation energy in the Random Phase Approximation\|RPA]], and [[BSE calculations\|BSE]]), a large number of empty orbitals is usually required, which can be much higher than the number of occupied states. The convergence of the calculations with a large number of empty states can be very slow. In such cases, we recommend performing an exact diagonalization ({{TAG\|ALGO}}=Exact) of the Hamiltonian with empty bands starting from a converged charge density.

	==== Parallelization ====		==== Parallelization ====

Tal at 09:49, 8 April 2022

2022-04-08T09:49:02Z

← Older revision		Revision as of 09:49, 8 April 2022
Line 2:		Line 2:
	{{DEF\|NBANDS\|max({{TAG\|NELECT}}/2+NIONS/2,{{TAG\|NELECT}}*0.6)\|}}		{{DEF\|NBANDS\|max({{TAG\|NELECT}}/2+NIONS/2,{{TAG\|NELECT}}*0.6)\|}}

	Description: {{TAG\|NBANDS}} specifies the number of ~~bands~~ in the calculation.		Description: {{TAG\|NBANDS}} specifies the total number of KS or QP orbitals in the calculation.
	----		----
	The right choice of {{TAG\|NBANDS}} strongly depends on the type of the performed		The right choice of {{TAG\|NBANDS}} strongly depends on the type of the performed calculation and the system. As a minimum, VASP requires all occupied states + one empty band, otherwise, a warning is given.
	calculation and the system. As a minimum, VASP requires one empty band,
	otherwise, a warning is given~~. Only in large gap insulators, accurate~~
	~~results can be obtained without empty bands and the warning might be ignored~~.

	~~Since in~~ iterative matrix-diagonalization ~~schemes~~ eigenvectors close to the top		==== Electronic minimization ====
	of the calculated number of states converge much slower than the lowest		In the electronic minimization calculations, empty states do not contribute to the total energy, however, empty states are required to achieve a better convergence.
	eigenstates, it is important to choose a sufficiently large {{TAG\|NBANDS}}. An		In iterative matrix-diagonalization algorithms (see {{TAG\|ALGO}}) eigenvectors close to the top of the calculated number of states converge much slower than the lowest eigenstates, thus it is important to choose a sufficiently large {{TAG\|NBANDS}}. Therefore, we recommend using the default settings for {{TAG\|NBANDS}}, i.e. ''NELECT/2 + NIONS/2'', which is a safe choice in most cases. In some cases, it is also possible to decrease it to ''NELECT/2+NIONS/4'', however, in some transition metals with open ''f'' shells a much larger number of empty bands might be required (up to ''NELECT/2+2*NIONS''). To check this parameter perform several calculations for a fixed potential ({{TAG\|ICHARG}}=12) with an increasing number of bands (e.g. starting from ''NELECT/2 + NIONS/2''). An accuracy of <math> 10^{-6}</math> should be obtained in 10-15 iterations.
	~~insufficient number of empty bands might result in a significant performance~~		{{NB\|tip\|Note that the {{TAG\|RMM-DIIS}} scheme ({{TAG\|ALGO}}{{=}}Fast) is more sensitive to the number of bands than the default Davidson algorithm ({{TAG\|ALGO}}{{=}}Normal) and can require more bands for fast convergence.\|}}
	~~loss.~~ Therefore, we recommend ~~setting~~ {{TAG\|NBANDS}} ~~to the default value of~~
	''NELECT/2 + NIONS/2'', which is safe in most cases. In some cases, it is also
	possible to decrease ~~the number of additional bands~~ to ''NIONS/4'' ~~and improve~~
	~~the performance slightly. On the other hand~~, some transition metals ~~and atoms~~ with open
	''f'' shells ~~might require~~ a much larger number of empty bands (up to
	''2*NIONS'').
	To check this parameter perform several calculations for a fixed potential
	({{TAG\|ICHARG}}=12) with an increasing number of bands (e.g. starting from
	''NELECT/2 + NIONS/2''). An accuracy of <math> 10^{-6}</math> should be obtained
	in 10-15 iterations.

	The convergence of the calculations with a large number of empty states can be		==== Many-Body Perturbation Theory ====
	slow. In such cases, we recommend performing an exact diagonalization		In the Many-Body Perturbation Theory calculations ([[Practical guide to GW calculations\|''GW'']],[[RPA/ACFDT: Correlation energy in the Random Phase Approximation\|RPA]], and [[BSE calculations\|BSE]]), a large number of empty orbitals is usually required and can be much higher than the number of occupied states. The convergence of the calculations with a large number of empty states can be very slow. In such cases, we recommend performing an exact diagonalization ({{TAG\|ALGO}}=Exact) of the Hamiltonian with empty bands starting from a converged charge density.
	({{TAG\|ALGO}}=Exact) of the Hamiltonian with empty bands starting from a
	converged charge density.
	~~{{NB\|tip\|Note that the {{TAG\|RMM-DIIS}} scheme ({{TAG\|ALGO}}{{=}}Fast) is more sensitive to the number of bands than the default Davidson algorithm ({{TAG\|ALGO}}{{=}}Normal).\|}}~~

	==== Parallelization ====		==== Parallelization ====
	When executed on multiple CPUs, VASP automatically increases the number of bands, so that {{TAG\|NBANDS}} is divisible by the number of CPU cores.		When executed on multiple CPUs, VASP automatically increases the number of bands, so that {{TAG\|NBANDS}} is divisible by the number of CPU cores. If {{TAG\|NCORE}} > 1, {{TAG\|NBANDS}} is increased until it is divisible by the number of cores in a group ({{TAG\|NCORE}}). If {{TAG\|KPAR}} > 1, {{TAG\|NBANDS}} is increased until it is divisible by the number of cores in a group.
	If {{TAG\|NCORE}} > 1, {{TAG\|NBANDS}} is increased until it is divisible by the number of cores in a group ({{TAG\|NCORE}}).
	If {{TAG\|KPAR}} > 1, {{TAG\|NBANDS}} is increased until it is divisible by the number of cores in a group.

	==== Spin-polarized calculation ====		==== Spin-polarized calculation ====
	In the case of spin-polarized calculations, the default value for {{TAG\|NBANDS}} is		In the case of spin-polarized calculations, the default value for {{TAG\|NBANDS}} is increased to account for the initial magnetic moments.
	increased to account for the initial magnetic moments.

	==== Noncollinear calculation ====		==== Noncollinear calculation ====
	In noncollinear calculations, the default {{TAG\|NBANDS}} value is doubled to account for the		In noncollinear calculations, the default {{TAG\|NBANDS}} value is doubled to account for the spinors components.
	spinors components.
			== Related tags and article ==
			{{TAG\|NCORE}}, {{TAG\|NBANDS}}, {{TAG\|NBANDSGW}}, {{TAG\|NBANDSV}}, {{TAG\|NBANDSO}}, {{TAG\|NPAR}},{{TAG\|KPAR}}

	----		----
			[[The_VASP_Manual\|Contents]]

	[[Category:INCAR]][[Category:Electronic minimization]][[Category:Dielectric properties]]		[[Category:INCAR]][[Category:Electronic minimization]][[Category:Dielectric properties]]

Tal at 12:13, 7 April 2022

2022-04-07T12:13:11Z

@@ Line 1: / Line 1: @@
 {{TAGDEF|NBANDS|[integer]}}
-{{DEF|NBANDS|{{TAG|NELECT}}/2+NIONS/2|non-spinpolarized|0.6*{{TAG|NELECT}}+NMAG|spinpolarized}}
+{{DEF|NBANDS|max({{TAG|NELECT}}/2+NIONS/2,{{TAG|NELECT}}*0.6)|}}
-Description: {{TAG|NBANDS}} determines the actual number of bands in the calculation.
+Description: {{TAG|NBANDS}} specifies the number of bands in the calculation.
 ----
-One should choose {{TAG|NBANDS}} such that a considerable number of empty bands is included in the calculation. As a minimum we require one empty band. VASP will give a warning, if this is not the case.
+The right choice of {{TAG|NBANDS}} strongly depends on the type of the performed
-{{TAG|NBANDS}} is also important from a technical point of view: In iterative matrix-diagonalization schemes eigenvectors close to the top of the calculated number of vectors converge much slower than the lowest eigenvectors. This might result in a significant performance loss if not enough empty bands are included in the calculation. Therefore we recommend setting {{TAG|NBANDS}} to {{TAG|NELECT}}/2 + NIONS/2, this is also the default setting of the makeparam utility and of VASP.4.X. This setting is safe in most cases. In some cases, it is also possible to decrease the number of additional bands to NIONS/4 for large systems without performance loss, but on the other hand transition metals do require a much larger number of empty bands (up to 2*NIONS).
+Since in iterative matrix-diagonalization schemes eigenvectors close to the top
-To check this parameter perform several calculations for a fixed potential ({{TAG|ICHARG}}=12) with an increasing number of bands (e.g. starting from {{TAG|NELECT}}/2 + NIONS/2). An accuracy of should be obtained in 10-15 iterations. Mind that the RMM-DIIS scheme ({{TAG|IALGO}}=48) is more sensitive to the number of bands than the default CG algorithm ({{TAG|IALGO}}=38).
+The convergence of the calculations with a large number of empty states can be
 ----
 [[Category:INCAR]][[Category:Electronic minimization]][[Category:Dielectric properties]]

Huebsch at 14:31, 6 April 2022

2022-04-06T14:31:11Z

← Older revision		Revision as of 14:31, 6 April 2022
Line 6:		Line 6:
	One should choose {{TAG\|NBANDS}} such that a considerable number of empty bands is included in the calculation. As a minimum we require one empty band. VASP will give a warning, if this is not the case.		One should choose {{TAG\|NBANDS}} such that a considerable number of empty bands is included in the calculation. As a minimum we require one empty band. VASP will give a warning, if this is not the case.

	{{TAG\|NBANDS}} is also important from a technical point of view: In iterative matrix-diagonalization schemes eigenvectors close to the top of the calculated number of vectors converge much slower than the lowest eigenvectors. This might result in a significant performance loss if not enough empty bands are included in the calculation. Therefore we recommend ~~to set~~ {{TAG\|NBANDS}} to {{TAG\|NELECT}}/2 + NIONS/2, this is also the default setting of the makeparam utility and of VASP.4.X. This setting is safe in most cases. In some cases, it is also possible to decrease the number of additional bands to NIONS/4 for large systems without performance loss, but on the other hand transition metals do require a much larger number of empty bands (up to 2*NIONS).		{{TAG\|NBANDS}} is also important from a technical point of view: In iterative matrix-diagonalization schemes eigenvectors close to the top of the calculated number of vectors converge much slower than the lowest eigenvectors. This might result in a significant performance loss if not enough empty bands are included in the calculation. Therefore we recommend setting {{TAG\|NBANDS}} to {{TAG\|NELECT}}/2 + NIONS/2, this is also the default setting of the makeparam utility and of VASP.4.X. This setting is safe in most cases. In some cases, it is also possible to decrease the number of additional bands to NIONS/4 for large systems without performance loss, but on the other hand transition metals do require a much larger number of empty bands (up to 2*NIONS).

	To check this parameter perform several calculations for a fixed potential ({{TAG\|ICHARG}}=12) with an increasing number of bands (e.g. starting from {{TAG\|NELECT}}/2 + NIONS/2). An accuracy of should be obtained in 10-15 iterations. Mind that the RMM-DIIS scheme ({{TAG\|IALGO}}=48) is more ~~sensible~~ to the number of bands than the default CG algorithm ({{TAG\|IALGO}}=38).		To check this parameter perform several calculations for a fixed potential ({{TAG\|ICHARG}}=12) with an increasing number of bands (e.g. starting from {{TAG\|NELECT}}/2 + NIONS/2). An accuracy of should be obtained in 10-15 iterations. Mind that the RMM-DIIS scheme ({{TAG\|IALGO}}=48) is more sensitive to the number of bands than the default CG algorithm ({{TAG\|IALGO}}=38).

	{{sc\|NBANDS\|Examples\|Examples that use this tag}}		{{sc\|NBANDS\|Examples\|Examples that use this tag}}
Line 14:		Line 14:
	[[The_VASP_Manual\|Contents]]		[[The_VASP_Manual\|Contents]]

	[[Category:INCAR]][[Category:Electronic ~~Minimization~~]][[Category:Dielectric ~~Properties~~]]		[[Category:INCAR]][[Category:Electronic minimization]][[Category:Dielectric properties]]

Karsai at 10:28, 19 April 2019

2019-04-19T10:28:33Z

← Older revision		Revision as of 10:28, 19 April 2019
Line 14:		Line 14:
	[[The_VASP_Manual\|Contents]]		[[The_VASP_Manual\|Contents]]

	[[Category:INCAR]][[Category:Electronic Minimization]]		[[Category:INCAR]][[Category:Electronic Minimization]][[Category:Dielectric Properties]]

← Older revision		Revision as of 14:46, 8 April 2022
Line 27:		Line 27:

	----		----
	~~[[The_VASP_Manual\|Contents]]~~		[[Category:INCAR tag]][[Category:Electronic minimization]][[Category:Dielectric properties]]

	[[Category:INCAR]][[Category:Electronic minimization]][[Category:Dielectric properties]]