Problem with geometry optimization

Top Page > Browsing

Problem with geometry optimization

Date: 2023/11/21 20:25
Name: Lim: Dear experts,

I have a question regarding the use of geometry optimization.
I'm trying to use geometry optimization for a Cu primitive cell.
After completing the full relaxation, I plan to calculate the band structure based on the results.
I am trying to perform relaxation on both atom and cells using the OptC5 method, but I am encountering a problem.
I am doing the relaxation of a simple Cu primitive cell with a single atom, but the relaxation is not converging well. Despite using a sufficient number of cores and processes, I am experiencing a shortage of memory.
I used the Materials Project to obtain the initial coordinates and initial unit cell data.
What parameters should I change to achieve better convergence of geometry optimization

Below is my input data.

#
# The file was generated by the OpenMX Viewer

#

System.CurrrentDirectory ./
System.Name Cu_primitive
level.of.stdout 1
level.of.fileout 1

Species.Number 1
<Definition.of.Atomic.Species
Cu Cu6.0H-s3p2d1 Cu_PBE19H
Definition.of.Atomic.Species>

Atoms.Number 1
Atoms.SpeciesAndCoordinates.Unit FRAC
<Atoms.SpeciesAndCoordinates
1 Cu 0.0000000 0.0000000 0.0000000 9.5 9.5
Atoms.SpeciesAndCoordinates>

Atoms.UnitVectors.Unit Ang
<Atoms.UnitVectors
2.5606189 0.0000000 0.0000000
1.2803095 2.2175610 0.0000000
1.2803095 0.7391870 2.0907366
Atoms.UnitVectors>

scf.XcType GGA-PBE
scf.SpinPolarization off
scf.ElectronicTemperature 300.0
scf.energycutoff 220.0
scf.maxIter 100
scf.EigenvalueSolver band
scf.Kgrid 9 9 9
scf.Mixing.Type rmm-diisk
scf.Init.Mixing.Weight 0.05
scf.Min.Mixing.Weight 0.01
scf.Max.Mixing.Weight 0.30
scf.Mixing.History 25
scf.Mixing.StartPulay 15
scf.criterion 1.0e-7 # default=1.0e-6 (Hartree)

MD.Type OptC5 # Nomd|Opt|DIIS|NVE|NVT_VS|NVT_NH
MD.Opt.DIIS.History 10 # default=7
MD.Opt.StartDIIS 10 # default=5
MD.Opt.EveryDIIS 10000 # default=10
MD.maxIter 400 # default=1
MD.Opt.criterion 0.00003 # default=1.0e-4 (Hartree/bohr)
<MD.Fixed.XYZ
1 1 1 1
MD.Fixed.XYZ>

#MD.Type OptC2
#MD.maxIter 300
#MD.TimeStep 1.0
#MD.Opt.criterion 0.0003

Dos.fileout off
Dos.Erange -25.0 20.0
Dos.Kgrid 8 8 8

Band.dispersion off
Band.Nkpath 5
<Band.kpath
40 0.000000 0.000000 0.000000 0.500000 0.000000 0.500000 G X
40 0.500000 0.000000 0.500000 0.500000 0.250000 0.750000 X W
40 0.500000 0.250000 0.750000 0.500000 0.500000 0.500000 W L
40 0.500000 0.500000 0.500000 0.000000 0.000000 0.000000 L G
40 0.000000 0.000000 0.000000 0.375000 0.375000 0.750000 G K
Band.kpath>

Page: [1]

Re: Problem with geometry optimization ( No.1 )

Date: 2023/11/21 20:41
Name: Masanobu Miyata <m-miyata@jaist.ac.jp>

Hi.

The crystal system of pure Cu is a face-centered cubic, indicating that the primitive lattice vector should be constrained under a = b = c. Therefore, you should not use OptC5(without constrain) but OptC3(a = b = c).

Best regards
Miyata.

Re: Problem with geometry optimization ( No.2 )

Date: 2023/11/22 11:17
Name: Lim

Thank you for your response.

I believe that OptC3 involves fixing the symmetry during optimization.
However, I think that if full relaxation like OptC5 also converges, it can yield the same results.
I believe that performing a full relaxation during DFT calculations for optimization is more reliable, which is why I chose to use OptC5.
I have observed that full relaxation for primitive cells of other metals like Ru and Mo converges in less than a minute, but I am curious why the Cu primitive cell takes much longer and does not converge as easily

Could you provide any tips on how to ensure good convergence through full relaxation optimization?

Re: Problem with geometry optimization ( No.3 )

Date: 2023/11/26 01:16
Name: Lim

When I perform OptC5 optimization on the Cu primitive cell, the calculation does not proceed after the phrase 'Invalid access in truncation.c' appears in the .out file. What could be the reason for this?

Page: [1]