IALGO=54 (damped MD, velocity quenched or quickmin) of a H2O molecule

Problems running VASP: crashes, internal errors, "wrong" results.


Moderators: Global Moderator, Moderator

Post Reply
Message
Author
jonghyun_jung
Newbie
Newbie
Posts: 1
Joined: Wed Jan 13, 2021 1:36 pm

IALGO=54 (damped MD, velocity quenched or quickmin) of a H2O molecule

#1 Post by jonghyun_jung » Thu Sep 25, 2025 11:35 am

Dear whom it may concern,

Thank you very much always for your help!

First I tried to converge r2SCAN in a 800 atom system (Al2O3-SiO2 pore) using ALGO=All and TIME=0.05. A slow convergence was observed, probably due to a large TIME, that is, the energy increases and decreases and fluctuates.

For a test calculation, I tested a H2O molecule in vacuum using ALGO=All and IALGO=54 (damped MD, velocity quenched or quickmin), which is less sensitive to TIME according to the vasp wiki. The energy did not increase much but decreased as expected. Interestingly the convered forces differ by an order of 0.1 eV/Ang. Is it okay? For the original 800 atom case, I'll try TIME=0.01, but this post is for others who may use IALGO=54.

Below are input files:

POSCAR

Code: Select all

O  H
 1.0000000000000000
    13.0000000000000000    0.0000000000000000    0.0000000000000000
     0.0000000000000000   13.0000000000000000    0.0000000000000000
     0.0000000000000000    0.0000000000000000   13.0000000000000000
 O   H
   1   2
Cartesian
  4.0000000000000000  6.0000000000000000  6.3975400000000002
  4.0000000000000000  6.7632399999999997  5.8012300000000003
  4.0000000000000000  5.2367600000000003  5.8012300000000003

POTCARS:
POTCAR: PAW_PBE O 08Apr2002
POTCAR: PAW_PBE H 15Jun2001

1) ALGO=All
INCAR

Code: Select all

ISTART = 0
SYSTEM = H2O
ENCUT = 600 ! 1.5 ENMAX of O ( 400 )
PREC = Accurate
ALGO = All  # 700 atoms, r2SCAN
TIME = 0.05
NELM = 400  # max electronic steps. slow convergence

GGA_COMPAT = F
LREAL = A
METAGGA = R2SCAN
LASPH   = T

ISMEAR = 0 ! Gaussian
SIGMA = 0.1 ! eV
KSPACING = 0.52  # a < 12 Ang: 2x2x2, otherwise gamma. For > 200 kpt atom.
NCORE = 18
ISIF = 2   ! calculate stress, relax ions, keep cell
LWAVE = F
LCHARG = F
NWRITE = 0  # For a large job, save space. Force at the first ionic step.

2) IALGO=54 (damped MD, velocity quenched or quickmin) : only the ALGO tag was removed and the IALGO tag was included.

Code: Select all

ISTART = 0
SYSTEM = H2O
ENCUT = 600 ! 1.5 ENMAX of O ( 400 )
PREC = Accurate
IALGO = 54  # quick min
TIME = 0.05
NELM = 400  # max electronic steps. slow convergence

GGA_COMPAT = F
LREAL = A
METAGGA = R2SCAN
LASPH   = T

ISMEAR = 0 ! Gaussian
SIGMA = 0.1 ! eV
KSPACING = 0.52  # a < 12 Ang: 2x2x2, otherwise gamma. For > 200 kpt atom.
NCORE = 18
ISIF = 2   ! calculate stress, relax ions, keep cell
LWAVE = F
LCHARG = F
NWRITE = 0  # For a large job, save space. Force at the first ionic step.

Below are converged forces: see the z component of the 1st atom changed from -0.365 to -0.44.
1) ALGO=All

Code: Select all

 POSITION                                       TOTAL-FORCE (eV/Angst)
 -----------------------------------------------------------------------------------
      4.00000      6.00000      6.39754        -0.000000      0.000000     -0.365369
      4.00000      6.76324      5.80123         0.000000     -0.159017      0.182685
      4.00000      5.23676      5.80123        -0.000000      0.159017      0.182685
 -----------------------------------------------------------------------------------
    total drift:                                0.000010     -0.000010     -0.001770

2) IALGO=54 (damped MD)

Code: Select all

 POSITION                                       TOTAL-FORCE (eV/Angst)
 -----------------------------------------------------------------------------------
      4.00000      6.00000      6.39754         0.000000     -0.000000     -0.441665
      4.00000      6.76324      5.80123        -0.000000     -0.159572      0.220833
      4.00000      5.23676      5.80123        -0.000000      0.159572      0.220833
 -----------------------------------------------------------------------------------
    total drift:                                0.000010     -0.000010     -0.135618

Additionally, these are OSZICAR's, and here we see that energy decreases for the both cases:
1) ALGO=All

Code: Select all

       N       E                     dE             d eps       ncg     rms          ort
DAV:   1     0.581741031911E+02    0.58174E+02   -0.18822E+03     8   0.501E+02
DAV:   2    -0.113070093572E+02   -0.69481E+02   -0.69481E+02     8   0.127E+02
DAV:   3    -0.197597343557E+02   -0.84527E+01   -0.84527E+01     8   0.620E+01
DAV:   4    -0.198934556826E+02   -0.13372E+00   -0.13372E+00     8   0.854E+00
DAV:   5    -0.198935847656E+02   -0.12908E-03   -0.12908E-03     8   0.260E-01
SDA:   6    -0.129442185079E+02    0.69494E+01   -0.12908E-03     8   0.161E+02 0.000E+00
CGA:   7    -0.148101982169E+02   -0.18660E+01   -0.16011E-04     8   0.285E+01-0.515E+01
CGA:   8    -0.151539219623E+02   -0.34372E+00   -0.32281E+00     8   0.372E+00 0.159E+00
CGA:   9    -0.152347824471E+02   -0.80860E-01   -0.80384E-01     8   0.163E+00 0.144E-02
CGA:  10    -0.152961538547E+02   -0.61371E-01   -0.28828E-02     8   0.128E+00 0.204E-04
CGA:  11    -0.153278318929E+02   -0.31678E-01   -0.31096E-01     8   0.267E-01 0.196E-02
CGA:  12    -0.153340004275E+02   -0.61685E-02   -0.61703E-02     8   0.665E-02-0.291E-04
CGA:  13    -0.153358452092E+02   -0.18448E-02   -0.18457E-02     8   0.215E-02-0.425E-05
CGA:  14    -0.153364554424E+02   -0.61023E-03   -0.60955E-03     8   0.560E-03 0.683E-05
CGA:  15    -0.153366075001E+02   -0.15206E-03   -0.15205E-03     8   0.143E-03 0.385E-06
CGA:  16    -0.153366552644E+02   -0.47764E-04   -0.47904E-04     8   0.531E-04-0.799E-06
   1 F= -.15336655E+02 E0= -.15336655E+02  d E =-.244881E-12

2) IALGO=54 (damped MD)

Code: Select all

       N       E                     dE             d eps       ncg     rms          ort
DAV:   1     0.155601673409E+03    0.15560E+03   -0.89953E+02     8   0.501E+02
DAV:   2     0.916370191975E+02   -0.63965E+02   -0.63964E+02     8   0.135E+02
DAV:   3     0.474284007478E+02   -0.44209E+02   -0.44207E+02     8   0.111E+02
DAV:   4     0.194504862141E+01   -0.45483E+02   -0.45482E+02     8   0.191E+02
DAV:   5    -0.138562881141E+02   -0.15801E+02   -0.15801E+02     8   0.120E+02
SDA:   6    -0.894666041969E+01    0.49096E+01   -0.11280E+01     8   0.226E+02 0.000E+00
QMN:   7    -0.100157802299E+02   -0.10691E+01   -0.18714E+01     8   0.172E+02 0.202E+02
QMN:   8    -0.116790697546E+02   -0.16633E+01   -0.19696E+01     8   0.104E+02 0.290E+02
QMN:   9    -0.132570986338E+02   -0.15780E+01   -0.14367E+01     8   0.524E+01 0.235E+02
QMN:  10    -0.141574985973E+02   -0.90040E+00   -0.54949E+00     8   0.405E+01 0.694E+01
QMN:  11    -0.141277629552E+02    0.29736E-01   -0.35276E+00     8   0.706E+01-0.126E+02
QMN:  12    -0.144492290462E+02   -0.32147E+00   -0.52224E+00     8   0.464E+01 0.581E+01
QMN:  13    -0.148746166428E+02   -0.42539E+00   -0.42407E+00     8   0.188E+01 0.660E+01
QMN:  14    -0.151519863679E+02   -0.27737E+00   -0.15287E+00     8   0.371E+00 0.269E+01
QMN:  15    -0.151536048171E+02   -0.16184E-02   -0.31563E-01     8   0.631E+00-0.289E+01
QMN:  16    -0.151831814476E+02   -0.29577E-01   -0.51005E-01     8   0.468E+00 0.552E+00
QMN:  17    -0.152275151193E+02   -0.44334E-01   -0.50476E-01     8   0.258E+00 0.752E+00
QMN:  18    -0.152665018838E+02   -0.38987E-01   -0.32585E-01     8   0.106E+00 0.546E+00
QMN:  19    -0.152850062681E+02   -0.18504E-01   -0.80012E-02     8   0.786E-01 0.814E-01
QMN:  20    -0.152797575114E+02    0.52488E-02   -0.89978E-02     8   0.180E+00-0.379E+00
QMN:  21    -0.152882074993E+02   -0.84500E-02   -0.14678E-01     8   0.136E+00 0.158E+00
QMN:  22    -0.153010398416E+02   -0.12832E-01   -0.14845E-01     8   0.769E-01 0.220E+00
QMN:  23    -0.153127032442E+02   -0.11663E-01   -0.10169E-01     8   0.327E-01 0.171E+00
QMN:  24    -0.153189388447E+02   -0.62356E-02   -0.34587E-02     8   0.207E-01 0.485E-01
QMN:  25    -0.153186367940E+02    0.30205E-03   -0.22117E-02     8   0.442E-01-0.768E-01
QMN:  26    -0.153207279267E+02   -0.20911E-02   -0.37232E-02     8   0.351E-01 0.394E-01
QMN:  27    -0.153240314024E+02   -0.33035E-02   -0.39743E-02     8   0.218E-01 0.577E-01
QMN:  28    -0.153272562413E+02   -0.32248E-02   -0.30319E-02     8   0.111E-01 0.495E-01
QMN:  29    -0.153293359807E+02   -0.20797E-02   -0.15293E-02     8   0.803E-02 0.226E-01
QMN:  30    -0.153299463725E+02   -0.61039E-03   -0.70096E-03     8   0.140E-01-0.612E-02
QMN:  31    -0.153306099841E+02   -0.66361E-03   -0.11803E-02     8   0.111E-01 0.125E-01
QMN:  32    -0.153316618785E+02   -0.10519E-02   -0.12795E-02     8   0.711E-02 0.185E-01
QMN:  33    -0.153327160560E+02   -0.10542E-02   -0.10379E-02     8   0.408E-02 0.167E-01
QMN:  34    -0.153334735810E+02   -0.75753E-03   -0.66105E-03     8   0.343E-02 0.979E-02
QMN:  35    -0.153338681101E+02   -0.39453E-03   -0.40987E-03     8   0.520E-02 0.299E-02
QMN:  36    -0.153340733798E+02   -0.20527E-03   -0.42142E-03     8   0.782E-02 0.612E-03
QMN:  37    -0.153343491220E+02   -0.27574E-03   -0.62748E-03     8   0.927E-02 0.328E-02
QMN:  38    -0.153348443286E+02   -0.49521E-03   -0.81438E-03     8   0.841E-02 0.788E-02
QMN:  39    -0.153354837923E+02   -0.63946E-03   -0.76865E-03     8   0.565E-02 0.973E-02
QMN:  40    -0.153360039924E+02   -0.52020E-03   -0.41322E-03     8   0.264E-02 0.562E-02
QMN:  41    -0.153361048646E+02   -0.10087E-03   -0.53433E-04     8   0.107E-02-0.422E-02
QMN:  42    -0.153361563750E+02   -0.51510E-04   -0.95742E-04     8   0.922E-03 0.993E-03
   1 F= -.15336156E+02 E0= -.15336156E+02  d E =-.333902E-12

I do not know whether a r2SCAN calculation always should start from reading WAVECAR of converged PBE and should not start from scratch.

Thank you very much!


alexey.tal
Global Moderator
Global Moderator
Posts: 510
Joined: Mon Sep 13, 2021 12:45 pm

Re: IALGO=54 (damped MD, velocity quenched or quickmin) of a H2O molecule

#2 Post by alexey.tal » Thu Sep 25, 2025 2:11 pm

Dear jonghyun_jung,

Thank you for your question.

It looks to me that this issue could be related to the electronic convergence. Have you tried using a more stringent convergence criterion? I would suggest that you try EDIFF = 1E-6 or maybe even lower.

Best wishes,
Alexey


jonghyun_jung
Newbie
Newbie
Posts: 1
Joined: Wed Jan 13, 2021 1:36 pm

Re: IALGO=54 (damped MD, velocity quenched or quickmin) of a H2O molecule

#3 Post by jonghyun_jung » Thu Sep 25, 2025 2:42 pm

Dear Alexey,

Thank you very for your help!

Reducing EDIFF to 1e-6 reduced the force difference to within 0.01 eV/Ang.

1) Algo=All, EDIFF=1e-6:
POSITION TOTAL-FORCE (eV/Angst)
-----------------------------------------------------------------------------------
4.00000 6.00000 6.39754 -0.000000 0.000000 -0.365353
4.00000 6.76324 5.80123 0.000000 -0.158920 0.182676
4.00000 5.23676 5.80123 -0.000000 0.158920 0.182676
-----------------------------------------------------------------------------------
total drift: 0.000010 -0.000010 -0.001065

2) IALGO=54, EDIFF=1e-6:
POSITION TOTAL-FORCE (eV/Angst)
-----------------------------------------------------------------------------------
4.00000 6.00000 6.39754 0.000000 0.000000 -0.372137
4.00000 6.76324 5.80123 -0.000000 -0.161975 0.186069
4.00000 5.23676 5.80123 -0.000000 0.161975 0.186069
-----------------------------------------------------------------------------------
total drift: 0.000010 -0.000010 -0.004732

Thank you very much!

Last edited by jonghyun_jung on Thu Sep 25, 2025 2:55 pm, edited 1 time in total.

Post Reply