Page 1 of 1

No convergence of SCF Hartree Fock on Pt-specific surfaces

Posted: Wed May 08, 2024 7:23 am
by junweilucasbao
Hello,

I am running a series of calculations on Pt and Cu surfaces requiring HF density. Cu results are fine and run as expected, however, Pt energies become positive and diverge. I have optimized my surfaces with PBE-D3BJ and used the converged structures as my POSCAR with gamma KPOINTS, standard PAW-PBE and WAVECAR from PBE-D3BJ. The following are an example of my POSCAR and INCAR:

POSCAR
1.00000000000000
13.5825507782909423 -0.0000286735715211 0.0243690973237042
6.7912622303935422 11.7628352216583565 0.0334447326871308
0.0528407405015798 0.0537534515123181 26.0766653504449941
Pt C O
100 1 1
Selective dynamics
Direct
0.1256565183384240 0.3135983057400793 0.3171499509283535 F F F
0.2587130146578515 0.4461613198998839 0.4068593979325831 T T T
0.2587132300282893 0.0461611326032080 0.4068601931053661 T T T
0.1248346003783661 0.3112932151945988 0.5850367889025364 T T T
0.3256565399072215 0.3135983347388418 0.3171521016228169 F F F
0.0587128506823162 0.4461610621451871 0.4068591289474868 T T T
0.1917803154188479 0.5787300325629350 0.4953267660290470 T T T
0.0587131229046408 0.0461611500219690 0.4068592449770280 T T T
0.1917800517494419 0.1787309557818930 0.4953256325548373 T T T
0.3256568398612458 0.5135980753329719 0.3171533634110162 F F F
0.0587129785211858 0.2461620290603148 0.4068584454296771 T T T
0.1917800216721832 0.3787306340921368 0.4953258953967620 T T T
0.3248345218362086 0.5112929782572969 0.5850388816678243 T T T
0.1256566909705015 0.5135981066578097 0.3171535533987040 F F F
0.1256572023623690 0.1135978193140232 0.3171535283887152 F F F
0.2587121845781521 0.2461618730496352 0.4068576755608660 T T T
0.1248347760938800 0.5112928577241886 0.5850390761407571 T T T
0.1248343760897652 0.1112934119298534 0.5850371732185414 T T T
0.5256567049199532 0.3135982624344962 0.3171519234609335 F F F
0.3917794486725540 0.1787311580274203 0.4953257978071106 T T T
0.5248340297010384 0.3112933869559912 0.5850369693788386 T T T
0.7256570140144802 0.1135977540934690 0.3171538747127869 F F F
0.8587123938082840 0.2461619742167400 0.4068591893949044 T T T
0.9917802448487741 0.3787306325555291 0.4953268164387158 T T T
0.7248337137992102 0.1112938852243750 0.5850381317296560 T T T
0.4587128484074257 0.4461612825870906 0.4068593781781568 T T T
0.4587126976976073 0.0461612344485999 0.4068593710489272 T T T
0.5917798282635749 0.1787310345468479 0.4953258119298277 T T T
0.9256566196595912 0.1135978740812078 0.3171537520603067 F F F
0.3248341673501684 0.3112934359362499 0.5850380127942844 T T T
0.6587128746010197 0.2461618936777825 0.4068585545132268 T T T
0.7917797504736158 0.3787304844446092 0.4953275173919216 T T T
0.9248343315456463 0.1112933809266221 0.5850382127825305 T T T
0.3256568320170246 0.1135976869802064 0.3171526898300456 F F F
0.4587124749351335 0.2461619528836311 0.4068584900080261 T T T
0.5917801133082403 0.3787304868695649 0.4953265475324709 T T T
0.9256564244306773 0.3135986851229617 0.3171537318236375 F F F
0.3248343618625606 0.1112933943830953 0.5850377206677161 T T T
0.6587130536126187 0.4461609244437579 0.4068596664211932 T T T
0.6587130837009545 0.0461613665987129 0.4068604853300556 T T T
0.7917797304487887 0.1787310886442656 0.4953267863057250 T T T
0.9248344948434264 0.3112930518613115 0.5850381953395964 T T T
0.5256569922379413 0.5135978034496418 0.3171529566916433 F F F
0.5256569962509019 0.1135977994285042 0.3171535085341191 F F F
0.3917798604153527 0.3787306543046185 0.4953263821853482 T T T
0.5248341044534136 0.5112929634460608 0.5850388167909443 T T T
0.5248339511167894 0.1112935048745635 0.5850374813049110 T T T
0.7256568705662261 0.3135981612436751 0.3171514343755035 F F F
0.8587124738821874 0.4461617435489960 0.4068614216747427 T T T
0.8587126881525675 0.0461613523562018 0.4068609732390263 T T T
0.9917801439589827 0.1787308249990113 0.4953260245522753 T T T
0.7248339933182587 0.3112933155448063 0.5850386180868661 T T T
0.1256568856663947 0.7135980111226274 0.3171530927498196 F F F
0.3917799976514493 0.5787303171273144 0.4953272622766575 T T T
0.4587130597100218 0.8461613777271370 0.4068603848545147 T T T
0.1248345215427066 0.7112930926711507 0.5850369654472984 T T T
0.0587131491160586 0.8461613389107157 0.4068597966647247 T T T
0.1917801489725158 0.9787302786363660 0.4953267978719392 T T T
0.3256567582694245 0.7135979918280029 0.3171527833768693 F F F
0.3248345518323070 0.7112929915422034 0.5850388000713722 T T T
0.2587126927687909 0.8461612597577349 0.4068595513121498 T T T
0.3917802514248077 0.9787304098833587 0.4953273531885757 T T T
0.0587130607183136 0.6461610749487587 0.4068601149610894 T T T
0.1917802227882705 0.7787305618973631 0.4953268327485181 T T T
0.1256568883335589 0.9135978856793088 0.3171522432697742 F F F
0.1248344749309994 0.9112933123854525 0.5850387315081869 T T T
0.2587129753665894 0.6461611716516130 0.4068596504000020 T T T
0.3917800251222427 0.7787302293465697 0.4953268196015767 T T T
0.3256570439059061 0.9135978913707916 0.3171544900238317 F F F
0.3248342832171955 0.9112929437303438 0.5850387726529831 T T T
0.5256570909999922 0.7135980937205417 0.3171540387144987 F F F
0.5248343284948294 0.7112928835997446 0.5850379221370097 T T T
0.5917800037602736 0.9787304533530706 0.4953271595012264 T T T
0.7256569549310328 0.5135981526340103 0.3171550963696035 F F F
0.8587125867423795 0.6461612583266771 0.4068609757440638 T T T
0.9256566675828282 0.9135980424743764 0.3171547714103795 F F F
0.7248341662471876 0.5112927514903186 0.5850395972177599 T T T
0.7917798645825784 0.7787306279587507 0.4953281358250289 T T T
0.9248344857406161 0.9112932425951480 0.5850392567868401 T T T
0.5917801208404685 0.5787300490443431 0.4953271229342263 T T T
0.6587128776720853 0.8461612667158871 0.4068601652750775 T T T
0.7917799243624408 0.9787305628635838 0.4953283999613264 T T T
0.9256563429713367 0.5135982493934321 0.3171542579423061 F F F
0.9248347270186658 0.5112929739496691 0.5850397669261514 T T T
0.9917800012303680 0.7787306292835167 0.4953272216493600 T T T
0.4587128438718983 0.6461610291189941 0.4068592997917543 T T T
0.5256570009983719 0.9135977979257390 0.3171528974162925 F F F
0.5248343661096535 0.9112932093306321 0.5850386621747167 T T T
0.7917801051835769 0.5787304921028789 0.4953286553166883 T T T
0.7256568497277058 0.7135980601001037 0.3171534020855233 F F F
0.8587126465297762 0.8461616459192740 0.4068614453462541 T T T
0.9917804621124413 0.9787303542596576 0.4953273897919650 T T T
0.7248343426634226 0.7112931153903511 0.5850391975435176 T T T
0.6587130895065290 0.6461613669857461 0.4068608330299584 T T T
0.7256570283824146 0.9135979663804363 0.3171546872591762 F F F
0.5917802779012181 0.7787306220471010 0.4953274348985464 T T T
0.7248341625063320 0.9112932317642749 0.5850398514743940 T T T
0.9917803577581089 0.5787301391611289 0.4953279819807894 T T T
0.9256565352953743 0.7135982492406328 0.3171548750411830 F F F
0.9248344435900980 0.7112932009553093 0.5850395234977963 T T T
0.4888389137216351 0.4568493065947529 0.8320673817815070 T T T
0.4887292783559234 0.4564606295343310 0.8758953629566144 T T T

ENCUT = 660
ISMEAR = 1 # Gaussian Smearing
SIGMA = 0.2
EDIFF = 1E-8 # high accuracy for electronic groundstate
MAXMIX = 80
ALGO = Damped
TIME = 0.5
NELM = 300

ISPIN = 2
GGA = PE
LHFCALC = .TRUE.
AEXX=1.0
NCORE=40

VASP 5.4.4 and vasp 6.2.0 std and gam both fail.
Pt_pv POTCAR fails as well.
Adding/removing surface adsorbates do not help.
I have tried modifying ISPIN = 1, ALGO = FAST, NORMAL, , SIGMA=0.1,0.2,0.5,0.9, increase NBANDS, and AEXX=0.1,0.01,1.0 all with no
success. Here is an example of the results:

the WAVECAR file was read successfully
initial charge from wavefunction
entering main loop
N E dE d eps ncg rms
ort
gam= 0.000 g(H,U,f)= 0.448E+01 0.472E+01 0.943E+01 ort(H,U,f) = 0.000E+00
0.000E+00 0.000E+00
SDA: 1 -0.719716220672E+03 -0.71972E+03 -0.93131E+01 1017
0.186E+02 0.000E+00
gam= 0.333 g(H,U,f)= 0.235E+01 0.244E+01 0.206E+00 ort(H,U,f) =-0.393E+00
0.266E+01-0.605E+00
DMP: 2 -0.723368838684E+03 -0.36526E+01 -0.27712E+01 1017
0.499E+01 0.166E+01
gam= 0.333 g(H,U,f)= 0.925E+01 0.182E+01 0.828E+00 ort(H,U,f) =-0.310E+01
0.171E+01 0.470E+00
DMP: 3 -0.724576075046E+03 -0.12072E+01 -0.57959E+01 1017
0.119E+02-0.920E+00
gam= 0.333 g(H,U,f)= 0.621E+02 0.487E+02 0.231E+01 ort(H,U,f)
=-0.210E+02-0.248E+01-0.613E+00
DMP: 4 -0.721125522048E+03 0.34506E+01 -0.52564E+02 1017
0.113E+03-0.240E+02
gam= 0.333 g(H,U,f)= 0.665E+03 0.363E+04 0.112E+03 ort(H,U,f)
=-0.160E+03-0.233E+03-0.198E+02
DMP: 5 -0.591744001015E+03 0.12938E+03 -0.21330E+04 1017
0.440E+04-0.412E+03
gam= 0.333 g(H,U,f)= 0.275E+05 0.181E+06 0.816E+04 ort(H,U,f)
=-0.215E+04-0.189E+05-0.928E+03
DMP: 6 0.852297261240E+04 0.91147E+04 -0.10456E+06 1017
0.216E+06-0.220E+05
gam= 0.333 g(H,U,f)= 0.349E+07 0.537E+06 0.216E+05 ort(H,U,f) =
0.100E+05-0.601E+05-0.365E+04
DMP: 7 0.111798088353E+06 0.10328E+06 -0.20131E+07 1017
0.404E+07-0.537E+05
gam= 0.333 g(H,U,f)= 0.717E+07 0.153E+06 0.184E+03 ort(H,U,f)
=-0.271E+05-0.194E+05 0.755E+02
DMP: 8 0.150651532435E+06 0.38853E+05 -0.36565E+07 1017
0.733E+07-0.464E+05
gam= 0.333 g(H,U,f)= 0.140E+08 0.236E+06 0.268E+03 ort(H,U,f)
=-0.281E+06-0.419E+04 0.387E+03
DMP: 9 0.234605263918E+06 0.83954E+05 -0.70576E+07 1017
0.142E+08-0.285E+06
gam= 0.333 g(H,U,f)= 0.235E+08 0.201E+06 0.136E+04 ort(H,U,f)
=-0.898E+06-0.366E+05 0.189E+03
DMP: 10 0.281815597152E+06 0.47210E+05 -0.11688E+08 1017
0.237E+08-0.935E+06
gam= 0.333 g(H,U,f)= 0.358E+08 0.251E+06 0.129E+04 ort(H,U,f)
=-0.163E+07-0.550E+05-0.186E+04
DMP: 11 0.373343665976E+06 0.91528E+05 -0.17748E+08 1017
0.361E+08-0.169E+07

I suspect this is an issue with SCF on Pt specifically.
Thanks in advance for your help,
Elijah

Re: No convergence of SCF Hartree Fock on Pt-specific surfaces

Posted: Wed May 08, 2024 9:56 pm
by junweilucasbao
Here is the ZIP file:
scratch.zip

Re: No convergence of SCF Hartree Fock on Pt-specific surfaces

Posted: Mon May 13, 2024 8:07 am
by jonathan_lahnsteiner2
Dear junweilucasbao,

I was checking your input files. Now it would be very helpful to know if you also get convergence issues when switching off the Hartree-Fock part of your calculation
by commenting

Code: Select all

  
#LHFCALC=.TRUE.
#AEXX=0.01
in your INCAR file. If the problem remains this will speed up the calculation and testing is faster for you.
To resolve the convergence issue I would try to use a different algorithm for electronic minimization. You can do this by setting the tag ALGO. I would suggest to try with ALGO=ALL.

I will move the post to the using-vasp forum because right now it does not look like there is a bug in the code.


All the best Jonathan

Re: No convergence of SCF Hartree Fock on Pt-specific surfaces

Posted: Thu May 16, 2024 1:49 pm
by junweilucasbao
Jonathan,

Switching off LHFCALC and running with ALGO=normal/all/fast all behave properly. Setting ALGO=ALL with LHFCALC=.TRUE. and AEXX=1.0/0.5 still does not give convergence for this system (Results are similar to previously reported non-convergence). Do you have any other suggestions? Generally, I have been first running PBE as an initial guess wavefunction for the Hartree-Fock calculation. I only have issues converging the latter with any ALGO setting I have tried previously.

Thank you very much!

Re: No convergence of SCF Hartree Fock on Pt-specific surfaces

Posted: Tue May 21, 2024 11:18 am
by jonathan_lahnsteiner2
Dear Elijah,

You could try to adjust the updating of the charge density during the optimization. You can achieve this by adjusting the mixing parameters AMIX, BMIX. You can also choose a mixing method with IMIX. By adjusting these parameters to the needs of the considered system you should be able to improve the electronic convergence.


All the best Jonathan

Re: No convergence of SCF Hartree Fock on Pt-specific surfaces

Posted: Tue May 21, 2024 3:44 pm
by alex
Hello Elijah,

some remarks about your simulation as is:
a) your CO molecule is about 7\AA away from the surface, so you PBE+D calculation went most likely wrong. I'm guessing, that's _not_ what you are after.
b) your Pt-Pt distance is a bit smaller than the one in Materialsproject (2.72 vs 2.79 \AA). This does not sound like much, _but_ underestimating bond lengths and having convergence troubles very often are twins.
c) a more general remark: HF is made for localized stuff like molecules and always has a band gap. So it's qualitatively wrong to even mix partially HF contributions into your energy. However, it's your choice.

If you like a converged result: put the CO in a proper position, forget about the PBE pre-optimization, extend the starting Pt-Pt bond lengths to the equilibrium distance of your method, e.g., from a bulk optimization which you do before. Build your slab from there. Use a smaller cell for testing.

Good luck!

alex