sleazy optimization to save time

Queries about input and output files, running specific calculations, etc.


Moderators: Global Moderator, Moderator

Post Reply
Message
Author
jasius
Newbie
Newbie
Posts: 11
Joined: Thu Nov 28, 2013 3:25 pm
License Nr.: 1266

sleazy optimization to save time

#1 Post by jasius » Thu Dec 05, 2013 2:52 pm

Dear all,

I am running big zeolite (289 atoms) periodic cell optimizing only atom positions and was wondering which parameters can I relax so it crudely optimizes for a following restart with "tight" optimization criteria criteria? I am sure I would save some time this way, I am not sure which parameters can be relaxed. I am running gamma point PAW with the following.

Also, do I need smearing in situations that contain closed shell zeolite with adsorbed molecules such as simple hydrocarbons

thanks much

ISTART = 0 ; ICHARG=2
ENCUT = 400
PREC = NORMAL

smearing:
ISMEAR = 0 ; SIGMA = 0.1

dynamic:
IBRION = 1
NSW = 200
POTIM = 0.2
EDIFFG = -0.05

thanks

Jonas
Last edited by jasius on Thu Dec 05, 2013 2:52 pm, edited 1 time in total.

jasius
Newbie
Newbie
Posts: 11
Joined: Thu Nov 28, 2013 3:25 pm
License Nr.: 1266

sleazy optimization to save time

#2 Post by jasius » Thu Dec 05, 2013 10:45 pm

as a note PREC = NORMAL would result in wild swings in SCF convergence so I increased to ACCURATE thus converging but also increasing the time, opposite of what I am lookign for
Last edited by jasius on Thu Dec 05, 2013 10:45 pm, edited 1 time in total.

alex
Hero Member
Hero Member
Posts: 586
Joined: Tue Nov 16, 2004 2:21 pm
License Nr.: 5-67
Location: Germany

sleazy optimization to save time

#3 Post by alex » Fri Dec 06, 2013 9:09 am

Hi Jonas,

what do you need the charge for? It should be neutral.

Smearing is fine, it improves convergence. Stick to it. But pay attention to the entropy you'll introduce by partial occupations (you need to check the DOS if it looks alright).

I would also go for ALGO = V

Cheers,

alex
Last edited by alex on Fri Dec 06, 2013 9:09 am, edited 1 time in total.

jasius
Newbie
Newbie
Posts: 11
Joined: Thu Nov 28, 2013 3:25 pm
License Nr.: 1266

sleazy optimization to save time

#4 Post by jasius » Fri Dec 06, 2013 10:01 am

not sure why I have charge 2. I am in a learning phase and doing some things mechanically. I will switch to ALGO=V, ICHARG=0. but what is a quick and dirty way of optimizing? That is, I would like to optimize with some sleazy criteria and then restart from that structure to do accurate optimization. What should I lower? Perhaps ENCUT?
Last edited by jasius on Fri Dec 06, 2013 10:01 am, edited 1 time in total.

alex
Hero Member
Hero Member
Posts: 586
Joined: Tue Nov 16, 2004 2:21 pm
License Nr.: 5-67
Location: Germany

sleazy optimization to save time

#5 Post by alex » Fri Dec 06, 2013 3:33 pm

Ehm, sorry, I mixed ICHARG with sth. else.

If you want to do it sloppy, use O_s in the POTCAR for O and go to 250 eV as well as PREC = Low and ALGO V.
For more precision switch to your first setup.
Oxygen atoms are usually hardest to describe, so it's safe to leave the other pseudo potentials as is.

Cheers,

alex
Last edited by alex on Fri Dec 06, 2013 3:33 pm, edited 1 time in total.

jasius
Newbie
Newbie
Posts: 11
Joined: Thu Nov 28, 2013 3:25 pm
License Nr.: 1266

sleazy optimization to save time

#6 Post by jasius » Fri Dec 06, 2013 11:28 pm

Alex, one more question, it is converging, but doing it in 40+ iterations and it looks more auspicious than systematic. Should it do that with PEC=LOW?
general:
SYSTEM = ZSM-H+CH3OH opt
ISTART = 0 ; ICHARG=2
ENCUT = 300
PREC = LOW
ALGO = V
NELM = 200
NELMIN = 10

smearing:
ISMEAR = 0 ; SIGMA = 0.1

dynamic:
IBRION = 1
NSW = 400
POTIM = 0.2
EDIFFG = -0.04

N E dE d eps ncg rms rms(c)
RMM: 1 -0.234257348188E+04 -0.23426E+04 -0.32280E+02 1872 0.139E+01 0.444E+00
RMM: 2 -0.233926665715E+04 0.33068E+01 -0.47955E+00 1872 0.156E+00 0.264E+00
RMM: 3 -0.233912954018E+04 0.13712E+00 -0.17187E-01 1871 0.299E-01 0.164E+00
RMM: 4 -0.233908405151E+04 0.45489E-01 -0.60283E-01 1872 0.479E-01 0.817E-01
RMM: 5 -0.233906491786E+04 0.19134E-01 -0.33243E-02 1803 0.133E-01 0.277E-01
RMM: 6 -0.233906105036E+04 0.38675E-02 0.40900E-03 1805 0.806E-02 0.863E-02
RMM: 7 -0.233906232527E+04 -0.12749E-02 -0.18045E-02 1732 0.450E-02 0.984E-02
RMM: 8 -0.233906198146E+04 0.34381E-03 -0.99953E-03 1675 0.426E-02 0.644E-02
RMM: 9 -0.233906314471E+04 -0.11632E-02 -0.23799E-02 1439 0.260E-02 0.162E-02
RMM: 10 -0.233906286508E+04 0.27963E-03 -0.18712E-02 1179 0.194E-02 0.776E-03
RMM: 11 -0.233906152426E+04 0.13408E-02 -0.52937E-03 1050 0.126E-02 0.405E-03
RMM: 12 -0.233906162425E+04 -0.99992E-04 -0.32133E-03 1106 0.163E-02 0.291E-03
RMM: 13 -0.233906247219E+04 -0.84794E-03 -0.52050E-03 993 0.128E-02 0.210E-03
RMM: 14 -0.233906194546E+04 0.52673E-03 0.12258E-04 919 0.978E-03 0.133E-03
RMM: 15 -0.233906243626E+04 -0.49080E-03 -0.83586E-03 806 0.689E-03 0.754E-04
RMM: 16 -0.233906223753E+04 0.19874E-03 -0.57833E-03 755 0.492E-03 0.244E-04
RMM: 17 -0.233906226996E+04 -0.32435E-04 -0.93021E-03 930 0.149E-02 0.449E-04
RMM: 18 -0.233906222897E+04 0.40987E-04 -0.68394E-03 965 0.157E-02 0.454E-04
RMM: 19 -0.233906188731E+04 0.34166E-03 -0.89959E-04 888 0.113E-02 0.457E-04
RMM: 20 -0.233906174617E+04 0.14114E-03 -0.14606E-03 824 0.940E-03 0.487E-04
RMM: 21 -0.233906162001E+04 0.12617E-03 -0.65952E-05 773 0.597E-03 0.511E-04
RMM: 22 -0.233906283217E+04 -0.12122E-02 -0.12537E-02 900 0.122E-02 0.549E-04
RMM: 23 -0.233906187803E+04 0.95414E-03 0.12215E-03 939 0.152E-02 0.583E-04
RMM: 24 -0.233906241255E+04 -0.53453E-03 -0.67361E-03 918 0.126E-02 0.600E-04
RMM: 25 -0.233906239435E+04 0.18202E-04 -0.83307E-03 834 0.893E-03 0.612E-04
RMM: 26 -0.233906201161E+04 0.38274E-03 -0.41917E-03 765 0.601E-03 0.614E-04
RMM: 27 -0.233906222502E+04 -0.21341E-03 -0.10594E-02 907 0.144E-02 0.993E-03
RMM: 28 -0.233906148838E+04 0.73664E-03 -0.52038E-04 1020 0.133E-02 0.581E-03
RMM: 29 -0.233906213448E+04 -0.64610E-03 -0.66725E-03 942 0.122E-02 0.556E-03
RMM: 30 -0.233906193197E+04 0.20252E-03 -0.74173E-03 873 0.105E-02 0.552E-03
RMM: 31 -0.233906176882E+04 0.16315E-03 -0.62122E-03 814 0.818E-03 0.552E-03
RMM: 32 -0.233906160446E+04 0.16436E-03 -0.27737E-03 864 0.104E-02 0.213E-02
RMM: 33 -0.233906009566E+04 0.15088E-02 0.11953E-02 1041 0.158E-02 0.882E-03
RMM: 34 -0.233906086489E+04 -0.76923E-03 0.67899E-03 960 0.129E-02 0.786E-03
RMM: 35 -0.233906078236E+04 0.82529E-04 0.69515E-03 861 0.824E-03 0.779E-03
RMM: 36 -0.233906157165E+04 -0.78929E-03 -0.25317E-04 778 0.638E-03 0.777E-03
RMM: 37 -0.233906205464E+04 -0.48300E-03 -0.79964E-03 867 0.123E-02 0.100E-02
RMM: 38 -0.233906057999E+04 0.14747E-02 0.13978E-02 937 0.119E-02 0.238E-03
RMM: 39 -0.233906207591E+04 -0.14959E-02 -0.28234E-03 900 0.128E-02 0.138E-03
RMM: 40 -0.233906128563E+04 0.79028E-03 0.48583E-03 873 0.127E-02 0.135E-03
RMM: 41 -0.233906179311E+04 -0.50748E-03 0.38904E-04 815 0.102E-02 0.134E-03
RMM: 42 -0.233906239841E+04 -0.60530E-03 -0.12770E-02 903 0.146E-02 0.128E-02
RMM: 43 -0.233905947739E+04 0.29210E-02 0.18423E-02 968 0.144E-02 0.809E-03
RMM: 44 -0.233906170929E+04 -0.22319E-02 -0.85294E-04 897 0.125E-02 0.781E-03
RMM: 45 -0.233906137445E+04 0.33484E-03 0.15459E-03 813 0.841E-03 0.778E-03
RMM: 46 -0.233906156848E+04 -0.19403E-03 -0.25520E-04 736 0.584E-03 0.777E-03
RMM: 47 -0.233906151508E+04 0.53399E-04 0.15708E-05 702 0.444E-03
3 F= -.23390615E+04 E0= -.23390615E+04 d E =-.338819E+01
BRION: g(F)= 0.228E+01 g(S)= 0.000E+00 retain N= 2 mean eig= 0.64
eig: 0.645 0.645
bond charge predicted
Last edited by jasius on Fri Dec 06, 2013 11:28 pm, edited 1 time in total.

alex
Hero Member
Hero Member
Posts: 586
Joined: Tue Nov 16, 2004 2:21 pm
License Nr.: 5-67
Location: Germany

sleazy optimization to save time

#7 Post by alex » Mon Dec 09, 2013 10:50 am

Hi jasius,

the challenge for the minimiser is the guess wavefunction. It's probably electron gas. So - far away from bonds as in zeolites.

On the other hand: is this your third iteration? 40 cycles are ok for the very first step, but then it should drop to twenty fastly. After about 5 cylcles I end up with below ten for that kind of system. Maybe your structure is far off from the equilibrium?!

And go to 250 eV. You'll gain nothing with 300eV for starters.

Cheers,

alex
Last edited by alex on Mon Dec 09, 2013 10:50 am, edited 1 time in total.

Post Reply