My Community

Dear all,

I am running GW calculations on a slab with benzene molecule on top. On our sludter we have a time limit of 24 hours. After that the job is killed. I wanted to ask you how can I restart my GW calculations. I am at the same time learning the theory behind thsi method so i do not really know much about the meanings of the terms here.

Here is the INCAR:
_________________________
ALGO = SCGW0
ENCUTGW = 150
NOMEGA = 72
ISMEAR = -5
ISPIN = 2
SIGMA = 0.01
KPAR = 4
NBANDS = 144
LREAL = False
MAXMEM = 13500
NELM = 4
_________________________


And my job folder has several new files of the kind.
___________________________________________________________________________
-rw-rw---- 1 ak498084 ak498084 287696 Jan 10 08:31 W0001.tmp
-rw-rw---- 1 ak498084 ak498084 291728 Jan 10 08:44 W0002.tmp
-rw-rw---- 1 ak498084 ak498084 290608 Jan 10 09:15 W0003.tmp
-rw-rw---- 1 ak498084 ak498084 290384 Jan 10 09:46 W0004.tmp
-rw-rw---- 1 ak498084 ak498084 290272 Jan 10 10:18 W0005.tmp
-rw-rw---- 1 ak498084 ak498084 289488 Jan 10 10:40 W0006.tmp
-rw-rw---- 1 ak498084 ak498084 290944 Jan 10 11:13 W0007.tmp
-rw-rw---- 1 ak498084 ak498084 289824 Jan 10 11:50 W0008.tmp
-rw-rw---- 1 ak498084 ak498084 289600 Jan 10 12:28 W0009.tmp
-rw-rw---- 1 ak498084 ak498084 287136 Jan 10 06:50 W0010.tmp
-rw-rw---- 1 ak498084 ak498084 388635840 Jan 7 13:08 WAVECAR
-rw-rw---- 1 ak498084 ak498084 0 Jan 8 05:53 WAVECAR.chi
-rw-rw---- 1 ak498084 ak498084 9953408 Jan 7 13:08 WAVEDER
-rw-rw---- 1 ak498084 ak498084 105678448 Jan 10 08:31 WFULL0001.tmp
-rw-rw---- 1 ak498084 ak498084 108659824 Jan 10 08:44 WFULL0002.tmp
-rw-rw---- 1 ak498084 ak498084 107827504 Jan 10 09:15 WFULL0003.tmp
-rw-rw---- 1 ak498084 ak498084 107661424 Jan 10 09:46 WFULL0004.tmp
-rw-rw---- 1 ak498084 ak498084 107578432 Jan 10 10:18 WFULL0005.tmp
-rw-rw---- 1 ak498084 ak498084 106998384 Jan 10 10:40 WFULL0006.tmp
-rw-rw---- 1 ak498084 ak498084 108076864 Jan 10 11:13 WFULL0007.tmp
-rw-rw---- 1 ak498084 ak498084 107246784 Jan 10 11:50 WFULL0008.tmp
-rw-rw---- 1 ak498084 ak498084 107081152 Jan 10 12:28 WFULL0009.tmp
-rw-rw---- 1 ak498084 ak498084 105267648 Jan 10 06:50 WFULL0010.tmp
___________________________________________________________________________

And at end of the logfile, what i get is:

___________________________________________________________________________
QP shifts <psi_nk| G(iteration)W_0 |psi_nk>: iteration 3
NQ= 1 operations 12
NQ= 5 operations 4
NQ= 9 operations 2
NQ= 1 operations 12
NQ= 5 operations 4
NQ= 9 operations 2
allocating 1 responsefunctions rank= 2610
allocating 72 responsefunctions rank= 2610
Doing 1 frequencies on each core in blocks of 72
NQ= 1 0.0000 0.0000 0.0000,
|....|.........|.........|....|.........|.........|.........|.........|.........|.........|.........|.........|.........
|.........|.........|.........
performed 2156000 updates of chi_q(r,r)
total number of BLAS operations 113684.18 Gflops
NQ= 2 0.1250 -0.0000 0.0000,
|....|.........|.........|....|.........|.........|.........|.........|.........|.........|.........|.........|.........
|.........|.........|.........
performed 7049000 updates of chi_q(r,r)
total number of BLAS operations 378946.35 Gflops
NQ= 3 0.2500 -0.0000 0.0000,
|....|.........|.........|....|.........|.........|.........|.........|.........|.........|.........|.........|.........
|.........|.........|.........
performed 11942000 updates of chi_q(r,r)
total number of BLAS operations 642174.30 Gflops
NQ= 4 0.3750 0.0000 0.0000,
|....|.........|.........|....|.........|.........|.........|.........|.........|.........|.........|.........|.........
|.........|.........|.........
performed 16835000 updates of chi_q(r,r)
total number of BLAS operations 904996.35 Gflops
NQ= 5 0.5000 -0.0000 0.0000,
|....|.........|.........|....|.........|.........|.........|.........|.........|.........|.........|.........|.........
|.........|.........|.........
performed 20484000 updates of chi_q(r,r)
total number of BLAS operations 1100847.97 Gflops
NQ= 6 0.1250 0.1250 0.0000,
|....|.........|.........|....|.........|.........|.........|.........|.........|.........|.........|.........|.........
|.........|.........|.........
performed 26124000 updates of chi_q(r,r)
total number of BLAS operations 1401893.85 Gflops
NQ= 7 0.2500 0.1250 0.0000,
|....|.........|.........|....|.........|.........|.........|.........|.........|.........|.........|.........|.........
|.........|.........|.........
performed 32759000 updates of chi_q(r,r)
total number of BLAS operations 1759639.47 Gflops
NQ= 8 0.3750 0.1250 0.0000,
|....|.........|.........|....|.........|.........|.........|.........|.........|.........|.........|.........|.........
|.........|.........|.........
performed 39394000 updates of chi_q(r,r)
total number of BLAS operations 2114634.27 Gflops
NQ= 9 0.2500 0.2500 0.0000,
___________________________________________________________________________

When i restarted this calculation in the exact same folder, the result at the end of 24 hours was exactly the same as the first run. Is there a way to do it from the point of last departure. I mean here it shows " QP shifts <psi_nk| G(iteration)W_0 |psi_nk>: iteration 3" . How can I restart from this point then ?

Thanks for your advice in advance

best,
askhetan

i'm sorry but isn't this a fairly simple yes or no answer - "is it possible to restart GW calculations"

No, mechanism of restarting GW is not programmed.

Thank you. As a suggestion, I think it would be a very useful idea to implement this, given the power of this method. This way one can apply it to systems of realistic sizes and compute it over longer periods of time.