Author Topic: restart of a ccsd(t) calc that hit walltime  (Read 7488 times)

Punix

  • Newbie
  • *
  • Posts: 3
  • Karma: +0/-0
restart of a ccsd(t) calc that hit walltime
« on: March 12, 2017, 11:09:44 AM »
Dear users,

I've found similar threads but not quite what I want.

If my CCSD(T) calculation (TM 7.0) hit the walltime during the calculation of triples (CCSD-part was completed). Can I simply restart that calculation or do I need to delete some files first or did I had to use a specific keyword to be able to restart it (which I probably did not as Im not very experienced with TM).

Or do I have to do the hardway and do the whole calculation again?

Here is my input control file:
$maxcor    10000
$coord    file=coord
$scfconv    10
$cbas    file=auxbasis
$atoms
c 2-2,6,9,13,16,20 \
   cbas = c aug-cc-pVTZ                     \
   basis = c aug-cc-pVTZ
o 5-5,12,19 \
   cbas = o aug-cc-pVTZ                     \
   basis = o aug-cc-pVTZ
h 1-1,3-4,7-8,10-11,14-15,17-18,21-23 \
   cbas = h aug-cc-pVTZ                     \
   basis = h aug-cc-pVTZ
$cosmo_out    file=job_notRun_1.cosmo
$title    tgbbgt
$lastdiag
$basis    file=basis
$operating system unix
$denconv    1d-8
$tmole
$optimize
 internal off
 redundant off
 cartesian on
 global off
$symmetry    c1
$grad    file=gradient
$scfmo   file=mos
$ricc2
ccsd(t)
$scfdamp     start=0.700 step=0.050 min=0.100
$freeze
implicit core= 9 virt= 0
$subtitle
SP,CCSD(T)/RI,charge 0,c1,aug-cc-pVTZ,
$rundimensions
   dim(fock,dens)=359304
   natoms=23
   nshell=252
   nbf(CAO)=845
   nbf(AO)=736
   dim(trafo[SAO<-->AO/CAO])=1063
   rhfshells=1
$closed shells
 a       1-37                                   ( 2 )
$scfiterlimit       30
$thize     0.10000000E-04
$thime        5
$scfdump
$scfintunit
 unit=30       size=0        file=twoint
$scfdiis
$scforbitalshift  automatic=.1
$drvopt
   cartesian  on
   basis      off
   global     off
   hessian    on
   dipole     on
   nuclear polarizability
$forceinit on
   diag=default
$forceapprox    file=forceapprox
$energy    file=energy
$lock off
$actual step      ccsdf12
$orbital_max_rnorm 0.10963372954648E-07
$last SCF energy change = -460.07079
$charge from dscf
         -0.000 (not to be modified here)
$dipole from dscf
  x     0.00271883745107    y    -0.01824927750582    z    -0.14632785715261    a.u.
   | dipole | =    0.3748761438  debye
$end



It looks like when I simply start to run the job again he kind of realizes that the CCSD part is already done and tries to go on as the output file says now the following:


"restarted calculation begins with wavefunction model nb. 1 which is CCSD(T)


   **************************************************************************
   *                                                                        *
   *          OPTIMIZATION OF THE GROUND STATE CLUSTER AMPLITUDES           *
   *                                                                        *
   **************************************************************************
     threshold for screening of BF intermediate: 0.56E-16
     restart CCSD(T) from CCSD amplitudes found on the R0   file.
     remaining residual norm is   0.14E-06
     CCSD restart energy:   -461.9209156901


   Iter.   CCSD energy      Norm(Omega)    Norm(t1)  Norm(t2)     cpu    wall"


But then it seems like it did not do anything further and got aborted on the computer cluster. Tried to check all the files to see what the problem is/was but kind find anything.


Thanks for any help,
Punix
« Last Edit: March 12, 2017, 11:18:42 AM by Punix »

Arnim

  • Developers
  • Sr. Member
  • *
  • Posts: 253
  • Karma: +0/-0
Re: restart of a ccsd(t) calc that hit walltime
« Reply #1 on: March 13, 2017, 12:11:14 AM »
Dear Punix,

you can just restart the program to continue after the CCSD part. For the restart the CC* and restart.cc files will be used.

The triples part itself can not be restarted. So, you have to have enough walltime for that.

Cheers,

Arnim

Punix

  • Newbie
  • *
  • Posts: 3
  • Karma: +0/-0
Re: restart of a ccsd(t) calc that hit walltime
« Reply #2 on: March 13, 2017, 09:52:14 AM »
Thanks for the quick answer :)

Then I guess I ran out of scratch filespace as all these temporarily files are quite large. I saved them overnight to my PC and now deleted some and will try to carry on the non-deleted jobs. Hopefully that works.