Author Topic: RI-CC2 excitation calculation  (Read 7330 times)

philippt

  • Newbie
  • *
  • Posts: 3
  • Karma: +0/-0
RI-CC2 excitation calculation
« on: November 16, 2015, 02:22:16 PM »
Hello,

at first, the version that I use is TURBOMOLE V7.0.1( 19978 ).

Now my problem: I am trying to do a single point CC2 excitation analysis to compare it to my DFT results, but the calculation always stops before any excitation is calculated.
I am starting the calculation on our cluster through a script using ricc2 > ricc2.out.

My control file:
Code: [Select]
$title
M0002
$operating system unix
$symmetry c1
$redundant    file=coord
$coord    file=coord
$user-defined bonds    file=coord
$atoms
ru 1                                                                           \
   basis =ru def2-SV(P)                                                        \
   ecp   =ru def2-ecp                                                          \
   cbas  =ru def2-SV(P)                                                        \
   jkbas =ru def2-SV(P)
n  2-3,26,32,43,51,62-63,72                                                    \
   basis =n def2-SV(P)                                                         \
   cbas  =n def2-SV(P)                                                         \
   jkbas =n def2-SV(P)
c  4-6,8,10,13-14,16,18,20,22,24,27-31,33,35-36,39,41,44-46,48-49,52,54-55,59, \
   61,64-69,73-78,80,85-86                                                     \
   basis =c def2-SV(P)                                                         \
   cbas  =c def2-SV(P)                                                         \
   jkbas =c def2-SV(P)
h  7,9,11-12,15,17,19,21,23,25,34,37-38,40,42,47,50,53,56-58,60,70-71,79,81-82 \
   87-92                                                                       \
   basis =h def2-SV(P)                                                         \
   cbas  =h def2-SV(P)                                                         \
   jkbas =h def2-SV(P)
o  83-84                                                                       \
   basis =o def2-SV(P)                                                         \
   cbas  =o def2-SV(P)                                                         \
   jkbas =o def2-SV(P)
$basis    file=basis
$ecp    file=basis
$rundimensions
   dim(fock,dens)=464460
   natoms=92
   nshell=424
   nbf(CAO)=962
   nbf(AO)=902
   dim(trafo[SAO<-->AO/CAO])=1082
   rhfshells=1
   nt1amt=101210
$scfmo   file=mos
$scfiterlimit     1024
$scfconv        8
$denconv      1d-8
$thize     0.10000000E-04
$thime        5
$scfdamp   start=1.000  step=0.050  min=0.100
$scfdump
$scfintunit
 unit=30       size=0        file=twoint
$scfdiis
$maxcor    49152
$scforbitalshift  automatic=.1
$drvopt
   cartesian  on
   basis      off
   global     off
   hessian    on
   dipole     on
   nuclear polarizability
$interconversion  off
   qconv=1.d-7
   maxiter=25
$optimize
   internal   on
   redundant  on
   cartesian  off
   global     off
   basis      off   logarithm
$coordinateupdate
   dqmax=0.3
   interpolate  on
   statistics    5
$forceupdate
   ahlrichs numgeo=0  mingeo=3 maxgeo=4 modus=<g|dq> dynamic fail=0.3
   threig=0.005  reseig=0.005  thrbig=3.0  scale=1.00  damping=0.0
$forceinit on
   diag=default
$energy    file=energy
$grad    file=gradient
$forceapprox    file=forceapprox
$lock off
$orbital_max_rnorm 0.27699012812758E-06
$last SCF energy change = -2530.7698
$charge from dscf
          2.000 (not to be modified here)
$dipole from dscf
  x    13.57501811728358    y     2.13764202353045    z     0.60266025986182    a.u.
   | dipole | =   34.9632623794  debye
$closed shells
 a       1-204                                  ( 2 )
$freeze
 implicit core=   59 virt=    0
$ricc2
  cc2
  restart
$excitations
  irrep=a  multiplicity=1 nexc= 30
  spectrum states=all operators=diplen,dipvel
  xgrad states=(a 30)
  exprop relaxed states=all
$spectrum ev
$cbas    file=auxbasis
$jkbas    file=auxbasis
$actual step      ricc2
$end
The xgrad states=(a 30) and exprop relaxed states=all commands have not been in my first version. Those are solutions from this forum to - at least in my untrained eyes -  similar problems.

And a little bit of my output - no error messages here:
Code: [Select]
   **************************************************************************
   *                                                                        *
   *          OPTIMIZATION OF THE GROUND STATE CLUSTER AMPLITUDES           *
   *                                                                        *
   **************************************************************************
     restart CC2 from CC2 amplitudes found on the R0   file.
     remaining residual norm is   0.37E-07


   Iter.   CC2 energy       Norm(Omega)    Norm(t1)  Norm(t2)     cpu    wall
   ...........................................................................
    1  -2538.8103196929    0.0000000198     0.49187   1.04274   13.49   16.45
   ...........................................................................

   last energy change         :  0.41E-11
   convergence threshold      :  0.10E-07
   norm of the vector function:  0.20E-07
   convergence threshold      :  0.10E-06

     time in total         cpu: 13 min 33 s  wall: 16 min 31 s  ratio:  0.8


     **********************************************************************
     *                                                                    *
     *   RHF  energy                             :  -2530.7698080890      *
     *   correlation energy                      :     -8.0405116039      *
     *                                                                    *
     *   Final CC2 energy                        :  -2538.8103196929      *
     *                                                                    *
     *   D1 diagnostic                           :      0.2520            *
     *                                                                    *
     **********************************************************************



 **************************************************************
 *                                                            *
 *    OPTIMIZATION OF THE EXCITED STATE CLUSTER AMPLITUDES    *
 *                                                            *
 **************************************************************


  preoptimize excitation energies/vectors using CCS:
  ==================================================


 ======== LINEAR CC RESPONSE SOLVER ========

    threshold for convergence: 0.10E-02
    maximum number of simultaneous jacoby matrix transformations: 30

     summary of start vectors generation:
     -------------------------------------------
     type of solution vectors      : RE0
     symmetry                      : a
     number of vectors requested   :    30
     number of vectors generated   :    30
       -- guessed as rhs/(e_a-e_i) :    30
     -------------------------------------------


    Iter  #Vectors  time (min)    max. residual
    --------------------------------------------

The queue system reports the following error message in my *.e* file:
Code: [Select]
MPI Application rank 19 killed before MPI_Finalize() with signal 9
forrtl: error (78): process killed (SIGTERM)
Image              PC                Routine            Line        Source
ricc2_mpi          00000000019D6DB1  Unknown               Unknown  Unknown
ricc2_mpi          00000000019D5507  Unknown               Unknown  Unknown
ricc2_mpi          000000000197AC84  Unknown               Unknown  Unknown
ricc2_mpi          000000000197AA96  Unknown               Unknown  Unknown
ricc2_mpi          00000000019228AF  Unknown               Unknown  Unknown
ricc2_mpi          000000000192910B  Unknown               Unknown  Unknown
libpthread.so.0    00002B00D672F4A0  Unknown               Unknown  Unknown
libpthread.so.0    00002B00D672E4D0  Unknown               Unknown  Unknown
ricc2_mpi          0000000001927A30  Unknown               Unknown  Unknown
ricc2_mpi          000000000193F8C5  Unknown               Unknown  Unknown
ricc2_mpi          000000000193E5C1  Unknown               Unknown  Unknown
ricc2_mpi          00000000004D5181  cc3idxread_              1243  cc3idxio.f
ricc2_mpi          00000000004D1256  cc3idx_get_genera         938  cc3idxio.f
ricc2_mpi          00000000004CF597  cc3idx_get_nopoin        1031  cc3idxio.f
ricc2_mpi          00000000008060F8  cc_jgterm_                510  cc_jgterm.f
ricc2_mpi          0000000000861ED3  cc_rhtr_                 1575  cc_rhtr.f
ricc2_mpi          0000000000724B12  cc_linsol_                922  cc_linsol.f
ricc2_mpi          000000000070FE4F  cc_exci_                  694  cc_exci.f
ricc2_mpi          0000000000423C3D  MAIN__                   2687  ricc2.f
ricc2_mpi          00000000004080BE  Unknown               Unknown  Unknown
libc.so.6          00002B00D695ECDD  Unknown               Unknown  Unknown
ricc2_mpi          0000000000407FC9  Unknown               Unknown  Unknown

[...] this is repeted (without counting) probably 24 times.

As I am a very beginner in using Turbomole, whatever I tried did not help. Actually I don't even know, if anything that I chose to be written into the control file makes sense. :)
I hope that you can help me finding my error and guide me a little bit through the use of this module. Am I right, that Cosmo is currently not available with CC2?

Thank you very much
Philipp

uwe

  • Global Moderator
  • Hero Member
  • *****
  • Posts: 558
  • Karma: +0/-0
Re: RI-CC2 excitation calculation
« Reply #1 on: November 16, 2015, 04:59:16 PM »
Hi,

Quote
MPI Application rank 19 killed before MPI_Finalize() with signal 9

signal 9 is segmentation fault. Usually a sign that the calculation required too much memory.

In your control file you asked for almost 50 GB memory ($maxcor    49152) - but in the case of an MPI job this is the amount of memory per process. The job was running on at least 19 CPUs (according to the output). I do not know how many processes per nodes you started, but 50GB times this number is probably too much for the node.

Reduce the value in $maxcor such that it fits into your memory.

Regards,

Uwe

philippt

  • Newbie
  • *
  • Posts: 3
  • Karma: +0/-0
Re: RI-CC2 excitation calculation
« Reply #2 on: November 17, 2015, 09:09:40 AM »
Hi,

as the calculation now runs smoothly, this must have been the error.

Thank you very much
Philipp

christof.haettig

  • Global Moderator
  • Sr. Member
  • *****
  • Posts: 291
  • Karma: +0/-0
    • Hattig's Group at the RUB
Re: RI-CC2 excitation calculation
« Reply #3 on: November 20, 2015, 02:32:24 PM »
xgrad and exprop have nothing to do with the memory management. If you don't need the results that you request with these
keywords (see the manual),  I recommend to remove them to save some computational time...

xgrad states=(a 30)       

requests a gradient calculation for state 30.

exprop relaxed states=all

requests the calculation of orbital relaxed dipole moments for the all excited states. This will increase the computational time
by about 50%.

In any case, I recommend strongly that you take the time to read the CC2 chapter of the manual.

Christof