NEB question

Bernd Ensing ensin... at gmail.com
Tue Mar 24 10:36:37 UTC 2015


Dear people,

I have three questions about a NEB calculation that I am running (see input 
and output of first replica attached).

1) The number of replicas is 5, with the first and last replicas fixed. 
Since the first replica is fixed (indeed the positions are constant in the 
trajectory output), why is the wave function optimized after every NEB 
geometry cycle? In the attached output the number of SCF cycles is more 
than 1 (often more than 100 actually), even though the geometry has not 
changed.

2) The number of processors per replica is 8, and I run on 2 nodes of 8 
processors each. I see that on node 2, there is an Silver-BAND6.out (and 
Silver-BAND2.out and Silver-BAND4.out) containing info of SCF cycles and 
energies and so forth. What is this 6th output doing, considering that 
there are only 5 replica's (there are only 5 trajectory outputs)?  I was 
hoping that cp2k would cycle all 5 replicas over the 2 nodes subsequently; 
instead it puts replica's 1,3, and 5 on node 1, and 2, 4 and 6(???) on node 
2. Is there a smarter way to use 2x8 procs on 5 replica's?  (the system 
does not really scale to 1x16 procs).

3) The number of SCF cycles is often more than 100 for all replicas. In 
"normal" geometry optimization runs the number of SCF cycles is typically 7 
to 15 (apart from the first round). Is this poor convergence in BAND normal 
(due to the "stretched" geometries maybe?).  Does anyone have advice on 
this?

many thanks in advance.
Bernd





-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20150324/392479a7/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Silver-BAND1.out
Type: application/octet-stream
Size: 635361 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20150324/392479a7/attachment.obj>
-------------- next part --------------
&FORCE_EVAL
  METHOD QS

  &DFT
    BASIS_SET_FILE_NAME /home/ensing/BASIS_SET/BASIS_COMBI
    POTENTIAL_FILE_NAME /home/ensing/BASIS_SET/GTH_POTENTIALS
    &MGRID
      CUTOFF 280
      NGRIDS 5
    &END MGRID
    &QS
      METHOD GPW
      EPS_DEFAULT 1.0E-12
    &END QS
    &SCF
      MAX_SCF 100
      EPS_SCF 5.0E-7
      SCF_GUESS RESTART
      &OUTER_SCF
            EPS_SCF 1.0e-5
            MAX_SCF 4
      &END
      &OT
        MINIMIZER DIIS
        N_DIIS 7
        PRECONDITIONER FULL_ALL
        ENERGY_GAP 0.001
      &END
    &END SCF
    &XC
      &XC_FUNCTIONAL PBE
      &END XC_FUNCTIONAL
      &VDW_POTENTIAL
        DISPERSION_FUNCTIONAL PAIR_POTENTIAL
         &PAIR_POTENTIAL
            TYPE DFTD2
            REFERENCE_FUNCTIONAL PBE
         &END PAIR_POTENTIAL
      &END VDW_POTENTIAL
      &XC_GRID 
         XC_SMOOTH_RHO NN10 
         XC_DERIV SPLINE2_SMOOTH 
      &END XC_GRID
    &END XC
  &END DFT

  &SUBSYS
    &CELL
      ABC                11.66 11.66 30.000  
      ALPHA_BETA_GAMMA   90.000 90.000 60.000
      PERIODIC XYZ
    &END CELL
    &TOPOLOGY
      COORDINATE XYZ
      COORD_FILE replica1.xyz
    &END TOPOLOGY
    &KIND Ag
      BASIS_SET DZVP-MOLOPT-SR-GTH-q11
      POTENTIAL GTH-PBE-q11
    &END KIND
    &KIND H
      BASIS_SET DZVP-GTH
      POTENTIAL GTH-PBE-q1
    &END KIND
    &KIND C
      BASIS_SET DZVP-GTH
      POTENTIAL GTH-PBE-q4
    &END KIND
    &KIND O
      BASIS_SET DZVP-GTH
      POTENTIAL GTH-PBE-q6
    &END KIND
    &KIND Si
      BASIS_SET DZVP-MOLOPT-SR-GTH-q4
      POTENTIAL GTH-PBE-q4
    &END KIND
  &END SUBSYS
&END FORCE_EVAL

&GLOBAL
  PROJECT test
  RUN_TYPE BAND
  PRINT_LEVEL LOW
  PREFERRED_FFT_LIBRARY FFTSG
&END GLOBAL

&MOTION

  &BAND
   NPROC_REP 8
   BAND_TYPE IT-NEB
   NUMBER_OF_REPLICA 5 
     &OPTIMIZE_BAND 
      OPT_TYPE DIIS
      &DIIS
      &END
      OPTIMIZE_END_POINTS FALSE
     &END

   &REPLICA
    COORD_FILE_NAME replica1.xyz
   &END
   &REPLICA
    COORD_FILE_NAME replica2.xyz
   &END
   &REPLICA
    COORD_FILE_NAME replica3.xyz
   &END
   &REPLICA
    COORD_FILE_NAME replica4.xyz
   &END
   &REPLICA
    COORD_FILE_NAME replica5.xyz
   &END
   &PROGRAM_RUN_INFO HIGH
   &END
   &CONVERGENCE_INFO HIGH
   &END
   &REPLICA_INFO HIGH
   &END
  &END BAND
  &CONSTRAINT
   &FIXED_ATOMS
    LIST 1..16
   &END
  &END CONSTRAINT
  &GEO_OPT
   MAX_ITER 100
  &END
  
  &PRINT
   &TRAJECTORY
    FORMAT XMOL
    &EACH
     GEO_OPT 1
    &END
   &END TRAJECTORY
   &RESTART
    &EACH
     GEO_OPT 0
    &END
    ADD_LAST NUMERIC
   &END RESTART
  &END PRINT
&END MOTION



More information about the CP2K-user mailing list