Bug in CELL_OPT with MD??

Joseph Han jhh... at gmail.com
Sat Jan 21 01:14:43 CET 2012


I'm not sure if this is a bug, or if it's a problem on my part.  I've
created an example file with a few water molecules:


&FORCE_EVAL
  METHOD QS
  STRESS_TENSOR ANALYTICAL
  &DFT
    BASIS_SET_FILE_NAME /data/simulation/QM/EC-cp2k/BASIS_MOLOPT
    POTENTIAL_FILE_NAME /data/simulation/QM/EC-cp2k/GTH_POTENTIALS
    LSD
    CHARGE 0
    MULTIPLICITY 1
    RELAX_MULTIPLICITY
    &QS
    &END QS
    &MGRID
      NGRIDS 5
      CUTOFF 500
    &END MGRID
    &SCF
      MAX_SCF 20
      SCF_GUESS ATOMIC
      &OUTER_SCF
      &END OUTER_SCF
      &OT ON
        MINIMIZER CG
        LINESEARCH 3PNT
        PRECONDITIONER FULL_ALL
      &END OT
      &MIXING
         METHOD BROYDEN_MIXING
         ALPHA   0.1
         BETA    1.5
         NBROYDEN  8
      &END
    &END SCF
    &POISSON
      PERIODIC XYZ
      &EWALD
        EWALD_TYPE SPME
      &END EWALD
    &END POISSON
    &XC
      &XC_FUNCTIONAL PADE
      &END XC_FUNCTIONAL
    &END XC
  &END DFT
  &SUBSYS
    &CELL
      ABC 5.0 5.0 5.0
      ALPHA_BETA_GAMMA 90.0 90.0 90.0
      PERIODIC XYZ
      MULTIPLE_UNIT_CELL 1 1 1
    &END CELL
    &COORD
 O     2.77221   3.21647   2.69404
 H     1.94142   2.86547   2.13372
 H     3.22523   2.33011   2.79097
 O     0.85659   0.98497   2.41065
 H     0.68775   1.40092   1.40335
 H     1.14355   0.14146   2.29227
 O     3.39876   4.54704   0.48275
 H     2.91815   5.14962   0.81851
 H     4.34149   4.71169   0.85690
    &END COORD
    &KIND H
       ELEMENT H
       BASIS_SET DZVP-MOLOPT-SR-GTH-q1
       POTENTIAL GTH-PADE-q1
    &END KIND
    &KIND O
       ELEMENT O
       BASIS_SET DZVP-MOLOPT-SR-GTH-q6
       POTENTIAL GTH-PADE-q6
    &END KIND
  &END SUBSYS
&END FORCE_EVAL
&MOTION
  &PRINT
    &TRAJECTORY
      &EACH
        MD 1
      &END EACH
    &END TRAJECTORY
    &VELOCITIES
      &EACH
        MD 1
      &END EACH
    &END VELOCITIES
  &END PRINT
  &GEO_OPT
    OPTIMIZER BFGS
  &END GEO_OPT
  &CELL_OPT
    KEEP_ANGLES TRUE
    OPTIMIZER CG
    TYPE MD
    EXTERNAL_PRESSURE 2.0
    PRESSURE_TOLERANCE 0.1
    &CG
      &LINE_SEARCH
        TYPE 2PNT
      &END LINE_SEARCH
    &END CG
  &END CELL_OPT
  &MD
    ENSEMBLE NVT
    STEPS 3
    TEMPERATURE 350
    &THERMOSTAT
      REGION GLOBAL
      TYPE NOSE
      &NOSE
      &END NOSE
    &END THERMOSTAT
  &END MD
&END MOTION
&GLOBAL
  PROJECT H2O
  FFTW_WISDOM_FILE_NAME /opt/fftw/wisdom.dat-E5645
  RUN_TYPE CELL_OPT
&END GLOBAL

The symptom is that the CELL_OPT steps after the first one seem to
have twice as many MD steps as specified.  This can be seen in the
following command:

$ cat *.ener
#     Step Nr.          Time[fs]        Kin.[a.u.]
Temp[K]            Pot.[a.u.]        Cons Qty[a.u.]        UsedTime[s]
         0            0.000000         0.013300624
350.000000000       -51.477954394       -51.464099577
0.000000000
         1            0.500000         0.022568927
593.891236059       -51.487793114       -51.464666360
20.410000000
         2            1.000000         0.037623061
990.034082790       -51.503438263       -51.465251262
5.983000000
         3            1.500000         0.049416939
1300.384710734       -51.515442559       -51.465452803
4.917000000
#     Step Nr.          Time[fs]        Kin.[a.u.]
Temp[K]            Pot.[a.u.]        Cons Qty[a.u.]        UsedTime[s]
         4            2.000000         0.053467176
1406.964888976       -51.519550159       -51.465499604
14.125000000
         5            2.500000         0.049501637
1302.613504501       -51.515490916       -51.465395255
7.043000000
         6            3.000000         0.039279586
1033.624786518       -51.505042377       -51.465159531
5.969000000
         7            3.500000         0.015007193
394.907585294       -51.506976193       -51.491362786
14.038000000
         8            4.000000         0.018992546
499.780371364       -51.511082408       -51.491480099
5.953000000
         9            4.500000         0.023437730
616.753421941       -51.515612703       -51.491560775
4.861000000
#     Step Nr.          Time[fs]        Kin.[a.u.]
Temp[K]            Pot.[a.u.]        Cons Qty[a.u.]        UsedTime[s]
        10            5.000000         0.027009440
710.741367181       -51.519009416       -51.491380497
12.996000000
        11            5.500000         0.028692745
755.036776087       -51.520757451       -51.491439382
7.031000000
        12            6.000000         0.027676065
728.283312941       -51.519788199       -51.491480877
4.856000000
        13            6.500000         0.012364380
325.363146359       -51.518811647       -51.505813305
12.995000000
        14            7.000000         0.012366870
325.428667734       -51.518799708       -51.505796270
5.956000000
        15            7.500000         0.013533596
356.130539514       -51.519993047       -51.505820155
4.854000000

Where the subsequent steps also seem to have 6 MD steps.  When trying
to run the method on larger systems for more time steps, eventually
there is a SEGFAULT.  I'm wondering if the SEGFAULT is because there
are too many attempted MD steps per CELL_OPT step.

Any ideas?

Joseph




More information about the CP2K-user mailing list