Orbital Transformation in MD

[August Melcher]

My system is a 128 water box, see input file below for the rest of the 
details. I'm running on Edison (Cray XC30) at 
NERSC https://www.nersc.gov/users/computational-systems/edison/

I'm running a 2nd generation Car-Parrinello molecular dynamics scheme 
(http://pubs.acs.org/doi/abs/10.1021/ct800417q)

Starting with the same input, sometimes once the restart history is filled 
each step stays around ~5-6s and I obtain what seems to be optimal timing. 
Sometimes however, the timing for certain steps spikes way up (see attached 
.ener files for details). Does anyone know why this happens and how it 
could possibly be avoided?

&FORCE_EVAL
METHOD Quickstep
  &DFT
  UKS
                WFN_RESTART_FILE_NAME 
/global/u1/a/amelcher/restarts_data/new2CPMD/128equil2CPMD-RESTART.wfn
                BASIS_SET_FILE_NAME 
/global/u1/a/amelcher/soft/cp2k/cp2k/tests/QS/GTH_BASIS_SETS
                POTENTIAL_FILE_NAME 
/global/u1/a/amelcher/soft/cp2k/cp2k/tests/QS/GTH_POTENTIALS
  &MGRID
                        NGRIDS 4
  CUTOFF 320
  &END MGRID
  &QS
  WF_INTERPOLATION ASPC
  EXTRAPOLATION_ORDER 7
  &END QS
  &SCF
  EPS_SCF 1.0E-7
                        EPS_SCF_HIST 1.0E-5
  SCF_GUESS atomic
  MAX_SCF 40
                        MAX_SCF_HIST 1
  &OT T
PRECONDITIONER FULL_KINETIC
                                ALGORITHM IRAC
  MINIMIZER DIIS
         STEPSIZE 0.1
  &END OT
                        &OUTER_SCF
                            EPS_SCF 1.0E-7
                            MAX_SCF 50
                        &END OUTER_SCF
                        &PRINT
                            &RESTART_HISTORY
                                ADD_LAST NUMERIC
                                BACKUP_COPIES 7
                                FILENAME 
=/global/u1/a/amelcher/newirac/128equil2CPMD-RESTART.wfn
                                &EACH
                                    __ROOT__ 1
                                    MD 1
                                    QS_SCF 0
                                &END EACH
                            &END RESTART_HISTORY
                        &END PRINT
  &END SCF
  &XC
  &XC_FUNCTIONAL PBE
  &END XC_FUNCTIONAL
  &END XC
  &PRINT
  &END PRINT
  &END DFT
  &SUBSYS
  &CELL
  ABC 15.745248 15.745248 15.745248
  &END CELL
                &TOPOLOGY
                    COORDINATE XYZ
                    COORD_FILE_NAME 
/global/u1/a/amelcher/coord_files/128water.xyz
                &END TOPOLOGY
&KIND O
  BASIS_SET TZV2P-GTH
  POTENTIAL GTH-PBE-q6
  &END KIND
  &KIND H
  BASIS_SET TZV2P-GTH
  POTENTIAL GTH-PBE-q1
  &END KIND
  &END SUBSYS
&END FORCE_EVAL
&GLOBAL
  PROJECT 128equil2CPMD
  RUN_TYPE MD
  PRINT_LEVEL LOW
&END GLOBAL
&MOTION
  &MD
  ENSEMBLE LANGEVIN
                &LANGEVIN
                    GAMMA 1.35E-5
                    NOISY_GAMMA 8.65E-5
                &END LANGEVIN
  STEPS 1000000
  TIMESTEP 0.5
  &END MD
&END MOTION

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140511/f21c9303/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 1-1.ener
Type: application/octet-stream
Size: 87246 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140511/f21c9303/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 2-1.ener
Type: application/octet-stream
Size: 61701 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140511/f21c9303/attachment-0001.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 3-1.ener
Type: application/octet-stream
Size: 67596 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140511/f21c9303/attachment-0002.obj>