[DFTB-Plus-User] Different trajectory and energies for the different run

[jsxz]

Dear all,

When I set different steps and MDrestartFrequency (other parameters do not change) a few days ago, there are different trajectory and different energies.



Later, I try the same system with the same setting with the serial code, the trajectory and energies are startling.


1. I try the same system with the same setting with the "same" serial code (running the serial code for two times), the trajectory and energies are different when the MD iter is bigger than 300 steps. The difference is more and more big with the MD iter increasing, I am very confused.
a)For the first time, the energies is as follows:
 MD step:         410
 Pressure:                                -0.234067E-05 au               -0.688648E+08 Pa
 Potential Energy:                -490.6914390914 H        -13352.3934 eV
 MD Kinetic Energy:                  6.5383963122 H             177.9188 eV
 Total MD Energy:                 -484.1530427791 H      -13174.4746 eV
 MD Temperature:                     0.0151879125 au         4795.9577 K

For the first time, the trajectory is as follows:
288
MD iter: 410
    C      8.03654320      0.95517386    -11.96195149      3.96702850     21.19206211    -18.42919763      3.18951657
    C      6.96850391      1.87276743    -11.34894777      4.04300662     -4.87773890    -10.13079429      3.28427545
    C      5.64341693      1.95995357    -12.03037618      4.00324029      0.03068755    -51.48245231     -0.04639527
...
...

b) For the second time, the energies is as follows:
MD step:         410
 Pressure:                          -0.233884E-05 au    -0.688111E+08 Pa
 Potential Energy:                -490.6912020839 H       -13352.3870 eV
 MD Kinetic Energy:                  6.5381300912 H          177.9116 eV
 Total MD Energy:                 -484.1530719927 H       -13174.4754 eV
 MD Temperature:                     0.0151872941 au        4795.7624 K

For the second time, the trajectory is as follows:
288
MD iter: 410
    C      8.03649985      0.95514783    -11.96194439      3.96699301     21.19308616    -18.42461428      3.19092069
    C      6.96845152      1.87279035    -11.34899297      4.04297989     -4.87944064    -10.12696930      3.28718816
    C      5.64340211      1.95993610    -12.03037754      4.00324518      0.03010413    -51.48061127     -0.04931038

...
...

2. The results  using the SCC-DFTB_MPI are different from that with the seiral code for the same system with  the setting,
    even if np is 1 for the dftbmpi+  (nodes=1:ppn=1)


3. I set the EwaldParameter with the value of 0.15, there is still the different  trajectory and energies.
    The  trajectory and energies with the value of 0.0 for the EwaldParameter is different from the results with the valuse of 0.15,
       whether for dftbmpi+ or the serial code.

4. I run the collision dynamics for the CNT. Whether it should blame the different behavior for the collsion dynamics?



I would be really thankful for any help.
Best regards

Xiaobao Zhang
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20150211/d6615b24/attachment.htm>