Why some overlapped steps appear in the new trajectory file after RESTART the MD in the CP2K?

[jia]

Hello Everyone, 
    When I calculate the MD with CP2K.  Some overlapped steps appear in the 
new trajectory file after RESTART the MD. 
For example, I already get the MD output of  i=3906, 
...

773
 i =     3906, time =     1953.000, E =     -4474.7541142663
  N        18.9864963170        8.7518562091       10.6256224398
  O        18.5387553608        9.8006093920       11.1299284364
  O        18.3563092587        7.6487763074       10.7699162566
  O        20.0918231195        8.7474007752        9.9520189788 
  ---
  H        11.0477142399       10.5711430820       17.8516880476
  H        10.3012118921        9.3683700815       18.6096133300


So when we restart the MD again, I expect that  the 3907th step will come 
out. But the result is 3901st step, 3902nd steps, 3903rd steps,.... then I 
have two  3901st,3902nd,..., 3906th step. Unfortunately, the data is 
different between the two "3901th step" trojectory (for, 3902nd, 
3903rd,...,3906th step as well). 

 773
 i =     3901, time =     1950.500, E =     -4474.7861914555
  N        18.9915342896        8.7315755847       10.6140689267
  O        18.5431556496        9.7796178151       11.1373780816
  O        18.3407752730        7.6450796766       10.7633629707
  O        20.1093185435        8.7631643653        9.9548188259 
  ...

773
 i =     3902, time =     1951.000, E =     -4474.7782316820
  N        18.9900734976        8.7359794467       10.6165886903
  O        18.5428202911        9.7832959747       11.1359394674
  O        18.3441426920        7.6460054209       10.7644822741
  O        20.1059971546        8.7607514539        9.9543488296 
  ...

  ...

Since there is a jump  between the energy (position) of the 3906th step in 
the last MD trajectory and  the energy (position) in the 3901st step in the 
current one, some calculation, such as time correlation function,  will be 
not accurate. So it is better to solve this problem.  Does anyone know the 
possible reasons for this overlapped trajectory?  Thanks a lot advanced. 

Jia


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20121211/45f7c33b/attachment.htm>