RESTART using REFTRAJ

[tobias.l...@gmail.com]

Hello CP2K'ler,

i do my CP2K calculations (Localization of Wannier centers) on a HPC Cluster with a maximum walltime of 24 hours. This is too short, so i have to divide the calculation. I want to continue/restart the previous CP2K run at the timestep and with parameters of the terminition timestep. i have tried it with

&EXT_RESTART
 RESTART_FILE_NAME myproject-1.restart
&END

Desired result as example:
Run 1: timestep 1-50
Run 2: timestep 51-100

Recieved results: (files appended)
Run 1: timestep 1-50
Run 2: timestep 1-50

all results written in one Wannier-file "myproject_wannier.xyz".

I guess that the "myproject_reftraj.xyz" file is restarted each time at timestep 1. (compare timestep 1 of each Run)

Is there any possibility to get a restart as desired?

I have tried the same with FIRST_SNAPSHOT in section MOTION/MD/REFTRAJ too,
or tried to handle it with RESTART_COUNTERS in section EXT_RESTART,
but everytime i recieved the same output in the wannier-file. (files appended)

All my Input-files are appended!

Hope to get help.

Regards,
Tobias Lechner

[Matthias Krack]

Hi,

what's about splitting the trajectory file in two parts (1-50 and 51-100) and performing two independent REFTRAJ runs? I think there is no need to restart any MD run properly, because you are just traversing a set of configuration calculating certain properties.

Regards,

Matthias

[tobias.l...@gmail.com]

Thanks for the reply. This should work fine but a little bit annoying for large molecular systems with 3000 timesteps and about 5000 file-rows per timestep.
But nevertheless this method works :)

[hut...@chem.uzh.ch]

Can't you use the
FIRST_SNAPSHOT keyword in the REFTRAJ section?
No need to edit the files?

regards

Juerg
--------------------------------------------------------------
Juerg Hutter                         Phone : ++41 44 635 4491
Institut für Chemie C                FAX   : ++41 44 635 6838
Universität Zürich                   E-mail: hut...@chem.uzh.ch
Winterthurerstrasse 190
CH-8057 Zürich, Switzerland
---------------------------------------------------------------

-----cp...@googlegroups.com wrote: -----To: cp...@googlegroups.com
From: tobias.l...@gmail.com
Sent by: cp...@googlegroups.com
Date: 01/21/2015 01:55PM
Subject: [CP2K:6022] Re: RESTART using REFTRAJ

Thanks for the reply. This should work fine but a little bit annoying for large molecular systems with 3000 timesteps and about 5000 file-rows per timestep.
But nevertheless this method works :)

--

You received this message because you are subscribed to the Google Groups "cp2k" group.

To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.

To post to this group, send email to cp...@googlegroups.com.

Visit this group at http://groups.google.com/group/cp2k.

For more options, visit https://groups.google.com/d/optout.

[huan...@mail.huji.ac.il]

Hi Tobias,

I suggest that you'd better to use the following parameters for restarting simulations.

(1) For external restart section:

&EXT_RESTART

    RESTART_DEFAULT TRUE   

    RESTART_FILE_NAME YOUR_restart_file.restart     # specified the restart file

    RESTART_COUNTERS TRUE    # set the step counter followed your restart point. If it is False or do not specified, the step number will start from 1.  

    RESTART_POS TRUE   # specified the velocity file to read the velocity values

    RESTART_VEL TRUE    # specified the velocity file to read the velocity values

    RESTART_RANDOMG TRUE   

&END EXT_RESTART

(2) For the DFT section:

&DFT

     WFN_RESTART_FILE_NAME YOUR_wavefunction_file.wfn   # specified the wave function restart file. The program will read the wave function from your specified file.

     ...

&DFT END

(3) For the SCF section:

&SCF

    SCF_GUESS RESTART  # set this keyword, then the program will read data from the restart file.

    ...

&END SCF

I also attached the input file added above improvements. I hope those suggestion will helpful. 

All the best, 

Huan

[tobias.l...@gmail.com]

Hi CP2K user,

i have traced the point which is not working when i try to restart my simulations. With the FIRST_SNAPSHOT directive CP2K loads the right timestep of the reference trajectory. But the calculation of the Wannier-Centers doesnt work at all. The Coordinates in the Wannier-file differ from the ones calculated in one CP2K-Run after the checkpointing timestep. In my appended Wannier-files the respective checkpoint is between timestep 27 and 28. First and second Run input-files are appended too.

Has anyone ideas how to restart a calculation of Wannier-centers correctly?




 

Hi all,

the suggestion of Huan works. Restarting the simulation writes the Wannier-files as desired,
 

Run 1: timestep 1-50
Run 2: timestep 51-10
 

but i have to add the FIRST_SNAPSHOT directive to get the appropriate result.

 

The final target is to calculate raman-spectra with the trajectories using the software TRAVIS interacting with CP2K. I calculated the same trajectory with a single CP2K-run and with a two step restarted CP2K-run. And everytime when TRAVIS analyses a trajectory of the single CP2K-run the recieved raman-spectrum looks like expected. But once analysing a trajectory of the restarted two step CP2K-run the spectrum differs from the other one in an inacceptable way.

Has anyone experiences with restarting CP2K calculations and calculating raman-spectra with the TRAVIS-CP2K-combination?

Hope to get another time help!

Regards,
Tobias