restarting NPT MD

Matt W mattwa... at gmail.com
Tue Oct 17 15:01:04 UTC 2017


Hi,

it seems to be a problem with the &VELOCITY and &MASS of &BAROSTAT section.

For NPT_I they are single valued.  For NPT_F they should be a 3x3 tensor, 
and CP2K dies when it doesn't find the extra elements.

As a hack you could try putting in a diagonal tensor with the isotropic 
element in the correct positions (I guess elements 1,5,9 with fortran 
indices).

This seems to work for me with a fairly up-to-date src.

Matt

On Tuesday, October 17, 2017 at 10:44:50 AM UTC+1, Luca wrote:
>
> Hi Matt
>
> please find attached the restart file I obtain from i1.inp.
>
> Luca 
>
> Il giorno martedì 17 ottobre 2017 11:37:50 UTC+2, Matt W ha scritto:
>>
>> Could you attach the actual restart (feel free to delete coords and 
>> velocities if confidential).
>>
>> Can you run an NPT_F (one or two steps fine), restart NPT_I, then restart 
>> NPT_F?
>>
>> I guess that some default settings needed for NPT_F must not be properly 
>> set during an NPT_I run, then cause the problem when you restart.
>>
>> Matt
>>
>> On Tuesday, October 17, 2017 at 10:32:15 AM UTC+1, Luca wrote:
>>>
>>> Dear Matt and Marcella
>>>
>>> I tried almost every kind of combination of the keywords in the external 
>>> restart section. However, cp2k always complains if I try to restart an 
>>> NPT_F from a NPT_I.
>>> I have attached 2 inputs which give similar errors. i1.inp is the 
>>> initial NPT_I run and i2.inp is the restart.
>>> If you find my mistake please let me know...
>>>
>>> Luca
>>> Best Wishes.
>>> Luca  
>>>  
>>>
>>> Il giorno martedì 17 ottobre 2017 09:38:17 UTC+2, Marcella Iannuzzi ha 
>>> scritto:
>>>>
>>>> Dear Luca,
>>>>
>>>> you can leave RESTART DEFAULTS F, and select what exactly has to be 
>>>> restarted with the specific keywords.
>>>> In a NPT run, for sure the CELL.
>>>> If the problem you have now is related to some constraints that need to 
>>>> be released, you can try with RESTART_CONSTRAINTS F
>>>> Hope this help.
>>>> Kind regards
>>>> Marcella
>>>>
>>>>
>>>> On Monday, October 16, 2017 at 10:08:57 PM UTC+2, Luca wrote:
>>>>>
>>>>> Dear Marcella
>>>>>
>>>>> thank you for your answer. Putting RESTART DEFAULTS T fixes the 
>>>>> oscillation in the average pressure.
>>>>> However, I am still not able to restart an NPT_F from the NPT_I. I 
>>>>> also added the key RESTART_CELL T, 
>>>>> but just after the first scf -loop I get the error 
>>>>>
>>>>> MD_ENERGIES| Initialization proceeding
>>>>>
>>>>>  ******************************** GO CP2K GO! 
>>>>> **********************************
>>>>>  INITIAL POTENTIAL ENERGY[hartree]     =                    
>>>>>  -0.220770726164E+04
>>>>>  INITIAL KINETIC ENERGY[hartree]       =                      
>>>>> 0.403768735935E+00
>>>>>  INITIAL TEMPERATURE[K]                =                              
>>>>>    284.281
>>>>>  INITIAL BAROSTAT TEMP[K]              =                      
>>>>> 0.482402310612E+02
>>>>>  INITIAL PRESSURE[bar]                 =                    
>>>>>  -0.416980916246E+04
>>>>>  INITIAL VOLUME[bohr^3]                =                      
>>>>> 0.223680134131E+05
>>>>>  INITIAL CELL LNTHS[bohr]   =      0.2817577E+02   0.2817577E+02  
>>>>>  0.2817577E+02
>>>>>  INITIAL CELL ANGLS[deg]    =      0.9000000E+02   0.9000000E+02  
>>>>>  0.9000000E+02
>>>>>  ******************************** GO CP2K GO! 
>>>>> **********************************
>>>>>
>>>>>
>>>>>  *******************************************************************************
>>>>>  *   ___                                                              
>>>>>          *
>>>>>  *  /   \                                                              
>>>>>         *
>>>>>  * [ABORT]                                                            
>>>>>          *
>>>>>  *  \___/                        Error in diagonalisation              
>>>>>         *
>>>>>  *    |                                                                
>>>>>         *
>>>>>  *  O/|                                                                
>>>>>         *
>>>>>  * /| |                                                                
>>>>>         *
>>>>>  * / \                                          
>>>>> common/eigenvalueproblems.F:88 *
>>>>>
>>>>>  *******************************************************************************
>>>>>
>>>>> which is a bit odd considering that I use the OT algorithm. 
>>>>> I also tried the restart by reducing the time step from 0.5 to 0.1 fs 
>>>>> but I always get the same error.
>>>>> If I leave the keyword NPT_I in my input everything goes fine.
>>>>> Any ideas...?
>>>>>
>>>>> Best Regards
>>>>> Luca 
>>>>>  
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>  ===== Routine Calling Stack =====
>>>>>
>>>>>             3 velocity_verlet
>>>>>             2 qs_mol_dyn_low
>>>>>             1 CP2K
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> Il giorno lunedì 16 ottobre 2017 09:49:41 UTC+2, Marcella Iannuzzi ha 
>>>>> scritto:
>>>>>>
>>>>>>
>>>>>>
>>>>>> Hi Luca,
>>>>>>
>>>>>> you have set RESTART DEFAULTS as False, hence it is probably not 
>>>>>> restarting the CELL.
>>>>>> I may want to restart it as well.
>>>>>>
>>>>>> Kind regards,
>>>>>> Marcella
>>>>>>
>>>>>>
>>>>>> On Sunday, October 15, 2017 at 7:58:17 PM UTC+2, Luca wrote:
>>>>>>>
>>>>>>> Dear all
>>>>>>> I tried to restart an NPT simulation by putting in my input file the 
>>>>>>> following section:
>>>>>>>
>>>>>>>  &EXT_RESTART ON
>>>>>>>   RESTART_DEFAULT F
>>>>>>>   RESTART_FILE_NAME myjob-1.restart
>>>>>>>   RESTART_POS T
>>>>>>>   RESTART_COUNTERS T
>>>>>>>   RESTART_VEL T
>>>>>>>   RESTART_THERMOSTAT T
>>>>>>>   RESTART_BAROSTAT T
>>>>>>>   RESTART_BAROSTAT_THERMOSTAT T
>>>>>>>  &END EXT_RESTART
>>>>>>>
>>>>>>> Notwithstanding the restart section, I get strong oscillation in the 
>>>>>>> average pressure at the beginning of the restarted simulation (please see 
>>>>>>> the attached). 
>>>>>>> On the contrary, the temperature and the potential energy do not 
>>>>>>> show (as expected) jumps. Could someone point out whether I am missing 
>>>>>>> something in the restart section?
>>>>>>> The simulation is an NPT_I simulation. I have also found the 
>>>>>>> calculation crashes at the first iteration if I try a restart by changing 
>>>>>>> NPT_I with NPT_F.  Is this normal? 
>>>>>>> Do I have to add something else in the restart section?
>>>>>>>
>>>>>>> Luca 
>>>>>>>
>>>>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20171017/628e627c/attachment.htm>


More information about the CP2K-user mailing list