[CP2K:5392] How can I get the temperature converged in MD calculation (NVT) for system having H atoms?

[Jingyun Ye]

Dear Ari

I really thank you for your suggestion. I tried TIMESTEP=0.1, and 
TIMECON=10 or 1 which controls the temperature very well. But do you think 
I need to test the TIMESTEP by doing NVE calculation first? 
Actually, I tried NVE calculation by using TIMESTEP=1.0, 0.5 or 0.1. Which 
one do you think more efficient for my system? I can not see the 
difference. Why does the temperature drop down to 150 K? 
Here is the input:
&MOTION
 &MD
  ENSEMBLE NVT
  STEPS 10000
  TIMESTEP 0.1
  TEMPERATURE 300
&END
&END

The results shows bellow. 

<https://lh4.googleusercontent.com/-RHvKvzFvr1w/U5zG1T5uORI/AAAAAAAAALk/InVHsOA2q4M/s1600/1.1.png> 
<https://lh5.googleusercontent.com/-_ZCNYINTMIs/U5zG-Ha47PI/AAAAAAAAALs/E4N-K9Q3kOM/s1600/2.1.png> 
<https://lh4.googleusercontent.com/-CxehrpNtGbY/U5zHBTLaM1I/AAAAAAAAAL0/UQEY9A9V1Zs/s1600/3.1.png>
TIMESTEP=1.0













<https://lh3.googleusercontent.com/-Gw6bi0OBboo/U5zHZPqyIII/AAAAAAAAAL8/C1Pg3MX8OuY/s1600/1.2.png> 
<https://lh4.googleusercontent.com/-4uzX3hztoZQ/U5zHdxesxxI/AAAAAAAAAMM/KBQy-oW6PMM/s1600/3.2.png> 
<https://lh5.googleusercontent.com/-cLBRyLKattA/U5zHbpKiXTI/AAAAAAAAAME/l0HFcTq8BZk/s1600/2.2.png>
TIMESTEP=0.5












<https://lh6.googleusercontent.com/-7BAs-M00rn4/U5zHrjPPHPI/AAAAAAAAAMU/bXxVea9TNmI/s1600/1.png> 
<https://lh4.googleusercontent.com/-G016MkaZ-xg/U5zHx-pULQI/AAAAAAAAAMk/zE4wwcctgtc/s1600/3.png> 
<https://lh6.googleusercontent.com/-duQ57c1oeIE/U5zHvuLzDrI/AAAAAAAAAMc/CZav5HK047M/s1600/2.png>
TIMESTEP=0.1














On Wednesday, June 11, 2014 5:07:24 PM UTC-4, Ari Paavo Seitsonen wrote:
>
>
> Dear Jingyun Ye,
>
>   If you start from a more or less optimised structure about half of the 
> initial kinetic energy first gets converted into an increase of potential 
> energy, and the thermostat is a bit slower: Your time constant now is 100 
> fs, but this does not mean that ionic temperature would reach this 
> temperature within the time constant (small value means faster 
> equlibration, at least in principle; if you use a value that does not 
> couple to the system - too small for example - then there will not occur 
> heat transfer). Rather it is a "characteristic frequency". If one needs to 
> equilibrate the initial part faster, one can try with the CSVR thermostat 
> and a smaller time constant, and later either increase the time constant or 
> then shift to the Nosé-Hoover thermostat chain.
>
>   By the way, 0.1 fs is quite small a time step - do you need such a small 
> value? Well, it should certainly be a safe choice for any kind of system.
>
>     Greetings,
>
>        apsi
>
> 2014-06-11 22:15 GMT+02:00 Jingyun Ye <jing... at gmail.com <javascript:>>:
>
>> Hi, all
>>
>> I am running a MD calculation for searching the most stable structure at 
>> room temperature. But the temperature goes from 300K to 150K after 1000 
>> steps and the energy goes up. My question is how can I get the temperature 
>> converged? Is there any thing wrong in my input? What is the reasonable 
>> TIMESTEP for the system having H atoms in it? Small TIMECON and big 
>> TIMECON, which one is more efficient for heat transfer? 
>>
>> I included the input and results. 
>>
>> Thanks very much.
>> &MOTION
>>  &MD
>>   ENSEMBLE NVT
>>   STEPS 10000
>>   TIMESTEP 0.1
>>   TEMPERATURE 300
>>   &THERMOSTAT
>>    TYPE NOSE
>>     &NOSE
>>      TIMECON 100
>>     &END NOSE
>>   &END THERMOSTAT
>>  &END MD
>> &END MOTION
>>
>>  -- 
>> 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+... at googlegroups.com <javascript:>.
>> To post to this group, send email to cp... at googlegroups.com <javascript:>
>> .
>> Visit this group at http://groups.google.com/group/cp2k.
>> For more options, visit https://groups.google.com/d/optout.
>>
>
>
>
> -- 
>
> -=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-
>   Ari Paavo Seitsonen / Ari.P... at iki.fi <javascript:> / 
> http://www.iki.fi/~apsi/
>   Institut für Chemie der Universität Zürich
>   Tel: +41 44 63 55 44 97  /  Mobile: +41 79 71 90 935
>  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20140614/fb2b9e1b/attachment.htm>