Strange behaviour by molecule in Electric field

[Pankaj Mishra]

Dear Matt,

This is really a serious bug. Now the reason is clear to me for translating 
the system in -ve direction of the field which i was trying to figure out 
fundamentally from last more than 2 months. After recompiling the code with 
this change, My system (Single Hydrogen atom in constant EF and single 
water molecule in Gaussian Pulse both for non-periodic box) is 
translationally invariant  in presence of EF. This is  really helpful. 

Thank you so much 

Pankaj

On Friday, 8 November 2013 19:14:32 UTC+1, Matt W wrote:
>
> Hi,
>
> You make a very good point! Now for your argument to hold your system 
> needs to be translationally invariant. There are two reasons (at least) why 
> it is not.
>
> (i) the EFIELD is non periodic (i.e. a saw tooth), so there is a huge 
> field at the cell edge. You need to centre your molecule in the box.
> (ii) your CUTOFF is not truly sufficient if you really want to see the 
> molecule stay centred - there's a post from Flo discussing this concerning 
> vibrational analysis from a month or so ago, but this is a minor point.
>
> BUT, I think you have also identified a bug. According to me, there is an 
> inconsistency between the way forces due to the field act on atoms cores 
> and electrons.
>
> changing line 233 (or thereabouts) of efield_utils.F to
>
> force(ikind)%efield(:,iatom)=force(ikind)%efield(:,iatom)-field*zeff
>
> i.e. changing the sign of the forces on the cores, seems to give good 
> conservation of force in my quick tests (plus comments above). You'll have 
> to recompile, of course.
>
> Let us know if this seems to work for you (it is possible that the force 
> on the electrons rather than the core should be inverted, but at the moment 
> they are both being pushed in the same direction).
>
> Matt
>
> On Friday, 8 November 2013 09:10:51 UTC, Pankaj Mishra wrote:
>>
>> Hello,
>>
>> As per my fundamental knowledge, polar molecule can rotate in electric 
>> field, not translate. Being more precise, here i am talking about 1 
>> molecule in non-periodic box. I didn't find any force to move in the field. 
>>
>> Even non-polar molecule and single atom (H and Ne) are also translating 
>> in EF in CP2K. In the case of water cluster, the center of mass of whole 
>> system is translating. I have not found such behavior in other package like 
>> GAMESS .
>>
>> This is a very serious issue for me. Please try to suggest something if i 
>> am missing anything in my input.
>>
>> Thank you
>>
>> Pankaj 
>>
>> On Thursday, 7 November 2013 20:23:58 UTC+1, Matt W wrote:
>>>
>>> Hi,
>>>
>>> just wondering: why would you not expect a polar molecule to move in an 
>>> electric field?
>>>
>>> Matt
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20131112/b134605e/attachment.htm>