[gmx-users] External electric field applied to water box

[Alex]

To quickly chime in: we routinely use anisotropic coupling for 
relaxation (constant box size in XY) without any issues. There is a good 
reason for that in our case (an XY-membrane splitting the box mid-Z, 
which we don't want to mangle). Again, no issues whatsoever. Water is 
TIP4P in our case. I suspect issues with setup in your case.

Alex


On 5/22/2019 11:17 AM, Justin Lemkul wrote:
>
>
> On 5/22/19 9:03 AM, Nidhin Thomas wrote:
>> Dear Justin,
>>
>> Thanks a lot for the prompt reply.
>>
>> I ran another simulation with a larger box size with and without 
>> external electric field (field direction is Y-axis). I used 
>> anisotropic pressure coupling for both simulations. Water box without 
>> external electric field had stable system in the beginning but failed 
>> after couple of nano seconds. However, the system with external 
>> electric field deformed continuously. System with EF fails once the 
>> dimension of the box vector (Y-axis) along the direction of electric 
>> field reduces below the minimum box dimension. The error message is 
>> copied below.
>>
>> Fatal error:
>> The Y-size of the box (2.641246) times the triclinic skew factor 
>> (1.000000) is
>> smaller than the number of DD cells (2) times the smallest allowed 
>> cell size
>> (1.322000)
>>
>> I feel that box is shrinking in the direction of applied electric 
>> field (Y-axis) like there is an external compressive stress applied 
>> in that direction. I do not understand how this type of a stress is 
>> generated in the system. I also tried the system with semiisotropic 
>> pressure coupling and the box deformed in Z direction and failed. 
>> When I used isotropic coupling, the system did not fail.
>>
>> I have also shared link to an image that shows how the box vectors 
>> are changing with simulation time. When EF is applied, simulation 
>> fails in few picoseconds. But water box without EF did not fail.
>>
>> Could you please help me understand why system with external electric 
>> field would behave different from water box without EF?
>>
>> https://www.dropbox.com/s/zu7vl2wr99sube2/Box%20vectors%20with%20and%20without%20EF.png?dl=0 
>> <https://www.dropbox.com/s/zu7vl2wr99sube2/Box%20vectors%20with%20and%20without%20EF.png?dl=0> 
>>
>
> I don't know if anyone has ever tested electric fields with 
> anisotropic coupling, but if a system of pure water with no external 
> influence fails with anisotropic coupling, I'd say that is an argument 
> for not using that type of coupling. There is no physical reason why a 
> box of pure water should be treated anisotropically - it is an 
> isotropic medium, and all water ever "sees" via PBC is water. So the 
> box deformation is not physically real or relevant.
>
> -Justin
>