[gmx-users] T-coupling of water and ion

[Jian Zou]

Anton Feenstra wrote:
> zouj01 at mails.tsinghua.edu.cn wrote:
>> tc-grps                  =  RNA SOL_Na
>> tau_t                    =  0.1 0.1
>> ref_t                    =  400 400
>[...]
>> #       System  RNA      SOL     Na      SOL_Na
>>  200    269     405      266     504     266
>[...]
>
>> tc-grps                  =  RNA SOL Na
>> tau_t                    =  0.1 0.1 0.1
>> ref_t                    =  400 400 400
>[...]
>> #       System  RNA     SOL     Na      SOL_Na
>>  200    268     248     268     350     268
>[...]
>
> I don't understand your results. You couple (in all cases) to 400K,
> while your system seems to be about 300K on average. You have
> 'gen_vel=no', meaning you use starting velocities (if any) from your
> input structure file. Where did you get that from?
>
I just generate 8 ADE, dissolve it in a water box, run EM, PR, then MD.
Therefore, the starting structure is taken from the PR-MD result.


> If you do not couple Na+ separately, and it remains bound to the RNA, it
> may not exchange heat enough with the water. This can explain the
> relative high T for your Na+ ions (and for the RNA).
>
This sounds reasonable to me.
But I'm still wondering why couple Na seperately will give the temperature 
for RNA 100 degree C lower than the ref_t.

>> I use different time step and bond option for these 2 tests.
>> And I use afm pulling options.
>> I'm not sure whether these mdrun options will also influence the 
>> T-coupling.
>
> Using 1fs timestep and no constraints vs. 2fs with constraints can
> already give different temperature behaviour, mainly due to more or less
> (roundoff) errors (which can heat), and the effect of constraints (which
> can cool).
>
Oh.


I will run some other test under dt=1fs constraints=all-bonds, and compare 
the results together.


Thank you very much for you advice and comments.


Regards,

Jian Zou