[gmx-users] Re: compressing the system

[Vitaly Chaban]

Dear Moeed:

> Actually I am exactly using the same settings Dr. Chaban suggested because I
> know you are the experts in the field and what Dr. Chaban suggested was
> indeed the only way of compressing my 8 chain-system to the density I want.
> Before that I tried more than 30 NPT (with berendsen) and NVT simulations in
> one week to get a more realistic shape of the chain but they never worked
> (crash) when box becomes small. When you experts suggest sth I follow that
> for sure (especially since unfortunately I am the only person in the group
> dealing with molecular simulations and the archive and literature are my
> only advisors! I myself had no background on MD either when I started...). I
> appreciate your help.
>
> (The only issue I had was unsimilar energy values (KJ/mol).. I just changed
> PME to shift because at the moment I am not doing simulation on our cluster
> and PME runs take weeks to finish on my local machine. so after I used the
> exact settings (with PME) as you suggested I will get back to you about
> energy values).

You probably haven't taken the energies PER ONE MOLE. Please search
the recent messages, I commented this issue already.


> The density of around 800 (SI) is just fine and I am not trying to make the
> system smaller at all. The issue at this time is that surprisingly, when I
> use the same mdp file for *only one single* chain simulation crashes and I
> get the error message that CUTOFF is less than LBOX/.

It is a normal situation. The system compresses to its right density
but the corresponding size of the box becomes too small. Your system
should not be smaller than 2x2x2nm to provide realistic results.


>You (Dr.Chaban)
> suggested 30 bar pressure for 8 chains. Given the error I got I reduced
> pressure to even 1 bar. with this pressure I am not the getting the cutoff
> error but simulations crash ...

With what error, by the way?


> The same problem I had before. ..
> (I also altered tau_p in the range of 0.5 to 10 and tried different
> combinations of tau_p and ref _p but didnt succeed)
>
> Could you please give me comments on appropraite P coupling settings?

We would suggest you not to simulate so small systems with NPT. Try at
least 1000 atoms per system and everything will be ok. Or use NVT to
simulate 1 PE in vacuum. Either make PE chain longer...


> I have also a very important question: I read yesterday somewhere in the
> archive of the year 2007 about validity of the forcefiled at low pressures
> (~1 bar). I am planning to do simulations at very high pressure (200bar) and
> 400K to calculate surface tension and solubility paramater of PE-haxane
> polymer solution. Is there any way of using ff parameters (which are
> calculated for low pressure) at higher pressures?

I would not worry about it .

Dr. Vitaly Chaban