[gmx-users] water box with prescribed volume and density

[Justin A. Lemkul]

grivet wrote:
> Thank you for your quick and detailed answer.
> 
>> First, begin by using the right genbox flags :)  The -nmol flag is for insertion 
>> of extra molecules (specified in -ci).  This operation is not what you want.  If 
>> you wish to cap the number of solvent molecules, you want the -maxsol option.
>>
>>> By the way, I notice that the number of molecules in topol.top is not
>>> updated; is that normal ?
>> Given your commands, yes.  You are not passing your topology to genbox using -p. 
>>   If you do, it will happily update it for you.
>>
> I followed your prescriptions with succes. 
> 
>> Also understand that the number of water molecules present in the box is 
>> dependent upon other species (i.e., solute).  If you are trying to compare 
>> results with some other system (with a different solute), the space available 
>> for solvent may be different, and really is an irrelevant consideration. 
>> Provided you have a reasonable unit cell that does not violate the minimum image 
>> convention, and you use the same run parameters, then the comparison should be 
>> legitimate.  If you simply wish to run a new simulation from a previous 
>> configuration, there is no need to re-build the system; just start a new 
>> simulation with different starting velocities.
> 
> No, I simply want to simulate a box of water and compare results with
> those obained with GROMOS and spc water.
> 
> Having cleared the first hurdle, I ran mdrun for steepest descent (5832
> molecules). The program stopped after 2067 steps, with maximum force =
> 1.58E02 on atom 4870. How important is that warning?
> 

What warning?  This one?

http://www.gromacs.org/Documentation/Errors#Stepsize_too_small.2c_or_no_change_in_energy._Converged_to_machine_precision.2c_but_not_to_the_requested_precision

> Anyway, I started mdrun again for a dynamics run (0.5 fs step, 25 ps
> long, temperature and pressure cooupling with Berendsen thermostats).
> How can I judge the quality of results ?
> 

Evaluate the resulting observable parameters and compare them to what is 
expected, either experimentally or from other simulations.  See the manual for a 
list of analysis tools and what they can do.

> The temperature quickly evolved from 300K to the prescribed 270 K and
> then fluctuated by 2-3 K. (One would think that velocities are generated
> according to the reference temperature, 270K?)
> 

The initial temperature will depend on the quality of the initial configuration 
(i.e., if two particles are extremely close, they may spin off a bit from each 
other, causing a small temperature spike), application of constraints, etc.

> The pressure fluctuated a lot and is -179 at the end of the run; what is
> the significance of this result?
> 

http://www.gromacs.org/Documentation/Terminology/Pressure

The end value is far less significant than the average value.  See the link for 
notes on what dictates instantaneous pressure.

> The volume and density evolved rather smoothly to reasonable values. I
> have two values for the density; according to the .log file, it is
> 0.9937E03 and according to g_energy, it is 1.015E03; which should I
> believe?
> 

Probably g_energy.  See the manual, Appendix C.  How many values in the .log 
file are you looking at?  Anything printed there is just an instantaneous value, 
and may not represent the actual average over the simulation.

> Can I consider my system to be fully equilibrated? Since the run lasted
> 50 mn on a signle node, I cans easily extend the equilibration. A last
> point is not clear; which file should I start from to extend the
> computation: .cpt, .trr, or .xtc?
> 

http://www.gromacs.org/Documentation/How-tos/Extending_Simulations

The exact time necessary will depend on how well-converged all of your desired 
observables are.  For pure water, this shouldn't take long, maybe a few hundred 
ps at most.

-Justin

> Thank you again for taking pains with an ignorant user
> JP Grivet
> 
> 

-- 
========================================

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
MILES-IGERT Trainee
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================