[gmx-users] Maximun time-step recommended in the grompp part for each system

[Justin Lemkul]

On 12/13/14 11:34 AM, Mario Fernández Pendás wrote:
> Some extra information about this topic:
>
> We are exicitely interested in NOT using any kind of constraints. As you
> may assume, the purpose of this study is purely theoretical.
>

As the comments at the top of check_bonds_timestep() note, it is not a check for 
the validity of the integration, more like a guideline to prevent people from 
incompatible settings.  If you are not using constraints, the upper limit of dt 
is likely 1 fs, but depending on what you're simulating, you may have to use a 
much lower value.  A few trial runs to check energy conservation and stability 
would likely solve this very quickly.

-Justin

> Thank you very much.
>
> Best regards,
> Mario
> El 11/12/2014 18:20, "Mario Fernández Pendás" <mariofp77 at gmail.com>
> escribió:
>
>> Dear all,
>>
>> Taking look at the check_bonds_timestep() function in src/kernel/grompp.c
>> used for checking the time-step used for the integration of the equations
>> of motion a few questions come to me:
>>
>> - In the comments of the function it is said that "The stability limit of
>> leap-frog or velocity verlet is 4.44 steps per oscillational period". Does
>> anyone know how this exact number is obtained?
>>
>> - I see that, inside of the function, there are two "limits": 5 and 10.
>> They are used for establishing a relation between the estimated oscillation
>> period of the bond between two atoms of the system in use and the time-step
>> used for integrating the equations of motion and giving a warning message
>> in case of using a non suitable time-step. I assume, looking at the
>> literature, that this limits are taking from the following paper by Mazur:
>> http://www.sciencedirect.com/science/article/pii/S0021999197957405
>> Am I right?
>>
>> I have also found an issue in redmine for the function
>> check_bonds_timestep() reporting that it only reports the first suspicious
>> bond. I think this does not affect my questions.
>>
>> Thank you very much.
>>
>> Best regards,
>> Mario
>>

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

Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

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