[gmx-users] constraints

[Justin Lemkul]

On 5/12/14, 9:03 AM, Nicola Staffolani wrote:
> Hi Justin,
>
> thank you for your clear reply!
>
> However, it still comes as surprising to me that a "standard" N-H bond
> (entered as a gb_2 one with the same parameters as described in Table
> 2.5.2.1 of the GROMOS96 Manual and User Guide) enters in conflict with the
> standard dt value: indeed dt = 0.002 → tau < 5 · dt and dt = 0.001 → tau <
> 10 · dt (where with tau I have meant the period associated to the harmonic
> frequency) but dt = 0.001 is right the default
> value<http://manual.gromacs.org/online/mdp_opt.html#pp>of dt! So it's
> like who constructed GROMACS didn't know that using the
> default dt value and by just bonding a H atom to a N one would produce a
> WARNING...  What am I missing?
>
> By the way, I have done an estimate of tau = 2 · pi · sqrt(µ/k), where µ =
> m_H · m_N / (m_H + m_N) is the reduced mass of H and N and it looks right
> what the WARNING is asserting...
>

Beware - you can't equate "default" with "standard."  Indeed there are "default" 
values for most settings in the case that the user does not specify values. 
These should not be viewed as "standard" in any case.  Those of us who 
constructed Gromacs have tried to make life relatively simple for the user, but 
we can't think of every possible combination; there's just no way to do that 
from a software engineering perspective.

If you are not using any constraints, you must adjust dt accordingly (0.0005 fs, 
probably).  Most applications will use constraints, as most force fields at 
least constraint X-H bonds nowadays, though the original parametrization may not 
have.

-Justin

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

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

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 601
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

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

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