[gmx-users] constraint pull code / gromacs 5.0.4

[Justin Lemkul]

On 6/23/15 4:28 PM, Alex wrote:
>>
>>
>> It's dangerous to make broad generalizations like "constraints should be
>> deprecated."  Sometimes constraints are not appropriate (materials,
>> crystals, etc).  But for the vast majority of biomolecular simulations, we
>> really don't care about different vibrational states of bonds involving H,
>> so there's zero purpose to running our simulations 4x slower just to
>> collect information we don't need :)
>>
>>
>> Here he is. :) My not entirely serious comment aside (and not to hijack
> the thread), how is the analytical calculation of forces in harmonic
> springs computationally heavier than something like SHAKE or LINCS? Are you
> guys doing something algorithmically superfancy with LINCS? I recall
> implementing both SHAKE and LINCS and for the life of me can't see where
> the x4 factor comes from.
>

It has nothing to do with the computational cost of bonds vs. constraints; it's 
the time step.  Without constraints, dt = 0.5 fs (maybe 1.0 fs if you're lucky) 
and with constraints (I don't care which algorithm, but LINCS is generally more 
stable), dt = 2.0 fs.

-Justin

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

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

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