[gmx-users] Cutoff influence on performance

[Justin Lemkul]

On 12/18/12 1:22 PM, zugunder wrote:
> Hi,
>
> I have a general question on cutoff values.
> I am using an AMBER03 ff and in a number of papers/tutorials the cutoffs for
> vdw/coulomb are reported as low as 0.8 (and up to 1.2 - 1.4 depending on the
> interaction type). As far as I understand, increasing this value a bit will
> not harm unless there are no errors and may increase the accuracy and
> reliability of the calculations. But how do the changes of this parameter
> affect computation performance? If I, say, set it 1.0, will I see any
> significant difference in speed compared to 0.8?
>

It is absolutely untrue that increasing the cutoff values will improve accuracy. 
  On the contrary, you may be breaking the physical model entirely.  Use what is 
prescribed by the original force field literature or any newer, fully validated 
protocol that demonstrates the new cutoff settings and/or algorithms are more 
suitable.

In general, larger cutoffs run slower because more neighbor searching is done. 
Do not base your choice of cutoffs on desired performance.

-Justin

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

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

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