On 16/02/2012 3:14 PM, Juliette N. wrote:
On 15 February 2012 23:05, Mark Abraham <mark.abra...@anu.edu.au <mailto:mark.abra...@anu.edu.au>> wrote:On 16/02/2012 2:08 PM, Juliette N. wrote:On 15 February 2012 21:00, Mark Abraham <mark.abra...@anu.edu.au <mailto:mark.abra...@anu.edu.au>> wrote: On 16/02/2012 12:22 PM, Justin A. Lemkul wrote: Juliette N. wrote: Hi all, I am trying to run simulation in vaccum using the the changes shown below to the usual mdp file. pbc = no ;coulombtype = PME ;vdw-type = Shift ; Cut-offs rlist = 0 rcoulomb = 0 rvdw = 0nstlist = 0 ns_type = simpleCan anyone help me with some short questions please? 1- for pbc=no, I need to comment ;coulombtype = PME ;vdw-type = Shift so it defaults to vdw-type = Cut-off which are not suitable algorithms. Is using cut offs justified for in vacu runs? Plain truncations in condensed-phase systems lead to artifacts. Neither of those conditions apply here, as you're using infinite cutoffs. 2- I am not clear about using infinite cutoffs. Why one refers to infinite cutoffs when rlist = 0 rcoulomb = 0 rvdw = 0 ? My understanding is that this settings means zero cutoff i.e no interaction is calculated. Why does this setting refer to infinite rc? That's the way the code works. There are various parameters that can be set to -1, for instance, and that doesn't mean quantities are calculated every -1 steps ;) Setting cutoffs to zero in this manner mean *all* interactions are calculated, not none. Prove it to yourself with a zero-step MD run. The nonbonded energy terms will not be zero, as they would in the case that no interactions would be calculated. Or read about pbc=no in manual section 3.4.9 like I suggested Juliette do earlier this week... Thanks Justin and Mark. I think you meant 7.3.9 which I did when you referred me to that. My problem was that I did not expect rc=0 is *just defined* as infinite cutoff in gromacs. To me rc=0 looked more equivalent to no interaction than infinite cutff off (all interactions).Sure, but reading the documentation is usually a better idea than making assumptions :) The underlying reason for this behaviour is that it is much easier for the person writing the code to have one parameter that occasionally has a "special" meaning when it takes a nonsense value (like rc<=0) then it is to have a slew of parameters that have to be managed when they are input (and checked, and documented) and then possibly passed through a cascade of functions (lots of bureaucracy and chances to make errors) before they are used. The alternative costs the programmer more time. In an ideal world there would be an infinite amount of such time, but given the amount most people are prepared to pay for scientific software, that time is severely limited.And also I dont see why do we need to change rc to infinite. I mean if force fields dictate cutoffs based on a distance where nonbonded interactions are close enough to zero (negligible), what purpose use of infinite cutoff serve?Efficiency, like I said in the first post in this thread. Given that your force field was parametrized with given cut-offs for the condensed phase, to what purpose do you wish to calculate in vacuo? The perturbation from calculating in vacuo will be much larger than the perturbation from the use of infinite cut-offs.Thank you. I am looking at potential of a single molecule in vacu for heat of vap purposes at different temperatures by changing ref_temp and gen_temp for each run.
So it's much more bureaucracy every nstlist>0 steps to take your N atoms and look at their distance from the other N-1 atoms and make lists of which ones are inside rc than it is to just compute them all every step and know that if they're further than rc then the effect is tiny. For small enough N, the later is guaranteed to be faster...
Mark
-- gmx-users mailing list gmx-users@gromacs.org http://lists.gromacs.org/mailman/listinfo/gmx-users Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/Search before posting! Please don't post (un)subscribe requests to the list. Use the www interface or send it to gmx-users-requ...@gromacs.org. Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
