From: Jim Phillips (jim_at_ks.uiuc.edu)
Date: Wed Nov 09 2016 - 16:48:00 CST
The rule NAMD uses is that a DCD frame is written when the corresponding
timestep is first reached. This avoids re-writing the initial state or
repeating frames when "run 0" is called. An explicit command to append a
frame to a DCD file would be useful.
Jim
On Wed, 9 Nov 2016, Brian Radak wrote:
> Ah yes, so this is a problem that I understand - I just gave the REMD
> developers the benefit of the doubt that they had this working as well. When
> I implemented simulated tempering, the solution was to invert the cycle so
> that the exchange is attempted first and then MD is run. This ensures that
> the final frame (which must be a multiple of DCDFreq) is drawn with the
> reported Hamiltonian parameters.
>
> HTH,
>
> Brian
>
>
> On 11/09/2016 02:05 AM, Norman Geist wrote:
>>
>> Thanks Brian,
>>
>> I know about the „output” command but wanted to avoid using 1. Many
>> thousand files and 2. Using external tools from within namd (random
>> segfault issue with REMD and GPUs). Looks I will have to postprocess my
>> trajectories to extract the desired frames afterwards. But anyway the REMD
>> algorithms should actually output frames after exchange attempts, not
>> before as it currently is, this is hard to do with namd and dcdfreq and is
>> somewhat wrong. This is because an accepted replica by the metropolis
>> criterion should contribute its current state to the accepted temperature,
>> not the one it was coming from. Depending on the length of one run, the
>> sampling error can be very high, which can result in non-boltzmann
>> statistics at the respective temperature trajectories.
>>
>> Since there are so many TCL commands related to multicopy algorithms, there
>> is likely an interest in having such a “dcdframe” command aswell.
>>
>> Norman Geist
>>
>> *Von:*owner-namd-l_at_ks.uiuc.edu [mailto:owner-namd-l_at_ks.uiuc.edu] *Im
>> Auftrag von *Brian Radak
>> *Gesendet:* Dienstag, 8. November 2016 16:22
>> *An:* namd-l_at_ks.uiuc.edu; Norman Geist <norman.geist_at_uni-greifswald.de>
>> *Betreff:* Re: namd-l: Trigger DCD frame from TCL?
>>
>> I think the DCD standard expects/requires evenly spaced frames, so,
>> probably not?
>>
>> A close alternative is the "output <basename>" command, which triggers a
>> writing of coordinates, velocities, and extended system files based on the
>> provided basename. It's not pretty, but you can paste together a bunch of
>> binCoordinate files with catdcd when you are done.
>>
>> HTH,
>>
>> Brian
>>
>> On 11/08/2016 09:12 AM, Norman Geist wrote:
>>
>> Hey there,
>>
>> is there any TCL command that can be used to trigger a DCD frame,
>> without using dcdfreq?
>>
>> --
>> Brian Radak
>> Postdoctoral Appointee
>> Leadership Computing Facility
>> Argonne National Laboratory
>>
>> 9700 South Cass Avenue, Bldg. 240
>> Argonne, IL 60439-4854
>> (630) 252-8643
>> brian.radak_at_anl.gov <mailto:brian.radak_at_anl.gov>
>>
>
> --
> Brian Radak
> Postdoctoral Appointee
> Leadership Computing Facility
> Argonne National Laboratory
>
> 9700 South Cass Avenue, Bldg. 240
> Argonne, IL 60439-4854
> (630) 252-8643
> brian.radak_at_anl.gov
>
This archive was generated by hypermail 2.1.6 : Sun Dec 31 2017 - 23:20:47 CST