VMD-L Mailing List
From: Chitrak Gupta (chgupta_at_mix.wvu.edu)
Date: Fri Sep 25 2015 - 15:43:45 CDT
- Next message: Mayne, Christopher G: "Re: FFTK Bond optimization"
- Previous message: Chitrak Gupta: "Re: FFTK Bond optimization"
- In reply to: Mayne, Christopher G: "Re: FFTK Bond optimization"
- Next in thread: Chitrak Gupta: "Re: FFTK Bond optimization"
- Reply: Chitrak Gupta: "Re: FFTK Bond optimization"
- Reply: Mayne, Christopher G: "Re: FFTK Bond optimization"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]
Hi Christopher,
Thanks for the explanation.
I am trying to optimize the dihedrals. The current FFTK version shows "QME"
and "MME" along with the fitting (so a total of 3). Could you tell me what
the difference between these are? Also, the QME and MME are varying at some
points. is that normal?
I am attaching a pdf of my plot for your reference. Please note that it
says "Orig" simply because I hit "Run Optimization" instead of "Run
Refinement" during my last run by mistake. This is the result of a few
rounds of optimization.
Regards,
Chitrak.
On Wed, Sep 23, 2015 at 10:56 AM, Mayne, Christopher G <cmayne2_at_illinois.edu
> wrote:
> Chitrak,
>
> I have a question though. When I do the "scan torsion", there is a "Open
>> Torsion Explorer" button that is not there in the video tutorial. Could you
>> tell me what it is supposed to be used for? I was trying to use its
>> "Analyze trajectory" button, but nothing seemed to happen.
>>
>
> The Torsion Explorer is a new tool in VMD 1.9.3 alpha builds that is
> designed to help map dihedral scans to the QM PES. At some point I was
> parameterizing a troublesome molecule and was unable to get the dihedrals
> to fit properly. It was difficult to identify which dihedral was causing
> problems. The Torsion Explorer shows the plot of the QM PES (or whatever
> scan log files are selected) and loads the scan trajectory into VMD. Each
> scan is color coded in the PES; while moving through the trajectory, a
> molecular representation identifying the dihedral that was scanned is shown
> using the appropriate color coding. The Analyze Trajectory button computes
> the current dihedral angle and the slope of the dihedral angle at every
> scan point. This will allow you to see which dihedrals are changing the
> most a particular part of the scan. Typically the dihedral that was being
> scanned should be changing the most. In highly coupled systems, however,
> this may not be the case, and it is often the source of problems when
> trying to fit the overall PES.
>
> The other question I have is, from the video it isn't clear to me how to
>> proceed on the optimization. For example, in the video, one of the
>> dihedrals was duplicated. How do I know which one to duplicate? Also, when
>> the video showed changing the periodicities to 3, is that a default setting
>> (more like the 0.09 charge on aliphatic hydrogens), or was it specific to
>> the molecule being discussed in the example?
>>
>
> The example molecule (pyrrolidine) in the screencasts contains
> sp3-hybridized carbons. If you map the dihedrals between the atoms using
> Newman projections, there are 3 local minima in the torsion (2 gauche and
> one anti); this type of torsional profile is best represented with a
> periodicity of 3.
>
> A quick follow-up on that. Is the "periodicity" the same thing as the
> "multiplicity" in the manual (i.e. the n in the cos (n*phi - delta) term)?
> In that case, I am assuming n has to be adjusted for each one based on the
> molecule at hand, correct?
>
>
> Multiplicity refers to the number of terms that are used to describe a
> single dihedral. Periodicity is the number of minima in the torsion (n).
>
>
> Regards,
> Christopher Mayne
>
> On Sep 22, 2015, at 6:18 PM, Chitrak Gupta wrote:
>
> A quick follow-up on that. Is the "periodicity" the same thing as the
> "multiplicity" in the manual (i.e. the n in the cos (n*phi - delta) term)?
> In that case, I am assuming n has to be adjusted for each one based on the
> molecule at hand, correct?
>
>
> Chitrak.
>
> On Tue, Sep 22, 2015 at 7:06 PM, Chitrak Gupta <chgupta_at_mix.wvu.edu>
> wrote:
>
>> Hi Christopher,
>>
>> Sorry for the late reply. I solved this issue, I basically had two NAMD
>> executables by the same name which was creating confusion. I have it fixed
>> now.
>>
>>
>> I have a question though. When I do the "scan torsion", there is a "Open
>> Torsion Explorer" button that is not there in the video tutorial. Could you
>> tell me what it is supposed to be used for? I was trying to use its
>> "Analyze trajectory" button, but nothing seemed to happen.
>>
>> The other question I have is, from the video it isn't clear to me how to
>> proceed on the optimization. For example, in the video, one of the
>> dihedrals was duplicated. How do I know which one to duplicate? Also, when
>> the video showed changing the periodicities to 3, is that a default setting
>> (more like the 0.09 charge on aliphatic hydrogens), or was it specific to
>> the molecule being discussed in the example?
>>
>>
>> Best regards,
>> Chitrak.
>>
>> On Mon, Sep 14, 2015 at 9:34 AM, Mayne, Christopher G <
>> cmayne2_at_illinois.edu> wrote:
>>
>>> Chitrak,
>>>
>>> From the Opt. Bonded, set the NAMD bin just as you would when trying to
>>> perform the optimization. Open the TkCon (Extensions -> TkConsole) and
>>> type the following:
>>>
>>> exec $::ForceFieldToolKit::BondAngleOpt::namdbin
>>>
>>> Please respond with the output from the tkcon
>>>
>>> Regards,
>>> Christopher Mayne
>>>
>>>
>>>
>>> On Sep 11, 2015, at 8:26 AM, Chitrak Gupta wrote:
>>>
>>> > Hi there!
>>> >
>>> > I am working on the FFTK Bond optimization, and I got this error:
>>> >
>>> >
>>> >
>>> > ------------- Processor 0 Exiting: Called CmiAbort ------------
>>> > Reason: FATAL ERROR: Pair interaction calculation is not supported in
>>> CUDA version
>>> >
>>> > Charm++ fatal error:
>>> > FATAL ERROR: Pair interaction calculation is not supported in CUDA
>>> version
>>> >
>>> > ------------- Processor 0 Exiting: Called CmiAbort ------------
>>> > Reason: FATAL ERROR: Pair interaction calculation is not supported in
>>> CUDA version
>>> >
>>> > Charm++ fatal error:
>>> > FATAL ERROR: Pair interaction calculation is not supported in CUDA
>>> version
>>> >
>>> >
>>> >
>>> > I verified the NAMD binary I selected was namd2 (which is the CPU
>>> version). Is there some other setting to get it to use the CPU version?
>>> >
>>> >
>>> >
>>> > Regards,
>>> > Chitrak.
>>>
>>>
>>
>
>
- application/pdf attachment: multiplot.pdf
- Next message: Mayne, Christopher G: "Re: FFTK Bond optimization"
- Previous message: Chitrak Gupta: "Re: FFTK Bond optimization"
- In reply to: Mayne, Christopher G: "Re: FFTK Bond optimization"
- Next in thread: Chitrak Gupta: "Re: FFTK Bond optimization"
- Reply: Chitrak Gupta: "Re: FFTK Bond optimization"
- Reply: Mayne, Christopher G: "Re: FFTK Bond optimization"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] [ attachment ]