- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]

From: Tom Kurtzman <simpleliquid.gmail.com>

Date: Thu, 8 Sep 2016 11:04:18 -0400

Hi Shunsuke, glad to see you are using GIST :). Just a couple of things

to add.

First, I assume you want to get the reference densities (used as a

reference for the translational entropy calculations) and the energies at

different temperatures. For both these quantities, you should use the

values over a reasonably long run for the ENTIRE SYSTEM. That is, take

the number of particles and divide by the total energy of the system to get

the average energy (you can then average this quantity over all the frames

to get good sampling). Do the same for the Number of particles and Volume

of the system to get the densities. You can get estimated errors for

these quantities by doing the analysis all stored frames of the

simulation. Don't use individual voxels for determining the reference

values since you'll just get worse sampling from them. However, for a

given voxel dimension, you can get an idea of the error in each voxel based

on deviations of the voxel quantities (energies and densities) from the

well determined system average. Since the system is homogeneous, the

voxel quantities should fluctuate about the system averages.

As Prof. Case said, a box length of 20-22 Angstroms should be sufficient.

One way to make sure your box is big enough uses the fact that the

quantities you are calculating should not be dependent on the box size. In

other words, you don't want any finite size system effects.

Tom

On Thu, Sep 8, 2016 at 8:02 AM, David A Case <david.case.rutgers.edu> wrote:

*> On Thu, Sep 08, 2016, Shunsuke Mizutani wrote:
*

*> >
*

*> > I am trying to calculate my own reference data of TIP3P water model at
*

*> > several temperatures.
*

*> > And I am wondering what conditions (box size, grid size, Integer number
*

*> of
*

*> > grid increment etc. ) I should set for the calculation.
*

*>
*

*> The box size is up to you, but the minimum should be about 20-22 Ang. on a
*

*> side. For pure water, that should be plenty large enough.
*

*>
*

*> Use the defaults for everything else. Number of grids, etc. is internally
*

*> calculated based on the cutoff you request. Of course, after you do an
*

*> intial
*

*> calculation, you can play with these values if you want to investigate what
*

*> effects changes will have.
*

*>
*

*> ...dac
*

*>
*

*>
*

*> _______________________________________________
*

*> AMBER mailing list
*

*> AMBER.ambermd.org
*

*> http://lists.ambermd.org/mailman/listinfo/amber
*

*>
*

Date: Thu, 8 Sep 2016 11:04:18 -0400

Hi Shunsuke, glad to see you are using GIST :). Just a couple of things

to add.

First, I assume you want to get the reference densities (used as a

reference for the translational entropy calculations) and the energies at

different temperatures. For both these quantities, you should use the

values over a reasonably long run for the ENTIRE SYSTEM. That is, take

the number of particles and divide by the total energy of the system to get

the average energy (you can then average this quantity over all the frames

to get good sampling). Do the same for the Number of particles and Volume

of the system to get the densities. You can get estimated errors for

these quantities by doing the analysis all stored frames of the

simulation. Don't use individual voxels for determining the reference

values since you'll just get worse sampling from them. However, for a

given voxel dimension, you can get an idea of the error in each voxel based

on deviations of the voxel quantities (energies and densities) from the

well determined system average. Since the system is homogeneous, the

voxel quantities should fluctuate about the system averages.

As Prof. Case said, a box length of 20-22 Angstroms should be sufficient.

One way to make sure your box is big enough uses the fact that the

quantities you are calculating should not be dependent on the box size. In

other words, you don't want any finite size system effects.

Tom

On Thu, Sep 8, 2016 at 8:02 AM, David A Case <david.case.rutgers.edu> wrote:

-- ************************************************ Tom Kurtzman, Ph.D. Assistant Professor Department of Chemistry Lehman College, CUNY 250 Bedford Park Blvd. West Bronx, New York 10468 718-960-8832 http://www.lehman.edu/faculty/tkurtzman/ <http://www.lehman.edu/faculty/tkurtzman/index.html> ************************************************ _______________________________________________ AMBER mailing list AMBER.ambermd.org http://lists.ambermd.org/mailman/listinfo/amberReceived on Thu Sep 08 2016 - 08:30:03 PDT

Custom Search