Re: [AMBER] Collective Variable not "Strictly" Enforced: Adaptively Biased MD Simulations of type "COM_ANGLE" using "FLOODING"

From: Feng Pan <fpan3.ncsu.edu>
Date: Fri, 2 Dec 2016 15:26:25 -0500

Hi, Emmanuel

In your situation, I think the problem is that your initial value of CV is
over 0.7. One possible way to improve this is to first steer the CV to a
lower value using SMD.
In your case the flooding rate seems not so fast because you only get 0.6
to 0.9, so increasing the resolution to a higher value first or decreasing
the timescale can be a good choice to accelerate the
flooding rate so that the whole CV range can be fully sample quickly. The
you can slower the flooding rate gradually to refine the sampling.

To restrain the CV COM_ANGLE in a specific range, if you are using sander,
one way is to use the NMR refinement, which is compatible with ABMD, check
the section 24.1.1 in the manual.
The other way I am developing now is to use nfe_pmd, which now you can
restrain the CV to one position but I will extend it so that you can
restrain it to a specific range. Also this will be available in PMEMD.
This should be available soon in this December.

Best
Feng Pan


On Thu, Dec 1, 2016 at 1:08 AM, Emmanuel <emmanuel.salawu.gmail.com> wrote:

> Dear Amber Users,
>
> Please, kindly help/guide me.
>
> Question in Brief
> ==============
> Is there any way I can (further) strictly enforce the range of the
> collectively (CV) in Adaptively Biased MD simulation? If it were to be
> Steered Molecular Dynamics Simulations, I would be able to increase the
> force/spring constant and drive/steer the system as I desire. Is there a
> way to achieve a similar thing in ABMD? In other words, is there any
> variable in ABMD that is analogous to the Force/Spring Constant in SMD and
> is tunable (so that I can adjust it to more strictly enforce the range I
> specified in the CV)?
>
> I am very sorry that my question is rather too naive (maybe even stupid).
> Please, also look at the additional details I provide below. Thank you very
> much.
>
>
> More Details
> ==========
> I am trying to do Adaptively Biased MD Simulations (of type "COM_ANGLE"
> using "FLOODING") and it appears that the range I specified for the
> collective
> variable (CV) not "strictly" enforced. For example, I want to adaptively
> force my system between cv_min = 0.3 and cv_max = 0.7 at resolution = 0.05.
> (All in radians). However, when I check the monitor file, "monitor.txt", I
> do NOT see my system to maintained within (or at least approximately
> within) my desired range for the CV. For example, the CV dumped in the
> monitor file has some values as high as 0.82 (which above 0.7 I specified)
> and there is no value of the dumped CV near 0.3 that I specified (the least
> value of the dumped CV I have seen so far is approximately 0.607). In other
> words, I want the system to sample between 0.3 and 0.7 radians, but the
> ABMD appears to be sampling between 0.6 and 0.9 radians.
>
>
> Please, kindly guide me. The content of my input file "cv.in" is shown
> below. I also include a section of my output/monitor file that contains
> some value of the sampled CV.
>
>
> ++++++++++++++++++++++
> #cv.in FILE#
> &colvar
> cv_type = 'COM_ANGLE'
> cv_min = 0.3, cv_max = 0.7, resolution = 0.05
> cv_ni=6
> cv_i = 784, 0, 780, 813, 0, 818
> /
> ++++++++++++++++++++++
>
> #
> # MD time (ps), CV #1, E_{bias} (kcal/mol)
> #
> 0.9990 0.8214852146 0.0000000000
> 1.9990 0.8203868858 0.0000000000
> 2.9990 0.7514800469 0.0000000000
> 3.9990 0.7794953384 0.0000000000
>
> ...
>
> 242.9990 0.7524856810 0.0000000000
> 243.9990 0.7634033212 0.0000000000
> 244.9990 0.6801598169 0.0257862627
> 245.9990 0.8028405691 0.0000000000
> 246.9990 0.6577946805 0.0127113489
> 247.9990 0.7853694051 0.0000000000
>
> ...
>
> 1328.9990 0.8280800141 0.0000000000
> 1329.9990 0.7782295206 0.0000000000
> 1330.9990 0.6904481393 0.5377608131
> 1331.9990 0.7436914461 0.0000000000
> 1332.9990 0.8638745221 0.0000000000
> 1333.9990 0.7542451081 0.0000000000
> 1334.9990 0.6696472834 0.4234273988
>
> ++++++++++++++++++++++
>
>
> With regards
> Emmanuel Salawu
> emmanuel.salawu.gmail.com
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>



-- 
Feng Pan
Ph.D. Candidate
North Carolina State University
Department of Physics
Email:  fpan3.ncsu.edu
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Dec 02 2016 - 12:30:02 PST
Custom Search