From: Jeff Comer (jeffcomer_at_gmail.com)
Date: Thu May 25 2017 - 16:06:33 CDT
Hi,
Each replica produces its own gradient, count, and PMF files, which are
synchronized every shareFreq steps. For usual values of shareFreq, these
files should be very similar. How different they are depends on shareFreq,
which is the number of steps between synchronization of the replicas. If I
remember correctly, output of the .grad, .count, etc. files happens before
synchronization, so if shareFreq and colvarsRestartFrequency are the same,
your files will be different by shareFreq samples.
Jeff
–––––––––––––––––––––––––––––––––––———————
Jeffrey Comer, PhD
Assistant Professor
Institute of Computational Comparative Medicine
Nanotechnology Innovation Center of Kansas State
Kansas State University
Office: P-213 Mosier Hall
Phone: 785-532-6311
Website: http://jeffcomer.us
On Thu, May 25, 2017 at 5:02 AM, Souvik Sinha <souvik.sinha893_at_gmail.com>
wrote:
> While running MW_eabf, I am actually getting 4 sets " grad " files for each
> replica [ e.g. colvar2.1.grad, colvar2.1.zgrad, colvar2.1.czar.grad
> & colvar2.eabf.1.grad ] & 3 sets of " count " files [colvar2.1.count,
> colvar2.1.zcount & colvar2.eabf.1.count]. There are significant difference
> in the values of those files.
>
> I understand that " *.zcount " is the z-histogram & " *.zgrad " contains
> z-averaged restraints forces. But, what is need of the other sets ? I
> guess, one of them is related to Z-Y estimator but which one is it ?
>
> I found " *outputName*.czar.grad: current estimate of the free energy
> gradient (grid), in multicolumn ". Then where this file is differing from
> outputName.[ReplicaID].grad & outputName.eabf.[ReplicaID].grad ?
>
> Part of the configuration file (related to input-output) is here :
>
> set outName colvar2
>
> structure vacuum.psf
> parameters common/par_all22_prot.inp
>
> paraTypeCharmm on
>
> coordinates equilvac.coor
> velocities equilvac.vel
>
>
> outputname $outName.[myReplica]
> restartname $outName.[myReplica]
>
> *****
> ***
> source ../eabf.tcl
> set eabf_inputname 0 ;# restart file name or
> "0"
> set eabf_outputname colvar2.eabf.[myReplica] ;# restart file name
> set eabf_temperature 300
> set eabf_outputfreq 10000
>
> **************************
> *****************
>
> # ABF
>
> colvars on
> colvarsConfig colvars2.in
>
> ##### Replica Exchange ####################
>
>
> source /apps/NAMD_2.12_Linux-x86_64-multicore/lib/selectionRules.tcl
> source /apps/NAMD_2.12_Linux-x86_64-multicore/lib/resampleWalkers.tcl
> source /apps/NAMD_2.12_Linux-x86_64-multicore/lib/minExchanges.tcl
>
>
> firsttimestep 0
> replicaUniformPatchGrids on
> set m 1000
> set sharedFreq 10000
> for {set i 0} {$i < $m} {incr i} {
> run $sharedFreq
> cv bias abf1 share
> }
>
>
> *************************************************************
>
>
> Please help me out.
>
> Thank you.
>
>
> On Fri, Apr 28, 2017 at 10:57 AM, Souvik Sinha <souvik.sinha893_at_gmail.com>
> wrote:
>
> > Thanks for the reply.
> >
> > On Fri, Apr 28, 2017 at 2:30 AM, Jérôme Hénin <jerome.henin_at_ibpc.fr>
> > wrote:
> >
> >> Hi Souvik,
> >>
> >> Indeed, you can always find the current documentation here:
> >> http://colvars.github.io/
> >>
> >> Multiple-walker eABF works fine. To estimate the free energy, you can
> >> merge CZAR results from different walkers. For that, run a single eABF
> >> simulation for zero steps, providing the output from all walkers as
> >> inputPrefix
> >> <http://colvars.github.io/colvars-refman-namd/colvars-
> refman-namd.html#x1-590006.1.2>.
> >> To get all the necessary input, you need to enable writeCZARwindowFile
> >> <http://colvars.github.io/colvars-refman-namd/colvars-
> refman-namd.html#x1-640006.2.1> during
> >> your calculations (if you forgot, you can also extract this information
> >> from state files, but that will an extra step).
> >>
> >> The NAMD config for combining might look like this, using Tcl to build
> >> the list of files:
> >>
> >> set f base_name
> >> set n_replicas 42
> >> set input ""
> >>
> >> for {set i 0} {$i < $n_replicas} {incr i} {
> >> append input "$f.$i "
> >> }
> >>
> >> colvars on
> >> cv config "
> >> (...)
> >> abf {
> >> inputPrefix $input
> >> (...)"
> >>
> >> Alternately, for the Zheng/Yang estimator, see the -mergemwabf option in
> >> the documentation. To do that, remember that you need to enable the
> >> Zheng/Yang estimator when running all the calculations!
> >>
> >> Best,
> >> Jerome
> >>
> >>
> >> On 26 April 2017 at 13:17, Souvik Sinha <souvik.sinha893_at_gmail.com>
> >> wrote:
> >>
> >>> Sorry, I got most of my answers in http://www.ks.uiuc.edu/Rese
> >>> arch/vmd/current/ug/node235.html
> >>>
> >>> Still, doubt remain regarding applicability of Multiple-walker strategy
>
This archive was generated by hypermail 2.1.6 : Mon Dec 31 2018 - 23:20:19 CST