[gmx-users] R.M.S.I.P for identical eigenvectors not equal to 1 ?

[raja]

Hi Tsjerk,
Oops, I am sorry and thanks for your help and providing me the program. 
With thanks!
B.Nataraj


On Fri, 28 Jul 2006 13:35:48 +0200, "Tsjerk Wassenaar"
<tsjerkw at gmail.com> said:
> Hi Raja,
> 
> Okay, you're not using the eigenvectors, you're using the
> eigenvalues... The eigenvectors are written to a .trr or .xpm file, as
> NxN matrices.
> 
> You can try the attached script (which I got from Isabella Daidone) on
> the .xpm file.
> 
> Good luck,
> 
> Tsjerk
> 
> On 7/28/06, raja <raja_28 at fastmail.us> wrote:
> > Hi Tsjerk,
> > Here is the script I am using. I already checked the value computed by
> > this script with manually computed value for small set of 3 eigen
> > vectors in each file. As you already know the formula, so please check
> > it up by yourself. More I am not a regular programmer so this script is
> > written for some quick solution, so please do not expect perfection in
> > usage of commands.
> > #---------------------------Perl script for computing R.M.S.I.P.
> > value----------------------------------------------------------
> > # each file "eig1.xvg" and "eig2.xvg" contain 10 eigenvectors
> > $F1 = "eig1.xvg";
> > $F2 = "eig2.xvg";
> > open(F1) or die("Could not open file.");
> > $N1 = 0;
> > foreach $line(<F1>)
> > {
> > $A = $line;
> > open(F2) or die("Could not open file");
> > foreach $line(<F2>)
> > {
> > $B = $line;
> > $N = ($A*$B)*($A*$B);
> > print " $N \n";
> > $N1 = $N1 + $N ;
> > #print " $N1 \n ";
> > }
> > close(F2)
> > }
> > print "The Value of N1 is $N1 \n";
> > $N1 = $N1/10;
> > $N1 = sqrt ($N1);
> > print "$N1";
> > close(F1) ;
> > #-------End of
> > code-------------------------------------------------------------------------------------------
> >
> > With thanks !
> > B.Nataraj
> >
> >
> >
> > On Fri, 28 Jul 2006 13:05:23 +0200, "Tsjerk Wassenaar"
> > <tsjerkw at gmail.com> said:
> > > Hi Raja,
> > >
> > > I really don't get it. From the definition it follows directly that if
> > > you have the same sets of eigenvectors, the RMSIP will be 1. Each
> > > eigenvector from a set gives an inproduct of 1 with itself and of 0
> > > with any of the others, yielding a total of 10 for the double sum...
> > > etc. If you use the same set twice, you probably have a bug in your
> > > script.
> > >
> > > Besides, from the definition it also follows that the RMSIP can not
> > > exceed 1. I think 3.5% error is quite a bit.
> > >
> > > Tsjerk
> > >
> > > On 7/28/06, raja <raja_28 at fastmail.us> wrote:
> > > > Hi Tsjerk,
> > > > Thanks for your reply. I used identical values of 10 eignevectors in two
> > > > different file to compute r.m.s.i.p. I have pasted the link of the
> > > > article from where I used the formula for r.m.s.i.p. More for some other
> > > > cases of non-identical values, it slightly exceeds the value of 1 (e.g.,
> > > > 1.0352), Which I am thinking due to some numerical problem. But I am
> > > > confident that my script exactly following what the formula indicated in
> > > > the article as pasted hereunder.
> > > > http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1304087
> > > >
> > > > With thanks !
> > > > B.Nataraj
> > > >
> > > > On Fri, 28 Jul 2006 11:32:04 +0200, "Tsjerk Wassenaar"
> > > > <tsjerkw at gmail.com> said:
> > > > > Hi Raja,
> > > > >
> > > > > Did you use the same eigenvectors from the same simulation over the
> > > > > same stretch of time? For two (parts of a) simulation(s) of the same
> > > > > system, the RMSIP value will only converge to one in the limit of
> > > > > infinite time, or on long time scales at least. The eigenvectors
> > > > > obtained from two different (parts of) simulations can vary if
> > > > > different parts of the conformational space are sampled or simply if
> > > > > the sampling is different, which will be almost always for a limited
> > > > > stretch of time.
> > > > >
> > > > > Best regards,
> > > > >
> > > > > Tsjerk
> > > > >
> > > > > On 7/28/06, raja <raja_28 at fastmail.us> wrote:
> > > > > > Dear all,
> > > > > >
> > > > > > I have written small perl script to compute R.M.S.I.P.(Root mean square
> > > > > > internal product) for comparing two eigen vectors set from two different
> > > > > > simulation. I compared first 10 eigen vectors of each set. The strange
> > > > > > thing is that while I compare eigen vectors of same simulation, it is
> > > > > > showing a value of 0.54, but literature says, r.m.s.i.p of 1 indicates
> > > > > > that the sets are identical, while a value of 0 indicates that the
> > > > > > eigenvectors are orthogonal. I also confirmed my script validity by
> > > > > > manually computing the R.M.I.P for small set of values.
> > > > > >
> > > > > > Could any body explain me a bit regarding this issue?
> > > > > >
> > > > > > With thanks!
> > > > > > B.Nataraj
> > > > > > --
> > > > > >   raja
> > > > > >   raja_28 at fastmail.us
> > > > > >
> > > > > > --
> > > > > > http://www.fastmail.fm - A fast, anti-spam email service.
> > > > > >
> > > > > > _______________________________________________
> > > > > > gmx-users mailing list    gmx-users at gromacs.org
> > > > > > http://www.gromacs.org/mailman/listinfo/gmx-users
> > > > > > Please don't post (un)subscribe requests to the list. Use the
> > > > > > www interface or send it to gmx-users-request at gromacs.org.
> > > > > > Can't post? Read http://www.gromacs.org/mailing_lists/users.php
> > > > > >
> > > > >
> > > > >
> > > > > --
> > > > >
> > > > > Tsjerk A. Wassenaar, M.Sc.
> > > > > Groningen Biomolecular Sciences and Biotechnology Institute (GBB)
> > > > > Dept. of Biophysical Chemistry
> > > > > University of Groningen
> > > > > Nijenborgh 4
> > > > > 9747AG Groningen, The Netherlands
> > > > > +31 50 363 4336
> > > > > _______________________________________________
> > > > > gmx-users mailing list    gmx-users at gromacs.org
> > > > > http://www.gromacs.org/mailman/listinfo/gmx-users
> > > > > Please don't post (un)subscribe requests to the list. Use the
> > > > > www interface or send it to gmx-users-request at gromacs.org.
> > > > > Can't post? Read http://www.gromacs.org/mailing_lists/users.php
> > > > --
> > > >   raja
> > > >   raja_28 at fastmail.us
> > > >
> > > > --
> > > > http://www.fastmail.fm - Does exactly what it says on the tin
> > > >
> > > > _______________________________________________
> > > > gmx-users mailing list    gmx-users at gromacs.org
> > > > http://www.gromacs.org/mailman/listinfo/gmx-users
> > > > Please don't post (un)subscribe requests to the list. Use the
> > > > www interface or send it to gmx-users-request at gromacs.org.
> > > > Can't post? Read http://www.gromacs.org/mailing_lists/users.php
> > > >
> > >
> > >
> > > --
> > >
> > > Tsjerk A. Wassenaar, M.Sc.
> > > Groningen Biomolecular Sciences and Biotechnology Institute (GBB)
> > > Dept. of Biophysical Chemistry
> > > University of Groningen
> > > Nijenborgh 4
> > > 9747AG Groningen, The Netherlands
> > > +31 50 363 4336
> > > _______________________________________________
> > > gmx-users mailing list    gmx-users at gromacs.org
> > > http://www.gromacs.org/mailman/listinfo/gmx-users
> > > Please don't post (un)subscribe requests to the list. Use the
> > > www interface or send it to gmx-users-request at gromacs.org.
> > > Can't post? Read http://www.gromacs.org/mailing_lists/users.php
> > --
> >   raja
> >   raja_28 at fastmail.us
> >
> > --
> > http://www.fastmail.fm - Email service worth paying for. Try it for free
> >
> > _______________________________________________
> > gmx-users mailing list    gmx-users at gromacs.org
> > http://www.gromacs.org/mailman/listinfo/gmx-users
> > Please don't post (un)subscribe requests to the list. Use the
> > www interface or send it to gmx-users-request at gromacs.org.
> > Can't post? Read http://www.gromacs.org/mailing_lists/users.php
> >
> 
> 
> -- 
> 
> Tsjerk A. Wassenaar, M.Sc.
> Groningen Biomolecular Sciences and Biotechnology Institute (GBB)
> Dept. of Biophysical Chemistry
> University of Groningen
> Nijenborgh 4
> 9747AG Groningen, The Netherlands
> +31 50 363 4336
-- 
  raja
  raja_28 at fastmail.us

-- 
http://www.fastmail.fm - Send your email first class