[gmx-users] Bug (?) in dgdl.xvg

[Berk Hess]

>From: Maik Goette <mgoette at mpi-bpc.mpg.de>
>Reply-To: Discussion list for GROMACS users <gmx-users at gromacs.org>
>To: GMX-mailinglist <gmx-users at gromacs.org>
>Subject: [gmx-users] Bug (?) in dgdl.xvg
>Date: Thu, 10 Mar 2005 16:00:39 +0100
>
>Dear Gromacs developer,
>
>I think, I found a bug-like problem in the FEP-Output-file dgdl.xvg.
>Have a look at this:
>
>---cut
>999.99 513.523
>999.992 530.992
>999.994 511.562
>999.996 453.502
>999.998 369.732
>1000 561.751
>1000 569.696
>1000 615.34
>1000.01 679.243
>1000.01 755.308
>1000.01 809.608
>1000.01 865.887
>1000.01 928.085
>---cut
>
>This is a part out of the file at the edge of 1ns.
>As you can see, the output format is to small (7 characters) to print the 
>last number (after 1ns) and therefore rounds the value. This is not the bag 
>of dwarves, as I like to say. I am not able to process the files any 
>further, without writing a tool, which fixes this. So are you able to 
>change this in the next version?
>I don't know, if no one ever did a FEP over 2ns before, but in my opinion 
>this is a major problem.

The format is %g, which apparently uses only 7 places.
I'll fix it.
Many people do FE over more than 2ns, as you always need a
lot of sampling. But normally you only need the average and
fluctuations and do not care about the exact times.
You can easily patch the dgdl file by setting time=step*0.002

Berk.

_________________________________________________________________
Play online games with your friends with MSN Messenger 
http://messenger.msn.nl/