[gmx-users] g_covar 3.2.0

[Berk Hess]

>Dear gromacs users and developers
>
>I tried g_covar of forthcoming version (3.2.0) of gromacs to generate
>a covariance matrix with -xpma option. Resulting matrix was converted
>to EPS but did not look reasonable. I edited the xpm file. Correlations
>were scaled from most negative as "A" - "Z", "a" - "z", "0" - "9" and
>"!" - ";". Zero correlation was corresponding to "r" - "s". There were
>no (!) positive correlations. Most abundant was "r" - lowest negative
>(it is OK), the second was "A" - most negative (it is strange). Absence
>of positive correlations and high abundance of most negative correlations
>indicates that there is some bug. I left "a" - "r" as negative values and 
>tried "A" - "Z" as positive values (by editing of scale of xpm file)
>and resulting diagram looks reasonable. Is it posible that it should be
>like that?
>
>Vojtech Spiwok

The values are incorrect.
Your fix is not correct either, but improves things somewhat.

I fixed similar problems just before the 3.2.0 beta release, but somebody
added several problems afterwards. I fixed things again in the cvs version
and it will also be correct in the final 3.2.0 release.

Thank you for reporting this problem,

Berk.

_________________________________________________________________
MSN Search, for accurate results! http://search.msn.nl