AMBER Archive (2005)

Subject: Re: AMBER: AMBER 8.0 problems with impose (tleap) on linux

From: David A. Case (case_at_scripps.edu)
Date: Tue Apr 05 2005 - 23:48:14 CDT

On Tue, Apr 05, 2005, Guillermo Mulliert Carl?n wrote:

> I compiled amber8 which Intel Fortran 8.0 and Intel(R) Math Kernel Library
> v7.0 on linux RedHat 9.0. I made some time ago a small tetrapeptide to
> teach computer students how to measure torsion angles in proteins with the
> following input file.
>
> ...............................................................................
> moltor = sequence { NALA ALA ALA CALA }
>
> impose moltor { 1 2 } { { N CA C N 0.0 } { CA C N CA 60.0 } { C N CA C 120.0 } }

....

OK, I can definitely confirm that there is a problem somewhere. With gcc
3.2.2 on Linux (Fedora Core) I get wrong results, just as you report. But with
icc 8.0.55, the correct results are obtained. I also get correct results with
gcc 3.3.3 on cygwin (?!?).

The problem appears to be in the "impose" command, not with "measuregeom".
Using the Linux gcc with -O0 (rather than the default -O2) has no effect.

So, we need to look hard at what is going on. Given that I can create both
"good" and "bad" results, I'm hoping it won't take too long to figure this
out. As always, anyone who wants to help out is welcome to contribute!

Thanks for the report...dac

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu