AMBER Archive (2009)
Subject: Re: [AMBER] Solved-Amber 1.2 tools test suite failure
From: Jim Parker (jim.parker_at_prismsciences.com)
Date: Sat Apr 11 2009 - 14:29:55 CDT
Ok, no csh = thick-headed, I guess. At first it was a challenge, now
very difficult. BTW, I only have Linux systems so bash is comfortable
so I installed csh, and g77-3.4.6 and the install went fine. But again
I crashed at
Original error was
>make -f Makefile_at test
--lots of PASSED tests :-)
cd antechamber/tp && ./Run.tp
./Run.tp: Program error
make: *** [test.antechamber.hasG77] Error 1
I traced the failure back to the mopac.sh script
# Edit the following to match your local environment; the syntax
# for calling this shell script should be: mopac.sh input output.
# Depending on which version of mopac you have, you may need to remove
# the "<" and ">" symbols below, or make other changes. For example,
# some versions of mopac expect the input file to be "FOR005" and the
# output to be "FOR006"; some versions require the input file to end in
# ".DAT", etc., etc.
ln mopac.in FOR005
/home/surfer/tmp/amber10/bin/mopac >& FOR006
mv FOR006 mopac.out
rm -f FOR0??
I needed to change the #!/bin/sh -> #!/bin/csh, and it worked, and was
good for the other antechamber tests as well.
I don't know if it will break other systems to explicitly state which
bin/shell you are using, but perhaps making the change will help others
whose default shell is not csh.
The remainder of the test suite works fine! I'm now off to look at
Ross Walker wrote:
> Hi Jim,
>> the test ptraj_rmsa/Run.rmsa script is written for 'csh'. In fact all
>> the scripts in test.ptraj are in csh format. (I was surprised there
>> were not more questions on the archives related to this)
>> I was able to convert them to bash style, to run on Ubuntu, which gets
>> me to the next series of tests.
> This just makes debugging any problems you have really really hard. Why not
> just do the simple solution and install csh?
>> Ubuntu does not install with g77, it uses gfortran-4.3.2, and so this
>> section would be skipped
>> as these depend upon mopac. But I was curious about them, and attempted
>> to figure out how to make them run.
> Mopac should compile out of the box in Ambertools 1.2 using gfortran.
>> mopac (the version with AmberTools) requires the input file to be named
>> FOR005 and generates 3 output files FOR011, FOR012, and FOR016.
>> There is a script mopac.sh provided by AmberTools as a wrapper for calls
>> to mopac so that the user can pass arguments on the command line,
>> something that the native Fortran program it was written in did not
>> support. And again it was written in csh, so I modified for use with bash.
> Again why not just install csh so that it would be easy for us to help.
>> The question I have is that although the test runs to completion, the
>> result is "possible FAILURE: check tp.mol2.diff"
>> Can anybody tell me if the these results are OK?
>> cat tp.mol2.diff
>> < 1 CG -1.9590 0.1020 0.7950 ca 1 TP
>>> 1 CG -1.9590 0.1020 0.7950 ca 1 TP
>> < 2 CD1 -1.2490 0.6020 -0.3030 ca 1 TP
>>> 2 CD1 -1.2490 0.6020 -0.3030 ca 1 TP
>> < 3 CD2 -2.0710 0.8650 1.9630 ca 1 TP
>>> 3 CD2 -2.0710 0.8650 1.9630 ca 1 TP
>> < 5 C6 -1.4720 2.1290 2.0310 ca 1 TP
>>> 5 C6 -1.4720 2.1290 2.0310 ca 1 TP
>> < 7 HE2 -1.5580 2.7190 2.9310 ha 1 TP
>>> 7 HE2 -1.5580 2.7190 2.9310 ha 1 TP
>> < 8 S15 -2.7820 0.3650 3.0600 sh 1 TP
>>> 8 S15 -2.7820 0.3650 3.0600 sh 1 TP
>> < 11 H30 0.3730 2.0450 -0.7840 ha 1 TP
>>> 11 H30 0.3730 2.0450 -0.7840 ha 1 TP
>> < 12 H31 -0.0920 3.5780 0.7810 ha 1 TP
>>> 12 H31 -0.0920 3.5780 0.7810 ha 1 TP
>> < 13 H32 -2.3790 -0.9160 0.9010 ha 1 TP
>>> 13 H32 -2.3790 -0.9160 0.9010 ha 1 TP
>> ### Maximum absolute error in matching lines = 6.00e-04 at line 13 field 9
>> ### Maximum relative error in matching lines = 2.45e-02 at line 9 field 9
>> I think that it probably is, but I'd like someone to check before I
>> continue modifying the other tests.
> Your charges are all off around 0.0002 to 0.0004. This is probably innocuous
> and likely comes from differences between the g77 compilation of MOPAC and
> gfortran. This version of mopac is pretty old and doesn't play nice with new
> compilers. It will be replaced soon.
>> Also, would there be any interest in posting my edits to the testsuite?
> I guess the issue is whether it makes sense to update everything to bash
> shell. Probably not since csh is generally more ubiquitous than bash which
> is typically only found on Linux systems. I would be interested to know why
> having access to csh on your machine is so hard as to justify taking the
> time to modify all the scripts.
> All the best
> |\oss Walker
> | Assistant Research Professor |
> | San Diego Supercomputer Center |
> | Tel: +1 858 822 0854 | EMail:- ross_at_rosswalker.co.uk |
> | http://www.rosswalker.co.uk | PGP Key available on request |
> Note: Electronic Mail is not secure, has no guarantee of delivery, may not
> be read every day, and should not be used for urgent or sensitive issues.
> AMBER mailing list
AMBER mailing list