nwchem 6.0: strange behaviour with input file argument
Just Got Here
9:32:47 AM PST - Tue, Feb 7th 2012
I am new in nwchem so it is very likely that something very stupid is going on here.
However I'm experimenting this very strange behaviour of the nwchem executable:

~$ export PATH=$PATH:/local/ARCHIVE/NWCHEM/nwchem-6.0/bin/LINUX64
~$ nwchem h2_bnl.nw

works fine and its resulting output file also passes the comparison with the QA output file. While the following command that should be equivalent to the previous one :

~$ /local/ARCHIVE/NWCHEM/nwchem-6.0/bin/LINUX64/nwchem h2_bnl.nw

fails with following error message:

~$ /local/ARCHIVE/NWCHEM/nwchem-6.0/bin/LINUX64/nwchem h2_bnl.nw 
argument  1 = h2_bnl.nw

 Unable to open h2_bnl.nw --- appending .nw

------------------------------------------------------------------------

nwchem: failed to open the input file        0

------------------------------------------------------------------------

------------------------------------------------------------------------

 current input line : 

    0: 

------------------------------------------------------------------------

------------------------------------------------------------------------

There is an error in the input file

------------------------------------------------------------------------

For more information see the NWChem manual at 

http://nwchemgit.github.io/index.php/NWChem_Documentation

Furthermore, if I give the full path for the input file, the command:
~$ /local/ARCHIVE/NWCHEM/nwchem-6.0/bin/LINUX64/nwchem $(pwd)/h2_bnl.nw
works fine with nwchem deploying the scratch files in the /local/ARCHIVE/NWCHEM/nwchem-6.0/bin/LINUX64/ directory.


The environment variables list I used in the nwchem6.0 build I'm testing follows:


export NWCHEM_MODULES=all
export NWCHEM_TOP=/local/ARCHIVE/NWCHEM/nwchem-6.0
export NWCHEM_TARGET=LINUX64

export USE_MPI=y
export USE_MPIF=y
export USE_MPIF4=y

export MPI_LOC=/opt/pgi/linux86-64/2011/mpi/mpich
export MPI_LIB=$MPI_LOC/lib
export MPI_INCLUDE=$MPI_LOC/include
export LIBMPI="-lfmpich -lmpich -lpmpich"

export LARGE_FILES=TRUE

export FC=pgf90

Clearly, this behaviour of the nwchem executable I have built is not acceptable or desirable .
Any help/hint would be greatly appreciated

Mario

Forum Vet
12:00:56 PM PST - Tue, Feb 7th 2012
mpirun
Mario,
Do you get the same failure if you start the job with mpirun, that is by typing

/opt/pgi/linux86-64/2011/mpi/mpich/bin/mpirun -np 1 /local/ARCHIVE/NWCHEM/nwchem-6.0/bin/LINUX64/nwchem h2_bnl.nw

?

Edo

Forum Vet
12:08:07 PM PST - Tue, Feb 7th 2012
I guess that if you give an absolute path, the starting directory of the binary is it's absolute location, while in the first case it believs it started form your running directory. There is not much that we can do, it's the OS.

What we generally do:

1. Copy the binary to the scratch directory and start the binary from there.

2. Use the PERMANENT_DIR and SCRATCH_DIR directives in the input to tell the binary where to write the files (http://nwchemgit.github.io/index.php/Release61:Top-level#SCRATCH_DIR_.2F_PERMANENT_DIR).

Note, we have released NWChem 6.1

Bert



Quote:Ma t Feb 7th 4:32 pm
I am new in nwchem so it is very likely that something very stupid is going on here.
However I'm experimenting this very strange behaviour of the nwchem executable:

~$ export PATH=$PATH:/local/ARCHIVE/NWCHEM/nwchem-6.0/bin/LINUX64
~$ nwchem h2_bnl.nw

works fine and its resulting output file also passes the comparison with the QA output file. While the following command that should be equivalent to the previous one :

~$ /local/ARCHIVE/NWCHEM/nwchem-6.0/bin/LINUX64/nwchem h2_bnl.nw

fails with following error message:

~$ /local/ARCHIVE/NWCHEM/nwchem-6.0/bin/LINUX64/nwchem h2_bnl.nw 
argument  1 = h2_bnl.nw

 Unable to open h2_bnl.nw --- appending .nw

------------------------------------------------------------------------

nwchem: failed to open the input file        0

------------------------------------------------------------------------

------------------------------------------------------------------------

 current input line : 

    0: 

------------------------------------------------------------------------

------------------------------------------------------------------------

There is an error in the input file

------------------------------------------------------------------------

For more information see the NWChem manual at 

http://nwchemgit.github.io/index.php/NWChem_Documentation

Furthermore, if I give the full path for the input file, the command:
~$ /local/ARCHIVE/NWCHEM/nwchem-6.0/bin/LINUX64/nwchem $(pwd)/h2_bnl.nw
works fine with nwchem deploying the scratch files in the /local/ARCHIVE/NWCHEM/nwchem-6.0/bin/LINUX64/ directory.


The environment variables list I used in the nwchem6.0 build I'm testing follows:


export NWCHEM_MODULES=all
export NWCHEM_TOP=/local/ARCHIVE/NWCHEM/nwchem-6.0
export NWCHEM_TARGET=LINUX64

export USE_MPI=y
export USE_MPIF=y
export USE_MPIF4=y

export MPI_LOC=/opt/pgi/linux86-64/2011/mpi/mpich
export MPI_LIB=$MPI_LOC/lib
export MPI_INCLUDE=$MPI_LOC/include
export LIBMPI="-lfmpich -lmpich -lpmpich"

export LARGE_FILES=TRUE

export FC=pgf90

Clearly, this behaviour of the nwchem executable I have built is not acceptable or desirable .
Any help/hint would be greatly appreciated

Mario

Just Got Here
1:40:54 AM PST - Wed, Feb 8th 2012
First of all, many thanks for the replies.

@Edo: using mpirun, as you have suggested, solves the problem
and, by setting the environment variables:
NWCHEM_TOP, NWCHEM_TARGET, MPIRUN_PATH


Mario


Forum >> NWChem's corner >> Running NWChem