[ase-users] Jacapo/Dacapo not working on Ubuntu 11.04

[Marcin Dulak]

Hi,

this looks like:
http://forum.wrfforum.com/viewtopic.php?f=5&t=1501
Can you post:
gfortran -v
and
ompi_info | head -50

Marcin

Alex Eftimiades wrote:
> I do not think the problem is not underscoring. I tried with 
> fsecond_underscore, and it would not compile.
>
> The serial version does compile without the patch. The mpi version 
> does not. It complains about -f90 being an "unrecognized option" an 
> promptly halts. With that option removed, in just complained noisily 
> about some unexpected tokens (I am going on memory--it has been a few 
> weeks since I compiled the code.)
>
> I have both working on my computer now. Here is the relevant section 
> of the installer script I use:
>
> cd /opt
> svn checkout https://svn.fysik.dtu.dk/projects/dacapo/trunk dacapo
> cd dacapo
> mv psp psp2
> mkdir psp
> for m in `find ./psp2 -name '*.pseudo'`;do mv $m ./psp;done
> rm -rf psp2
> cd src
> export NETCDF="/usr/lib"
> export DACAPOPATH="/opt/dacapo/psp"
> export BLASLAPACK="-L/usr/lib -lnetcdff -lnetcdf -L/usr/lib -lblas 
> -llapack -L/usr/lib -lblas -llapack -lgfortran"
> export FFTW="/usr/lib"
> export MPIDIR="/usr/lib/openmpi"
> export MPI_LIBDIR="$MPIDIR/lib"
> export MPI_INCLUDEDIR="$MPIDIR/include"
> export MPI_BINDIR="/usr/bin"
> make gfortran_fnosecond_underscore
> patch Makefile < ~/Downloads/dacapo_patch_for_ubuntu.patch #This is a 
> simple patch I made that seems to allow the program to compile the mpi 
> version with gfortran
> make gfortran_fnosecond_underscore MP=mpi
> echo "
> export DACAPOPATH=\"/opt/dacapo/psp\"
> export 
> PATH=\$PATH:/opt/dacapo/src/gfortran_fnosecond_underscore_serial #For 
> the mpi version, change serial to to mpi
> " >> ~/.bashrc
> . ~/.bashrc
>
>
> This and the patch are posted 
> here: http://sourceforge.net/projects/aeftimiamisc/files/installstuff/
>
> Dacapo works just fine in serial
>
> On Aug 24, 2011, at 12:18 PM, Marcin Dulak wrote:
>
>> Hi,
>>
>> i think we need to split the problem into smaller ones:
>> 1. please try to install the serial version of dacapo first.
>> You should not need any patches - maybe the right underscoring 
>> conventions for you system is gfortran_fsecond_underscore
>> and not gfortran_fnosecond_underscore. Please check you underscoring 
>> convention as described here 
>> https://listserv.fysik.dtu.dk/pipermail/ase-users/2009-September/000362.html
>> 2. are you able to run the simple example CO from 
>> https://wiki.fysik.dtu.dk/ase/ase/calculators/jacapo.html?
>> To the problem of the optimizer I get in a separate post.
>>
>> Best regards,
>>
>> Marcin
>>
>> Alex Eftimiades wrote:
>>> I am trying to run the unit cell optimization provided here: 
>>> http://gilgamesh.cheme.cmu.edu/jacapo/jacapo/2-bulk/2.2-unitcell-optimization/2.2.0-unit-cell-optimization.html 
>>>
>>>
>>> I get:
>>> >>> print bulk.get_stresses()
>>> None
>>>
>>> That happens for all the examples on the page. Much worse errors 
>>> occure if I try bulk.get_forces()
>>> File "<stdin>", line 1, in <module>
>>> File "/usr/lib/python2.7/dist-packages/ase/atoms.py", line 571, in 
>>> get_forces
>>> forces = self._calc.get_forces(self)
>>> File 
>>>  "/usr/lib/python2.7/dist-packages/ase/calculators/jacapo/jacapo.py", 
>>> line 2309, in get_forces
>>> forces = nc.variables['DynamicAtomForces'][-1]
>>> KeyError: 'DynamicAtomForces'
>>>
>>> I should note I manually transcribed that error output (as well as 
>>> the ones described later in this message.) from another computer 
>>> screen, so please try to look over any typos.
>>>
>>> Now, I installed everything except decapo from apt-get repositories 
>>> (including ase and gpaw.) Decapo did not install willingly. Here is 
>>> the sequence of commands I used:
>>> cd /opt
>>> DIR=~/Downloads
>>> svn checkout https://svn.fysik.dtu.dk/projects/dacapo/trunk dacapo
>>> cd dacapo
>>> mv psp psp2
>>> mkdir psp
>>> for m in `find ./psp2 -name '*.pseudo'`;do mv $m ./psp;done
>>> rm -rf psp2
>>> cd src
>>> export NETCDF="/usr/lib"
>>> export DACAPOPATH="/opt/dacapo/psp"
>>> export BLASLAPACK="-L/usr/lib -lnetcdff -lnetcdf -L/usr/lib -lblas 
>>> -llapack -L/usr/lib -lblas -llapack -lgfortran"
>>> export FFTW="/usr/lib"
>>> export MPIDIR="/usr/lib/openmpi"
>>> export MPI_LIBDIR="$MPIDIR/lib"
>>> export MPI_INCLUDEDIR="$MPIDIR/include"
>>> export MPI_BINDIR="/usr/bin"
>>> patch Makefile < $DIR/dacapo_patch_for_ubuntu.patch.patch #This is a 
>>> simple patch I made that seems to allow the program to compile with 
>>> gfortran
>>> make gfortran_fnosecond_underscore MP=mpi
>>> echo "
>>> export PATH=\$PATH:/opt/dacapo/src/gfortran_fnosecond_underscore_mpi
>>> " >> ~/.bashrc
>>> . ~/.bashrc
>>> You can download the patch, dacapo_patch_for_ubuntu.patch.patch, 
>>> from: 
>>> http://sourceforge.net/projects/aeftimiamisc/files/installstuff/dacapo_patch_for_ubuntu.patch.patch/download
>>> It deletes the compiler option 
>>> -f90=$(GFORTRAN_FNOSECOND_UNDERSCORE_FC90)' 
>>> MPI_LIBDIR=${GFORTRAN_FNOSECOND_UNDERSCORE_MPIDIR}
>>> which the mpi compiler does not recognize. I will try again with the 
>>> serial version tomorrow. In the mean time has anyone gotten this 
>>> successfully working on Ubuntu (recently.) I have looked through the 
>>> listserves and followed everything they suggested.
>>>
>>> *I really just need /something/ that can optimize the geometry and 
>>> the unit cell.* If anyone knows of any other (free) software that 
>>> can do that, please let me know!
>>>
>>>
>>> Thanks,
>>> Alex
>>>
>>>
>>> warning: extra tokens at end of #endif directive
>>> and Warning: Deleted feaure: End of expression in DO loop at (1) 
>>> must be integer
>>
>>
>> -- 
>> ***********************************
>> Marcin Dulak
>> Technical University of Denmark
>> Department of Physics
>> Building 307, Room 229
>> DK-2800 Kongens Lyngby
>> Denmark
>> Tel.: (+45) 4525 3157
>> Fax.: (+45) 4593 2399
>> email: Marcin.Dulak at fysik.dtu.dk <mailto:Marcin.Dulak at fysik.dtu.dk>
>>
>> ***********************************
>>
>


-- 
***********************************
 
Marcin Dulak
Technical University of Denmark
Department of Physics
Building 307, Room 229
DK-2800 Kongens Lyngby
Denmark
Tel.: (+45) 4525 3157
Fax.: (+45) 4593 2399
email: Marcin.Dulak at fysik.dtu.dk

***********************************