a2y conversion with "large" KSS file

Forum for discussing issues arising with interfacing the Abinit code (KSS) file with Yambo.

Moderator: Conor Hogan

Post Reply
elena.mol
Posts: 17
Joined: Mon Oct 04, 2010 10:04 am

a2y conversion with "large" KSS file

Post by elena.mol » Tue Oct 05, 2010 10:44 am

Dear all,
I am using abinit + yambo to calculate optical spectra of a large molecule.
I have a problem with a2y converting the KSS file (generated with abinit6.2.2). For "small" KSS files ("small" bands, cutoff, convergence treshold, ...) a2y works well, while with a KSS file generated with converged parameters it stops with something like:

DB1
==WF splitter Blocks/bands/block size (Mb)


Can anyone help me with this problem?


thanks in advance
Elena Molteni
Department of Physics
University of Milan
via Celoria, 16
20133, Milan, Italy

User avatar
andrea marini
Posts: 325
Joined: Mon Mar 16, 2009 4:27 pm
Contact:

Re: a2y conversion with "large" KSS file

Post by andrea marini » Tue Oct 05, 2010 10:57 am

Dear Elena, we really need more information to try to trace back the problem :( Can you follow the instructions given on this link providing as much info as possible ? In the worst case we will need to reproduce the problem ourselves.

Andrea
Andrea MARINI
Istituto di Struttura della Materia, CNR, (Italy)

elena.mol
Posts: 17
Joined: Mon Oct 04, 2010 10:04 am

Re: a2y conversion with "large" KSS file

Post by elena.mol » Tue Oct 05, 2010 1:24 pm

Dear all,
i'm attaching a .tar (FOR_YAMBOFORUM_2.tar) containing:

* writeKSS.in, writeKSS.files: the input files used to generate the KSS file in abinit
* abinit_config.log: compilation details of the abinit version (6.2.2) i compiled and use on cineca sp6 (sp.sp6.cineca.it, the machine where i'm finding this problem with a2y); i miss the corresponding detailed information for yambo, since for this latter i'm using the standard cineca version, but for abinit i followed compilation hints given by the cineca helpdesk, so i guess the two installations should be "coherent"
* l_stderr: standard log of the a2y run which produces the error

In /gpfs/scratch/userinfm/cne0fm2s/elena, i have put an "enlarged" version of this directory (FOR_YAMBOFORUM_1), containing also some bigger files, which might be useful, but are too large to be attached (i guess), i.e.:
* the /SAVE dir produced by the a2y run
* writeKSSo_KSS: the kss file i'm using as input for a2y

let me know if you have problems in accessing them, or if there is some missing info

thanks in advance
Elena Molteni
Department of Physics
University of Milan
via Celoria, 16
20133, Milan, Italy
You do not have the required permissions to view the files attached to this post.

User avatar
Conor Hogan
Posts: 111
Joined: Tue Mar 17, 2009 12:17 pm
Contact:

Re: a2y conversion with "large" KSS file

Post by Conor Hogan » Tue Oct 05, 2010 2:00 pm

Hi Elena, thanks for that.
I copied the KSS and managed to run a2y myself without problems through the batch,
and adding -S to split the wavefunctions. Try that.

#!/bin/bash
#
# @ job_name = molt
# @ output = $(jobid).out
# @ error = $(jobid).err
# @ wall_clock_limit = 0:10:00
# @ total_tasks = 1
# @ task_affinity=cpu(1)
# @ job_type = parallel
# @ resources = ConsumableMemory(1700Mb)
#
# @ queue

module load yambo
a2y -S -F test.KSS

Gives:
<---> [01] A(binit) 2 Y(ambo)
<---> Checking input file ...
<---> DBs path set to :.
<---> KSS Header...
<---> :: Atom's informations...
<---> Symmetries...
<---> RL vectors...
<---> Energies...
<---> Report:
<---> :: Electrons : 471.0000
<---> :: Temperature [ev]: 0.000000
<---> :: Lattice factors [a.u.]: 79.36850 92.59658 62.73891
<---> :: K-points : 1
<---> :: Bands : 540
<---> :: Spinor components : 1
<---> :: Spin polarizations : 1
<---> :: Spin orbit coupling : no
<---> :: Symmetries [spatial]: 1
<---> :: [T-rev]: yes
<---> :: Max WF components : 176165
<---> :: RL vectors (WF): 176165
<---> :: RL vectors (CHARGE): 176165
<---> :: XC potential : Perdew & Wang (xc)
<---> :: Atomic species : 4
<---> :: Max atoms/species : 92
<---> == DB1 ...
<---> :: WF splitter Blocks/bands/block size(Mb): 2 270 362

and creates

sp0201 62% ls SAVE
ns.db1 ns.wf ns.wf_fragments_1_2
ns.kb_pp ns.wf_fragments_1_1

Let me know if you still have problems.
Conor.
Dr. Conor Hogan
CNR-ISM, via Fosso del Cavaliere, 00133 Roma, Italy;
Department of Physics and European Theoretical Spectroscopy Facility (ETSF),
University of Rome "Tor Vergata".

Post Reply