The ABINIT Users Mailing List ( CLOSED )

[james raynolds <jamesraynolds@yahoo.com>]

Matthieu,

Very nice to hear from you.  In fact I was planning on
writing to you about something else.

I abandoned that compiler and machine.  On an IBM
cluster running Linux and xlf90 I was able to get the
whole thing running.

Now for a physics question (I think):  I am running Si
just for the matrix elements but also for a metallic
silicon wire where the Eliashberg stuff is applicable.


I changed from running the fcc metals Al and Pb and
now I'm using Si but it is telling me that I haven't
computed enough matrix elements.

I'm assuming I haven't done enough values for phonon
polarization vectors.  With Al and Pb, (100) (010) and
(001) were good enough.

Do I now need six?

Thanks again,
Jim

--- Matthieu Verstraete <verstraete@pcpm.ucl.ac.be>
wrote:

> Hello James, welcome to electron-phonon...
> 
> I have also had some segmentation faults of that
> type. Most should be eliminated
> in 4.6.4. To be sure, when you run the debugger,
> could you type "up" several
> times after it crashes? gdb should tell you which
> successive calling routines
> are involved. Send me the whole list, and we should
> be able to tell where in the
> code it crashes (what you gave me is too low level-
> it's in the library routine
> __intel_new_memcpy). It may also be the ifc compiler
> - one of the buggiest on
> the market (being free doesn't give them any
> rights).
> 
> Have fun and please report any successes and
> especially comparisons to
> litterature. Much obliged.
> 
> Matthieu
> 
> Quoting james raynolds <jamesraynolds@yahoo.com>:
> 
> > To all,
> > 
> > I'm running some benchmark calculations on the
> > electron-phonon interaction for Al.  All steps run
> > correctly for small values of the convergence
> > parameters but I get a segmentation fault error
> from
> > anaddb for larger systems.
> > 
> > For example, with ecut=8, I increase the values of
> > ngkpt as follows
> > ngkpt  4 4 4
> > ngkpt  8 8 8
> > ngkpt 12 12 12   <----  This is where I get the
> > segmentation fault
> > 
> > I'm running the pre-compiled binaries for Linux
> with
> > Intel Fortran Compiler (IFC).  I ran anaddb with
> the
> > gdb debugger and the error 
> > message suggests a memory problem
> > 
> > Program received signal SIGSEGV, Segmentation
> fault.
> > 0x08225411 in __intel_new_memcpy ()
> > 
> > Attached to this e-mail you will find a gzipped
> tar
> > archive containing all of the  input and output
> files
> > for abinit, mrgddb, mrggkk and anaddb as well as
> the
> > last page of output from the debugger (gdb.out).
> > 
> > If anyone can help me with this problem I would
> > greatly appreciate it.
> > 
> > Sincerely,
> > Jim Raynolds, Ph.D.
> > College of Nanoscale Science and Engineering
> > University at Albany, State University of New York
> > 
> > __________________________________________________
> > Do You Yahoo!?
> > Tired of spam?  Yahoo! Mail has the best spam
> protection around 
> > http://mail.yahoo.com 
> 
> 
> -------------------------------------------------
> This mail sent through IMP: atlas.pcpm.ucl.ac.be
> 



                
__________________________________ 
Yahoo! Mail - PC Magazine Editors' Choice 2005 
http://mail.yahoo.com