[Octopus-users] EM response error

[Robertson Burgess]

David,

It looks like my problem is similar to yours. I'm also getting that error during some of the test runs, in particular my TD, the em-resp, and Fock-Darwin spectrum. My ID- Neon MPI ran through however. I think my problem must be in some of my compiler flags. I'm using gfortran and gcc. I think your problem can be solved by looking at the Compiler Appendix in the Long Instructions for compiling in the manual. It has identified the problem with the PGI compiler and MPI, and tells you how to fix it.

In the file config.h, replace the line

/* #undef MPI_H */

by

#define MPI_H 1

and remove the line

#define MPI_MOD 1

 I hope that helps.

Robertson Burgess
University of Newcastle


>>> David Strubbe <dstrubbe at berkeley.edu> 10/09/2008 4:34 pm >>>
Robertson,

For an executable compiled with -O4 with the PGI compiler, when I ran
the "make check" earlier today, I saw that very assertion failure for
every MPI run.  There were other problems too and it seems -O2 is the
highest level of optimization that gives correct results on this
computer.

Does this situation bear any resemblance to yours?

David Strubbe
UC Berkeley

On Tue, Sep 9, 2008 at 7:16 PM, Robertson Burgess
<Robertson.Burgess at newcastle.edu.au> wrote:
> Dear Octopus Users,
>
> I am trying to model the EM response of small metallic clusters. The system I am working on at the moment is a cluster of 20 Gold atoms. I ran a ground state calculation successfully, and then tried to perform a em-resp calculation.
> After a short time it crashed, the end of the log file reads:
>
> Input: [AbsorbingBoundaries = no_absorbing]
> Info: Setting up Hamiltonian for linear response
> Assertion "ubound(psi, DIM=1) == gr%m%np_part" failed in line    71 in file "h_inc.F90"
>
> This problem does not occur when I do not input an imaginary component of the frequency.
> The inp file that produces this crash is below:
>
> systemname = 'au20'
> calculationmode = 8
>
> units=2
> %species
> 'Au'|196.9666|spec_ps_psf|79|3|0
> %
>
> extrastates=1
>
> XYZCoordinates='Au.ANI'
>
> ConvAbsDens=2.5e-5
> EigensolverFinalTolerance=1.e-8
> BoxShape=1
>
> Radius=10
>
> SpinComponents=2
> Spacing=0.2
>
> %Boxoffset
>  10|10|-10
> %
>
> FromScratch=yes
>
> periodicdimensions=0
>
> #XCFunctional=lda_x
> #Cfunctional=lda_c_pz
>
> %EMFreqs
> 100 | 1 | 4
> %
> EMEta=0.1
>
> MaximumIter=200
>
> I got the value of 0.1eV for the imaginary frequency at the recommendation of the manual. Is there something wrong with my input, that is causing this error only for EMEta/=0? Any help with this would be greatly appreciated.
>
> Looking forward to your reply,
> Robertson Burgess
> University of Newcastle
>
> _______________________________________________
> Octopus-users mailing list
> Octopus-users at tddft.org 
> http://www.tddft.org/mailman/listinfo/octopus-users 
>