SEGSIGV using LDA+U and lowdin population with the regtest example

Matthias Krack matthia... at psi.ch
Wed Feb 17 13:29:08 CET 2010


Hi Valerio

there are specific regtests for Mulliken and Lowdin and the current
regtest is clean w.r.t. DFT+U. Please, try to reproduce the error with
the current CP2K version and report it as suggested by the rules for
help requests (http://groups.google.com/group/cp2k/web/general-rules-
for-posting-help-requests)

Matthias

On Feb 17, 12:09 pm, Valerio <valerio... at unimore.it> wrote:
> Hello everybody,
> I tried to make a calculation with LDA+U, and first checked
> one regtest input, namely H2O-uks-otcg-mulliken.inp in  the /tests/QS/
> regtest-plus_u directory.
> With mulliken pupulation I have no problem, If I use instead Lowdin
> I got a  SIGSEGV - Segmentation violation.
> Which can be the problem? could be due to optimization flags in the
> compilation?
> the output stops after
>
>   ----------------------------------- OT
> ---------------------------------------
>
>   Step     Update method      Time    Convergence         Total
> energy    Change
>
> ------------------------------------------------------------------------------
>
>   Total electronic density (r-space):          -7.9999969202
> 0.0000030798
>   Total core charge density (r-space):          7.9999999539
> -0.0000000461
>   Total charge density (r-space):
> 0.0000030337
>   Total charge density (g-space):
> 0.0000030337
>
> did somebody tried LDA+U+lowdin in the last version of the code?
> I am using the version dowloaded on 28 October 2009.
> Regards,
> Valerio



More information about the CP2K-user mailing list