[CP2K:7866] Re: "Cholesky failed" problems new in cp2k 3.0 and trunk versions

[Simiam Ghan]

Hello, 
By removing the redundant basis set declaration I also was able to fix the 
Cholesky errors.   Just as a comment here - I observed that the calculation 
(MD of a box of 64 waters with K+ and Cl- ions) sped up by roughly a factor 
of 2 (!) due to this change and so would have saved time had version 3.0 
ignored the redundancy by default.

Greetings,
Simiam





On Tuesday, June 28, 2016 at 6:21:11 PM UTC+3, Christian Tuma wrote:
>
> Dear Marcella, dear Jürg,
>
> thank you very much for your swift replies. Indeed, removal of the 
> redundant basis set declarations in the input file solved the problem:
>
> ...
>      &KIND Au
>        BASIS_SET DZVP-MOLOPT-SR-GTH
>        POTENTIAL GTH-PBE-q11
>      &END KIND
> ...
>
> From this, MD restart files created with version 2.6.2 introduce 
> redundancy again:
>
>      &KIND Au
>        BASIS_SET DZVP-MOLOPT-SR-GTH
>        POTENTIAL GTH-PBE-q11
>        &BASIS
>  ...
>        &END BASIS
>        &POTENTIAL
> ...
>        &END POTENTIAL
>      &END KIND
>
> whereas with the trunk version only the potential is repeated in the MD 
> restart files:
>
>      &KIND Au
>        BASIS_SET DZVP-MOLOPT-SR-GTH
>        POTENTIAL GTH-PBE-q11
>        &POTENTIAL
> ...
>        &END POTENTIAL
>      &END KIND
>
>
> From this I conclude that one has to be careful when using cp2k 3.0 (or 
> trunk) with MD restart files created by older versions of cp2k.
>
>
> Kind regards,
> Christian
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20160817/21f5f168/attachment.htm>