a GEO_OPT calculation of calcite

S Ling lingsa... at gmail.com
Fri May 8 19:31:30 UTC 2015

I have tried to run your calculation with fractional coordinates instead of 
Cartesian coordinates, and it looks like the calculation is running fine. A 
combination of Cartesian coordinates and HEXAGONAL cell could be the reason 
of your job failure, and your initial error message "ERRORL2 in 
cp_fm_cholesky:cp_fm_cholesky_decompose processor" might indicate that you 
have very bad geometry at that GEO_OPT step.

In addition, can you explain why you set the mass of carbon atoms to 2.016 
in your input?


On Friday, May 8, 2015 at 2:36:09 PM UTC+1, Mengjia He wrote:
> Thank you Matt.
> YES, the input started running OK, but stopped at 17 steps with the error 
> mentioned above.
> I have carried out the modified inp file removing the STRESS_TENSOR 
> NUMERICAL and increasing the cutoff to 500 Ry. The result seems to be more 
> far away from convergence, though it isn't ending now. 
> ( I've add the xyz to attanchment.)
> Actually I wanna a CELL_OPT for calcite calculation at first. But the 
> result turned out badly. The Convergence check always showed NO of all and 
> the structures of xyz file became stranger , likely the whole crystal 
> structure dispersing ( almost like the attanchment). So I turned  GEO_OPT 
> to check first. 
> Could you please suggest me what else can be done for the GEO_OPT?
> PS: As a new user of CP2K, would you like to tell me how to set the 
> cutoff correctly and fast? (I learned the tutorial 
> http://www.cp2k.org/howto:converging_cutoff. And I wonder if there's any 
> file to summarize the cutoff set?)
> Cheers.
> Mengjia
> 在 2015年5月7日星期四 UTC+8下午9:16:43,Matt W写道:
>> Hi,
>> your input started running OK for me. Couple of things that might help:
>> Remove the STRESS_TENSOR NUMERICAL - you never want that, maybe 
>> STRESS_TENSOR ANALYTICAL if doing a cell optimization.
>> Your cutoff is much too low for the BLYP functional and the elements in 
>> the system - this might be the cause of the problem later on. Increase the 
>> cutoff to at least 500 Ry, probably more. Check that the "Electronic 
>> density on regular grids:       -767.9999808982        0.0000191018" is 
>> correct to at least 6 or 7 figures as a rough idea.
>> Matt 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20150508/56d6ae8c/attachment.htm>

More information about the CP2K-user mailing list