failed in routine check_diag/warning in fm/cp_fm_diag. F:310

[Yu Zhang]

Dear CP2K users,

     I am trying to use CP2K to do a geometry optimization of a 4*2*1 supercell of a Rh(211) cell. I therefore have 4*2*1*24=192 Rh atoms in my system. I keep receiving warnings saying that the eigenvectors returned by SYEVD are not orthonormal and failed in routine check_diag. I saved a screen shoot of the error and added to this conversation for you to check. The inp. out, as well as the coordination files are also attached. 

   Some information: I had done the same calculation for only 1 Rh(211) unit cell and it finished smoothly. I also tried a 3*2*1supercell and it is currently running well. 

    I am looking forward to comments and suggestions from all of you :).

    Thanks a lot for reading.

     Best regards,

     Yu

   

[Krack Matthias (PSI)]

Hi

The —D__CHECK_DIAG compilation option triggers these warnings. It is a debug option which is usually not activated for the compilation of production binaries. The deviations from orthonormality in your screenshot are very small (close to machine precision) and can be neglected.

HTH

Matthias

Am 19.11.2021 um 07:37 schrieb Yu Zhang <yuzha...@gmail.com>:

Dear CP2K users,

<Screen shoot of the error.PNG>

 Yu

   

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/50488c89-b9b4-41fa-9d75-e9e5a1c6703cn%40googlegroups.com.

<Screen shoot of the error.PNG>

<Rh_mp-74_(2_1_1).inp>

<Rh_mp-74_(2_1_1).out>

<211_421.xyz>

[Yu Zhang]

Hi Matthias,

   Thanks a lot for your reply. I will neglect it and let it continue.

   Thanks.

   Best regards,

   Yu