[CP2K-user] [CP2K:21984] Re: Issues with MO warnings and Cholesky decomposition failure in CP2K with def2-TZVP

[Richard李]

Dear Marcella (and CP2K developers),

Thank you very much for your detailed and helpful reply.

I followed your suggestion and tested the MOLOPT series at the DZVP level — 
this solved the ill-conditioned S-matrix problem for the Cu slab (the 
calculation now runs stably). For the CO molecule, your explanation about 
the limited number of basis functions (and thus the maximum number of MOs) 
clarified my confusion — I had overlooked that detail before.

Regarding the MO warning, I will set ADDED_MOS -1 when I want to explicitly 
request all available MOs to suppress the warning (or otherwise request a 
number within the available range).

I really appreciate your time and warming guidance. Thanks again for the 
advice — it was kind of you helping a lot.

Best regards,
Sun

在2025年11月19日星期三 UTC+8 16:55:59<Marcella Iannuzzi> 写道：

Hi ..

The warning for the CO calculation is not a real problem. Since the number 
of orbital functions, and consequently the size of  H, is 62, it is not 
possible to generate more than 62 MOs, including occupied and virtual.  The 
advise provided by the warning message is to set ADDED_MOS -1 to request 
the maximum number of available MOs. On the other hand, the code already 
makes this choice and the calculation runs to the end.

The ill conditioned S matrix in the calculation of  the slab is due to the 
choice of the basis set, which is not really suitable for condensed matter 
calculations. To successfully run such systems,  basis sets optimised to 
reduce linear dependencies, as the MOLOPT series (but you can find others 
in the literatures), need to be adopted. 

Regards
Marcella

On Wednesday, November 19, 2025 at 9:38:22 AM UTC+1 scarlet... at gmail.com 
wrote:

Dear CP2K developers,

I am using CP2K with PBE-D3(BJ) to calculate single-point energies and try 
to make an output of AO overlap, density matrices, Mulliken and Löwdin 
populations for CO and CO adsorbed on a Cu slab.

I am facing two problems:

   1. 
   
   *MO warning for CO calculations:*
   When using the def2-TZVP basis set with smearing, CP2K reports:
   *** WARNING in qs_environment.F:1486 :: More added MOs requested than 
   *** *** available. The full set of unoccupied MOs will be used. Use *** *** 
   'ADDED_MOS -1' to always use all available MOs and to get rid of this *** 
   *** warning. 
   
   This occurs even though I explicitly requested 30 MOs in the input.
   2. 
   
   *Cholesky decomposition failure for Cu slab:*
   When using def2-TZVP for Cu (with or without smearing and CO), the 
   calculation aborts with:
   [ABORT] Cholesky decompose failed: matrix is not positive definite or 
   ill-conditioned fm/cp_cfm_basic_linalg.F:877 
   
I would like to ask for guidance on how to avoid these issues. Should I 
adjust the number of MOs, EPS_DEFAULT, smearing, or other SCF parameters 
for metal systems with large basis sets? I have no idea about this.

I have attached the input XYZ files and example output files for reference.

Thank you very much for your time and help.

-- 
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+unsubscribe at googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/782ee59a-9bae-42dd-a71f-a2710f75f0afn%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20251120/c72d1298/attachment-0001.htm>