LNO-CCSD(T) calculation converged to wrong HF energy (def2-qzvpp basis)

Dear MRCC support,I recently ran a series of LNO-CCSD(T) calculations on systems containing a ferrocene group and potentially an additional Ni or Pd atom. The only auxiliary basis set available for Fe, Ni and Pd atoms is the def2-qzvpp. (def2-qzvpp-ri and def2-qzvpp-ri-jk).Most of the calculations had problems with the HF converge, i.e. the energy converged to a higher value than it was obtained in the first or second iteration step. A fewer of them converged correctly (compared to what we expect based on DFT results)I attached a successful and a “failed” case to my topic entry with the complete output folder.I have also noticed that the aug-cc-pvtz basis set in a simple HF calculation converged to the right energy for one structure that I tested on, but then I could not read in the SCFDENISTIES properly to use it as an initial SCF guess for the LNO-CCSD(T) job.Therefore, I would like to ask if there is a way to avoid this convergence failure by using an additional parameter, apply a better SCF guess, or a different basis set or by any other possible solution.Thank you for your help,Eszter

Dear Eszter,
Could you also post the relevant output files?

Excuse me, I was supposed to attach the following files to my previous entry:This is the output file which converged to the wrong HF energy:

 
 And this is the one that worked:  Thank you for your response,Eszter

Dear Eszter,
The best solution is to restart the HF from the density evaluated with the def2-SVP basis set. To that end, you need:
calc=LNO-CCSD(T)
basis=def2-qzvpp
scfiguess=small
basis_sm=def2-SVP
 

Thank you for your answer, indeed this solved the convergence problem.

Best wishes,
Eszter