Very high RSME and RRSME in RESP charge fitting

[Prateek Chowdhury]

Hello...

I am new to cp2k. I started with geometric optimization and partial charge calculation of MoS2 (hexagonal) and CdS (cubic) crystals (2*2*1 supercells for the time being) respectively. For charge calculation, I was using RESP fitting method. For MoS2, the fitting was good showing sufficiently low RRMSE (~ 0.09 - 0.1) and RMSE (~0.00005), respectively, and the fitted data matched with literature too. 

Conversely, during the same exercise for CdS, inspite of achieving physically reasonable partial charges for both Cd (~ +2.00) and S (~ -2.00), the RRMSE (~ 0.98 - 0.99) and RMSE (~ 0.0008) came out to be extremely high. I wonder why this is happening in doing the similar exercise, despite using the similar PPs and Basis sets. I shall be highly obliged if anyone helps me out with this. I have attached my input files and the PPs and Basis sets used for both the cases below.

Cd (PP): GTH-PBE-q12

Cd (BS): DZVP-MOLOPT-SR-GTH-q12

S (PP): GTH-PBE-q6

S (BS): TZVP-MOLOPT-GTH-q6

Mo (PP): GTH-PBE-q14

Mo (BS): DZVP-MOLOPT-SR-GTH

S (PP): GTH-PBE-q6 

S (BS): DZVP-MOLOPT-GTH

I am adding RESP charge printed sections for both the cases in two separate files too.

Thanks in advance...

Prateek.