possible inaccuracies in calculating ECP integrals

Dear Mihály,
It seems that there are some instabilities in the procedure of calculating ECP integrals for the case of ECP with "rich expansion" (which is required for a good accuracy) such as GRECP.

Please, see attached example - Th^+4 cation in minimal basis set and GRECP:
If I use CFOUR to compute RHF I get SCF=-403.540018494225194 (see OUT_CFOUR)
If I use MRCC to compute RHF I get SCF=-489.9096178732527846 (see mrcc.out_MRCC)

Addition qualitative test: if I use basis set from GRECP but Stuttgart ECP (with very modest expansion) I get SCF=-402.9681956104190022 (see mrcc.out_Stutt_MRCC). It is in qualitative agreement with CFOUR result for GRECP as expected (basis set for GRECP is not optimal for Stuttgart ECP, besides, we compare results for different ECPs which is in principle a very qualitative operation).

The problem is that MRCC result for GRECP is not in even in a qualitative agreement with CFOUR one.

Thus, it seems that there could be some inaccuracies in calculating ECP integrals in some cases?

Best,
Leonid Skripnikov.

Dear Leonid,
It was an error at the reading of the ECPs. Please edit file integ.f, and in subroutines readecp and getecpnames change the line
do while(c1.eq.' ')
as
do while(c1.eq.' '.or.(ichar(c1).ge.43.and.ichar(c1).le.57))

Thank you very much for reporting this.

Dear Mihály,
thank you!
All is working now.