Error with transition metal calculations?

I am getting inaccurate results for CCSD(T) calculation on Br as compared to CCCBDB and a couple of other software packages.  The input file is:

basis=aug-cc-pVTZ
calc=CCSD(T)
mem=5000MB
core=frozen
scftol=8
unit=angs
geom=xyz
1

   Br        0.0000000000    0.0000000000   -0.0755311173

The HF energy is correct; but both the MP2 energy and CCSD(T) are not.  I have been trying to figure out the problem with no success.

I ran the same calculation for Cl with no issues (HF, MP2, CCSD(T) match benchmarks).  I also ran Ca using a cc-pVTZ basis set with no issues (HF, MP2, CCSD(T) match benchmarks).

I ran Br all electron (core=corr) with no issues (HF, MP2, CCSD(T) all match benchmarks).  That led me to wonder whether there was an issue with frozen core.

But then I tried Sc with both all electron and frozen core, and HF, MP2 and CCSD(T) were all inaccurate.

I also tried without density-fitting approximation (ccprog=mrcc, dfbasis_scf=none, and dfbasis_cor=none), which did not solve the problem.

I also tried different basis sets and used freshly downloaded basis sets with no success.

I would greatly appreciate any assistance.

Regards,

Berj
 

In mrcc, only the 4p electrons of Br are correlated, while in other programs, the 4s and 3d electrons can also be correlated. Please set core=9 to reproduce those results.

Thank you very much. That was exactly the issue. For these types of calculations, it seems like MRCC's default frozen core for Br (core=14) is more appropriate to the cc basis sets that are being used than the other programs. But it is helpful to have the flexibility.

Thank you, again.

Regards,

Berj Parseghian