ccprog=ccsd fails for dens=2

Dear Colleague,

I was running a CCSD gradient calculation using both of "ccsd" and "mrcc" modules in MRCC. I found a strange error of end-of-file during read using ccprog=ccsd with dens=2, while ccprog=mrcc could finish the density/gradient calculation normally. Here is the input and outputs are attached.
MINP:
unit=bohr
mem=2GB
basis=cc-pVDZ
dfbasis_cor=none
dfbasis_scf=none
calc=ccsd
ccprog=mrcc
dens=2
geom=xyz
4

C 1.006316674509 -2.002278344981 0.000000000000
H 3.043585491612 -2.415224157655 0.000000000000
O -0.620167656881 -3.843378105614 0.000000000000
H 0.353819664117 -0.056868714559 0.000000000000

Error message with ccprog=ccsd:
Number of integral batches: 1
forrtl: severe (24): end-of-file during read, unit 26, file /temp/jsyu/mtest/CCDENSITIES

Fatal error in prop.
Program will stop.


The 2016-07-15 binary was download from the MRCC website. Even I tried to re-compile from the source code the same problem remains. Could I get some possible solutions to this issue? Thank you!

Sincerely,
Kenny Yu

Dear Kenny,
Gradients are not implemented in the ccsd module, that is, ccprog=ccsd and dens=2 are not compatible.

Dear Prof. Kallay,

Actually my purpose was to get the conventional (non-DF) MP2 gradient via the "ccsd" module as described in the manual. If the "ccsd" module cannot be used to calculate the analytic gradient, is there any keyword combination available to do that? Thank you for your help in advance.

Sincerely,
Kenny Yu

Dear Kenny,
Conventional MP2 gradients are not available. However, you can simulate conventional MP2 gradient calculations with very small error using the DF-MP2 gradient code and very large auxiliary basis sets.
E.g.:
basis=cc-pVDZ
dfbasis_cor=aug-cc-pV6Z-RI
dfbasis_scf=aug-cc-pV6Z-RI