[molpro-user] ump2 method in molpro is different from the one in gaussian?

[lydia]

Hi all,
I found the ump2 calculation in molpro get different energy from the
one of Gaussian, but they have the same UHF energies.

This is from molpro:
The input file is
{uhf;orbprint,10;maxit,200;
 wf,85,1,1;closed,16,10,11,5;occ,17,10,11,5;}
{ump2;
orbital,IGNORE_ERROR;
closed,16,10,11,5;
occ,17,10,11,5;
 wf,85,1,1;}
and output file is
1PROGRAM * UMP2 (Unrestricted open-shell)
 Reference energy                   -1463.161083731141
 Correlation energy                    -2.548208935082
 !UMP2 STATE 1.4 Energy             -1465.709292666223

This is from gaussian:
HF=-1463.1610804
MP2=-1467.12587

Since they have the same UHF energy, the wavefunction, orbitals etc
are the same. The difference can only come from the correlation energy
calculation.
So I want to know the methodolody for UMP2 is different in Molpro and
gaussian? I checked the molpro manule and did not get any useful
information for that. If you have some experiences, please share with
me. Thanks in advance.

lydia