Often the largest source of error in a WF-in-DFT calculation, relative to a WF calculation on the full system, results from evaluation of the exchange-correlation contributions to the embedding potential at the DFT level. This error can be largely removed by evaluating the exchange-correlation contributions to the embedding potential using MP2 theory. It is thus possible to calculate an MP2-level correction to the WF-in-DFT energies, as described in the following paper:

J. D. Goodpaster, T. A. Barnes, F. R. Manby, and T. F. Miller III, J. Chem. Phys. 140, 18A507 (2014)

The MP2 correction can be invoked by running the `EMBED,PROJ,MP2_CORR` command:

`EMBED,PROJ,MP2_CORR``{ks}``{embed,proj;atoms,atom1,atom2,...}``{hf}``{ccsd(t)}``{embed,proj,mp2_corr}`

By default, the program will calculate the MP2 correction using LMP2; however, any local method can be used. The program will search for a procedure named `PROC MP2_CORR` and will use any location correlation method specified there. For example, including the following procedure at the top of the input file:

`PROC MP2_CORR`

`{df-lmp2,THRBP=1;orbital,5300.2} ! Must be 5300.2`

`ENDPROC`

will calculate the MP2 correction with all pairs correlated using density fitting.

Similarly,

`PROC MP2_CORR`

`{df-lccsd,THRBP=0.99,throsv=1.d-4;orbital,5300.2} ! Must be 5300.2`

`ENDPROC`

will calculate the MP2 correction using the orbital-specific virtual LCCSD method.

molpro@molpro.net 2019-03-19