*MP2CAL

Introduction

This section gives directives for the integral-direct closed-shell second-order Møller-Plesset perturbation theory (MP2) calculation. The default algorithm is described here and is a modification of the original algorithm published in Ref. [Laerdahl1997].

Note that default is to include all positive-energy (electronic) orbitals in the MP2 calculation. You specify .OCCUPIED and .VIRTUAL if you want to do the MP2 calculation with a subset of these orbitals.

Basic keywords

.OCCUPIED

Specify the active subset of the doubly occupied positive-energy (electronic) orbitals from the SCF.

For each fermion irrep, give an Specification of orbital strings of active occupied orbitals.

Default: All the occupied electronic orbitals.

.VIRTUAL

Specify the active subset of the empty virtual orbitals from the SCF.

For each fermion irrep, give an Specification of orbital strings of active virtual orbitals.

Default: All the unoccupied virtual orbitals.

Advanced options

.INTFLG

Specify what two-electron integrals to include in the direct MP2 calculation (default: .INTFLG under **HAMILTONIAN).

.SCREEN

Screening threshold in the 4-index transformation.

A negative number deactivates screening.

Default:

.SCREEN
 1.0D-14

Programmers options

.PRINT

Print level.

Default:

.PRINT
 0

.VIRTHR

Threshold for neglecting high-lying virtual orbitals. Neglect all orbitals with energy above this threshold. This option is retained for backwards compatibility, it is advised to use the more general energy option in the Specification of orbital strings input.

Arguments: Real DMP2_VIRTHR.

Default: No threshold.

.IJTSK

Max. number of I (or J) orbitals (occupied orbitals) in each calculation batch. Usually one batch is sufficient but larger calculations may be possible by reducing the batch size. Reducing the batch size to half will approximately reduce the memory requirements to 1/4 and increase the CPU time by a factor 4.

Arguments: Integer IJTSK.

Default: All active occupied electronic solutions.

.SCLMEM

Internal memory available for scalar untransformed integrals. Larger batches will be written to disk.

Arguments: Integer MAXSCL.

Default: Set by program and is usually sufficient.

.ORGALG

Use original alorithm as described in Ref. [Laerdahl1997].

Default: Use new algorithm.