## Name

PSOLVE — Set the sparse matrix solver method.

## Synopsis

PSOLVE=CGDMS

PSOLVE=QNDMS

PSOLVE=PDM

PSOLVE=CEM

PSOLVE=CEM=`n`

PSOLVE=SIGN

## Description

These methods solve for the density matrix directly, without diagonalization of the Fock matrix
or the formation of molecular orbitals. As such, these are different from the standard SCF engine.
Generally, these methods are used on large molecues along with the keyword
SPARSE to take advantage of the
natural sparsity of the system. If SPARSE
is not specified, the calculation will be performed using full matrices. (The full matrix case is
primarily for testing purposes rather than serious calculations.) For details on these methods and on
sparse matricies, see sparse matrix.

### PSOLVE=CGDMS

Solve for the density matrix using the conjugate-gradient density matrix (CGDMS) method.
The first SCF iteration, however, is done using the PDM method.

### PSOLVE=QNDMS

Solve for the density matrix using the quasi-newton density matrix search (QNDMS) method.
This is similar to CGDMS but avoids the line seach step. The first SCF iteration, however, is done
using the PDM method.

### PSOLVE=PDM

Solve for the density matrix using the the purification of the density matrix (PDM) method.

### PSOLVE=CEM

Solve for the density matrix using the the Chebyshev expansion method (CEM). The Chebyshev
expansion is carried out to order 5.

### PSOLVE=CEM=`n`

Solve for the density matrix using the the Chebyshev expansion method (CEM). The Chebyshev
expansion is carried out to order `n`

.

### PSOLVE=SIGN

Solve for the density matrix using the the sign-matrix method. This is very similar to PDM.