CPHF
References: DelBen2013
Parameters influencing the solution of the Z-vector equations in MP2 gradients calculations. [Edit on GitHub]
Keywords
Keyword descriptions
- DO_POLAK_RIBIERE
Type: logical
Default: F
Lone keyword: T
Usage: DO_POLAK_RIBIERE TDescription: Use a Polak-Ribiere update of the search vector in CG instead of the Fletcher Reeves update. Improves the convergence with modified step sizes. Ignored with other methods than CG.
- ENFORCE_DECREASE
Type: logical
Default: F
Lone keyword: T
Usage: ENFORCE_DECREASE TDescription: Restarts if residual does not decrease.
- EPS_CONV
Type: real
Default: 1.00000000E-004
Usage: EPS_CONV 1.0E-6Description: Convergence threshold for the solution of the Z-vector equations. The Z-vector equations have the form of a linear system of equations Ax=b, convergence is achieved when |Ax-b|<=EPS_CONV.
- MAX_ITER
Type: integer
Default: 30
Aliases: MAX_NUM_ITER
Usage: MAX_ITER 50Description: Maximum number of iterations allowed for the solution of the Z-vector equations.
- RECALC_RESIDUAL
Type: logical
Default: F
Lone keyword: T
Usage: RECALC_RESIDUAL TDescription: Recalculates residual in every step.
- RESTART_EVERY
Type: integer
Default: 5
Usage: RESTART_EVERY 5Description: Restart iteration every given number of steps.
- SCALE_STEP_SIZE
Type: real
Default: 1.00000000E+000
Usage: SCALE_STEP_SIZE 1.0Description: Scaling factor of each step.
- SOLVER_METHOD
Type: enum
Default: POPLE
Usage: SOLVER_METHOD POPLEDescription: Chose solver of the z-vector equations.
Valid values:
POPLEPople’s method (Default).CGConjugated gradient method (equivalent to Pople).RICHARDSONRichardson iterationSDSteepest Descent iteration