6.22 Variational Two-Electron Reduced-Density-Matrix Methods

6.22.3 v2RDM Job Control

(June 30, 2021)

RDM_POSITIVITY
       Indicates positivity conditions enforced in the v2RDM optimization.
TYPE:
       STRING
DEFAULT:
       DQG
OPTIONS:
       DQG, Two-electron conditions DQGT1 Two-electron conditions plus the T1 partial three-electron conditions DQGT2 Two-electron conditions plus the T2 partial three-electron conditions DQGT1T2 Two-electron conditions plus the T1 and T2 partial three-electron conditions DQG3POS Two-electron conditions plus the full three-electron conditions
RECOMMENDATION:
       For high-accuracy, use DQG3POS or DQGT2, although such computations become impractical for large active spaces. For large active spaces (e.g., n > 16 for CAS(n, n)), use DQG.

RDM_CONSTRAIN_SPIN
       Indicates if the spin-constraints are enforced.
TYPE:
       BOOLEAN
DEFAULT:
       TRUE
OPTIONS:
       TRUE Enforce spin-constraints. FALSE Do not enforce spin-constraints.
RECOMMENDATION:
       Use default.

RDM_E_CONVERGENCE
       The threshold for the primal-dual energy gap.
TYPE:
       INTEGER
DEFAULT:
       4
OPTIONS:
       N for a threshold of 10-N
RECOMMENDATION:
       Increase for gradient computations.

RDM_EPS_CONVERGENCE
       The threshold for the error in the primal and dual constraints.
TYPE:
       INTEGER
DEFAULT:
       4
OPTIONS:
       N for a threshold of 10-N
RECOMMENDATION:
       Increase for gradient computations.

RDM_MAXITER
       Maximum number of diagonalization steps in the BPSDP solver.
TYPE:
       INTEGER
DEFAULT:
       50000
OPTIONS:
       N>0
RECOMMENDATION:
       Increase for computations that are difficult to converge.

RDM_CG_CONVERGENCE
       The minimum threshold for the conjugate gradient solver.
TYPE:
       INTEGER
DEFAULT:
       12
OPTIONS:
       N for a threshold of 10-N
RECOMMENDATION:
       Should be at least (RDM_EPS_CONVERGENCE+2).

RDM_CG_MAXITER
       Maximum number of iterations for each conjugate gradient computations in the BPSDP algorithm.
TYPE:
       INTEGER
DEFAULT:
       1000
OPTIONS:
       N>0
RECOMMENDATION:
       Use default unless problems arise.

RDM_TAU
       Step-length parameter used in the BPSDP solver.
TYPE:
       INTEGER
DEFAULT:
       10
OPTIONS:
       N for a value of 0.1 * N
RECOMMENDATION:
       RDM_TAU should range between 10 and 16 for 1.0τ1.6.

RDM_MU_UPDATE_FREQUENCY
       The number of v2RDM iterations after which the penalty parameter μ is updated.
TYPE:
       INTEGER
DEFAULT:
       200
OPTIONS:
       N>0
RECOMMENDATION:
       Change if convergence problems arise.

RDM_TPDM_GUESS
       Initial guess for the RDMs
TYPE:
       STRING
DEFAULT:
       HF_GUESS
OPTIONS:
       HF_GUESS Use RDMs from Hartree-Fock calculations as the initial density for the semidefinite solver RANDOM_GUESS Use random numbers as the initial density for the semidefinite solver
RECOMMENDATION:
       Use default unless convergence problems arise.

RDM_DIAGONALIZER
       The algorithm used to diagonalize matrices inside semidefinite programming.
TYPE:
       INTEGER
DEFAULT:
       11
OPTIONS:
       0 Use parallel LAPACK function DSYEV 1 Use parallel LAPACK function DSYEVD 10 Use multiple simultaneous calls to serial LAPACK function DSYEV 11 Use multiple simultaneous calls to serial LAPACK function DSYEVD
RECOMMENDATION:
       Use default. Under certain circumstances (e.g., low symmetry), algorithm 1 may be faster.

RDM_PRINT
       Controls the amount of printing.
TYPE:
       INTEGER
DEFAULT:
       0
OPTIONS:
       0 Print minimal information. 1 Print information about all iterations.
RECOMMENDATION:
       Use 1 to analyze convergence issues.

RDM_ORBOPT_GRADIENT_CONVERGENCE
       The threshold for the orbital gradient during orbital optimization.
TYPE:
       INTEGER
DEFAULT:
       4
OPTIONS:
       N for threshold of 10-N
RECOMMENDATION:
       Tighten for gradient computations.

RDM_ORBOPT_ENERGY_CONVERGENCE
       The threshold for energy convergence during orbital optimization.
TYPE:
       INTEGER
DEFAULT:
       8
OPTIONS:
       N for threshold of 10-N
RECOMMENDATION:
       Tighten for gradient computations.

RDM_ORBOPT_MAXITER
       The maximum number of orbital optimization steps each time the orbital optimization routine is called.
TYPE:
       INTEGER
DEFAULT:
       20
OPTIONS:
       N>0
RECOMMENDATION:
       Use default unless convergence problems arise.

RDM_ORBOPT_FREQUENCY
       The number of v2RDM iterations after which the orbital optimization routine is called.
TYPE:
       INTEGER
DEFAULT:
       500
OPTIONS:
       N>0
RECOMMENDATION:
       Use default unless convergence problems arise.