SCF¶
Performs self consistent field (Hartree-Fock and Density Functional Theory) computations. These are the starting points for most computations, so this code is called in most cases.
General Wavefunction Info¶
BASIS¶
Primary basis set
Type: string
Possible Values: basis string
Default: No Default
CHOLESKY_TOLERANCE¶
Tolerance for Cholesky decomposition of the ERI tensor
Type: conv double
Default: 1e-4
DF_BASIS_SCF¶
Auxiliary basis set for SCF density fitting computations. Defaults to a JKFIT basis.
Type: string
Possible Values: basis string
Default: No Default
DF_SCF_GUESS¶
GUESS¶
The type of guess orbitals. See Initial Guess for what the options mean and what the defaults are.
Type: string
Possible Values: AUTO, CORE, GWH, SAD, SADNO, SAP, SAPGAU, HUCKEL, MODHUCKEL, READ
Default: AUTO
GUESS_MIX¶
Mix the HOMO/LUMO in UHF or UKS to break alpha/beta spatial symmetry. Useful to produce broken-symmetry unrestricted solutions. Notice that this procedure is defined only for calculations in C1 symmetry.
Type: boolean
Default: false
GUESS_PERSIST¶
If true, then repeat the specified guess procedure for the orbitals every time - even during a geometry optimization.
Type: boolean
Default: false
INTS_TOLERANCE¶
Screening threshold for the chosen screening method (SCHWARZ, CSAM, DENSITY) Absolute value below which TEI are neglected.
Type: conv double
Default: 1e-12
MAX_MEM_BUF¶
Max memory per buf for PK algo REORDER, for debug and tuning
Type: integer
Default: 0
MOLDEN_WRITE¶
Do write a MOLDEN output file? If so, the filename will end in .molden, and the prefix is determined by WRITER_FILE_LABEL (if set), or else by the name of the output file plus the name of the current molecule.
Type: boolean
Default: false
ORBITALS_WRITE¶
File name (case sensitive) to which to serialize Wavefunction orbital data.
Type: string
Default: No Default
PRINT_BASIS¶
Do print the basis set?
Type: boolean
Default: false
PRINT_MOS¶
Do print the molecular orbitals?
Type: boolean
Default: false
QCHF¶
Do perform a QCHF computation?
Type: boolean
Default: false
REFERENCE¶
Reference wavefunction type. Cfour Interface: Keyword translates into CFOUR_REFERENCE
Type: string
Possible Values: RHF, ROHF, UHF, CUHF, RKS, UKS
Default: RHF
SAPGAU_BASIS¶
The potential basis set used for the SAPGAU guess
Type: string
Default: sap_helfem_large
SAVE_JK¶
Keep JK object for later use?
Type: boolean
Default: false
SCF_MEM_SAFETY_FACTOR¶
Memory safety factor for allocating JK
Type: double
Default: 0.75
SCF_PROPERTIES¶
SCF Properties to calculate after an energy evaluation. Note, this keyword is not used for property evaluations.
Type: array
Default: No Default
S_CHOLESKY_TOLERANCE¶
Tolerance for partial Cholesky decomposition of overlap matrix.
Type: conv double
Default: 1e-8
S_ORTHOGONALIZATION¶
SO orthogonalization: automatic, symmetric, or canonical?
Type: string
Possible Values: AUTO, SYMMETRIC, CANONICAL, PARTIALCHOLESKY
Default: AUTO
S_TOLERANCE¶
Minimum S matrix eigenvalue to allow before linear dependencies are removed.
Type: conv double
Default: 1e-7
Convergence Control/Stabilization¶
BASIS_GUESS¶
Accelerate convergence by performing a preliminary SCF with this small basis set followed by projection into the full target basis. A value of
TRUE
turns on projection using the Defaults small basis set 3-21G, pcseg-0, or def2-SV(P).
Type: string
Default: FALSE
DAMPING_CONVERGENCE¶
The density convergence threshold after which damping is no longer performed, if it is enabled. It is recommended to leave damping on until convergence, which is the default. Cfour Interface: Keyword translates into CFOUR_SCF_DAMPING
Type: conv double
Default: 1.0e-18
DAMPING_PERCENTAGE¶
The amount (percentage) of damping to apply to the early density updates. 0 will result in a full update, 100 will completely stall the update. A value around 20 (which corresponds to 20% of the previous iteration’s density being mixed into the current density) could help to solve problems with oscillatory convergence.
Type: double
Default: 0.0
DF_BASIS_GUESS¶
When BASIS_GUESS is active, run the preliminary scf in density-fitted mode with this as fitting basis for the small basis set. A value of
TRUE
turns on density fitting with the default basis, otherwise the specified basis is used.
Type: string
Possible Values: basis string
Default: FALSE
DIIS¶
Do use DIIS extrapolation to accelerate convergence?
Type: boolean
Default: true
DIIS_MAX_VECS¶
Maximum number of error vectors stored for DIIS extrapolation
Type: integer
Default: 10
DIIS_MIN_VECS¶
Minimum number of error vectors stored for DIIS extrapolation. Will be removed in v1.7.
Type: integer
Default: 2
DIIS_RMS_ERROR¶
Use RMS error instead of the more robust absolute error?
Type: boolean
Default: true
DIIS_START¶
The minimum iteration to start storing DIIS vectors and performing ADIIS/EDIIS.
Type: integer
Default: 1
D_CONVERGENCE¶
Convergence criterion for SCF density, defined as the RMS or maximum absolute value of the orbital gradient. See Table SCF Convergence & Algorithm for default convergence criteria for different calculation types. Cfour Interface: Keyword translates into CFOUR_SCF_CONV
Type: conv double
Default: 1e-6
E_CONVERGENCE¶
Convergence criterion for SCF energy. See Table SCF Convergence & Algorithm for default convergence criteria for different calculation types.
Type: conv double
Default: 1e-6
FAIL_ON_MAXITER¶
Fail if we reach maxiter without converging?
Type: boolean
Default: true
INCFOCK¶
Do perform incremental Fock build?
Type: boolean
Default: false
INCFOCK_CONVERGENCE¶
The density threshold at which to stop building the Fock matrix incrementally
Type: conv double
Default: 1.0e-5
INCFOCK_FULL_FOCK_EVERY¶
Frequency with which to compute the full Fock matrix if using INCFOCK . N means rebuild every N SCF iterations to avoid accumulating error from the incremental procedure.
Type: integer
Default: 5
LEVEL_SHIFT¶
Do use a level shift?
Type: double
Default: 0.0
LEVEL_SHIFT_CUTOFF¶
DIIS error at which to stop applying the level shift
Type: double
Default: 1e-2
LINK_INTS_TOLERANCE¶
The screening tolerance used for ERI/Density sparsity in the LinK algorithm
Type: conv double
Default: 1.0e-12
MAXITER¶
Maximum number of iterations. Cfour Interface: Keyword translates into CFOUR_SCF_MAXCYC
Type: integer
Default: 100
MOM_OCC¶
The absolute indices of orbitals to excite from in MOM (+/- for alpha/beta)
Type: array
Default: No Default
MOM_START¶
The iteration to start MOM on (or 0 for no MOM)
Type: integer
Default: 0
MOM_VIR¶
The absolute indices of orbitals to excite to in MOM (+/- for alpha/beta)
Type: array
Default: No Default
SCF_INITIAL_ACCELERATOR¶
Use a method to accelerate initial SCF convergence? Use
NONE
for DIIS alone (if enabled) andEDIIS
orADIIS
to have both the chosen accelerator and DIIS (if enabled). For restricted-open references,EDIIS
andADIIS
have no effect.
Type: string
Possible Values: NONE, EDIIS, ADIIS
Default: ADIIS
SCF_INITIAL_FINISH_DIIS_TRANSITION¶
SCF error at which to complete the linear interpolation between DIIS steps and steps of the initial SCF accelerator Value taken from Garza and Scuseria, DOI: 10.1063/1.4740249
Type: double
Default: 1.0e-4
SCF_INITIAL_START_DIIS_TRANSITION¶
SCF error at which to start the linear interpolation between DIIS steps and steps of the initial SCF accelerator. Value taken from Garza and Scuseria, DOI: 10.1063/1.4740249
Type: double
Default: 1.0e-1
SOLVER_CONVERGENCE¶
Convergence threshold (max 2-norm) for numerical solvers (instability analysis and CPHF/CPKS).
Type: conv double
Default: 1.0e-6
SOLVER_MAXITER¶
Maximum iterations for numerical solvers (instability analysis and CPHF/CPKS).
Type: integer
Default: 100
SOLVER_N_GUESS¶
Number of guess vectors per root for instability analysis.
Type: integer
Default: 1
SOLVER_N_ROOT¶
Number of roots to converge for all irreps during instability analysis. (Overridden by SOLVER_ROOTS_PER_IRREP.)
Type: integer
Default: 1
SOLVER_ROOTS_PER_IRREP¶
Number of roots to converge, per irrep, during instability analysis. (Overrides SOLVER_N_ROOT.)
Type: array
Default: No Default
SOSCF¶
Do use second-order SCF convergence methods?
Type: boolean
Default: false
SOSCF_CONV¶
Second order convergence threshold. Cease microiterating at this value.
Type: conv double
Default: 5.0e-3
SOSCF_MAX_ITER¶
Maximum number of second-order microiterations to perform.
Type: integer
Default: 5
SOSCF_MIN_ITER¶
Minimum number of second-order microiterations to perform.
Type: integer
Default: 1
SOSCF_PRINT¶
Do we print the SOSCF microiterations?.
Type: boolean
Default: false
SOSCF_START_CONVERGENCE¶
When to start second-order SCF iterations based on gradient RMS.
Type: conv double
Default: 1.0e-2
STABILITY_ANALYSIS¶
Whether to perform stability analysis after convergence. NONE prevents analysis being performed. CHECK will print out the analysis of the wavefunction stability at the end of the computation. FOLLOW will perform the analysis and, if a totally symmetric instability is found, will attempt to follow the eigenvector and re-run the computations to find a stable solution.
Type: string
Possible Values: NONE, CHECK, FOLLOW
Default: NONE
Fractional Occupation UHF/UKS¶
FRAC_DIIS¶
Do use DIIS extrapolation to accelerate convergence in frac?
Type: boolean
Default: true
FRAC_LOAD¶
Do recompute guess from stored orbitals?
Type: boolean
Default: false
FRAC_OCC¶
The absolute indices of occupied orbitals to fractionally occupy (+/- for alpha/beta)
Type: array
Default: No Default
FRAC_RENORMALIZE¶
Do renormalize C matrices prior to writing to checkpoint?
Type: boolean
Default: true
FRAC_START¶
The iteration to start fractionally occupying orbitals (or 0 for no fractional occupation)
Type: integer
Default: 0
FRAC_VAL¶
The occupations of the orbital indices specified above (\(0.0\le {\rm occ} \le 1.0\))
Type: array
Default: No Default
Environmental Effects¶
EXTERN¶
An ExternalPotential (built by Python or nullptr/None)
Type: boolean
Default: false
ONEPOT_GRID_READ¶
Read an external potential from the .dx file?
Type: boolean
Default: false
PERTURB_DIPOLE¶
An array of length three describing the magnitude (atomic units) of the dipole field in the {x,y,z} directions
Type: array
Default: No Default
PERTURB_H¶
Do perturb the Hamiltonian?
Type: boolean
Default: false
PERTURB_MAGNITUDE¶
Size of the perturbation (applies only to dipole perturbations). Deprecated - use PERTURB_DIPOLE instead
Type: double
Default: 0.0
PERTURB_WITH¶
The operator used to perturb the Hamiltonian, if requested. DIPOLE_X, DIPOLE_Y and DIPOLE_Z will be removed in favor of the DIPOLE option in the future
Type: string
Possible Values: DIPOLE, DIPOLE_X, DIPOLE_Y, DIPOLE_Z, EMBPOT, SPHERE, DX
Default: DIPOLE
PHI_POINTS¶
Number of azimuthal grid points for spherical potential integration
Type: integer
Default: 360
RADIUS¶
Radius (bohr) of a hard-sphere external potential
Type: double
Default: 10.0
R_POINTS¶
Number of radial grid points for spherical potential integration
Type: integer
Default: 100
THETA_POINTS¶
Number of colatitude grid points for spherical potential integration
Type: integer
Default: 360
THICKNESS¶
Thickness (bohr) of a hard-sphere external potential
Type: double
Default: 20.0
DFSCF Algorithm¶
DF_BUMP_R0¶
Bump function min radius
Type: double
Default: 0.0
DF_BUMP_R1¶
Bump function max radius
Type: double
Default: 0.0
DF_DOMAINS¶
FastDF geometric fitting domain selection algorithm
Type: string
Possible Values: DIATOMIC, SPHERES
Default: DIATOMIC
DF_INTS_NUM_THREADS¶
Number of threads for integrals (may be turned down if memory is an issue). 0 is blank
Type: integer
Default: 0
DF_METRIC¶
FastDF Fitting Metric
Type: string
Possible Values: COULOMB, EWALD, OVERLAP
Default: COULOMB
DF_THETA¶
FastDF SR Ewald metric range separation parameter
Type: double
Default: 1.0
COSX Algorithm¶
COSX_INTS_TOLERANCE¶
Screening criteria for integrals and intermediates in COSX
Type: conv double
Default: 1.0e-11
COSX_MAXITER_FINAL¶
Controls SCF iteration behavior for the larger (i.e., final) COSX grid. -1 fully converges the SCF on the final grid if possible, ending early if MAXITER total SCF iterations are reached (failure). 0 disables the final COSX grid entirely. n runs up to n iterations on the final COSX grid, ending early if SCF convergence is reached (success) or if MAXITER total SCF iterations are reached (failure).
Type: integer
Default: 1
COSX_RADIAL_POINTS_FINAL¶
Number of radial points in final COSX grid.
Type: integer
Default: 35
COSX_RADIAL_POINTS_INITIAL¶
Number of radial points in initial COSX grid.
Type: integer
Default: 25
COSX_SPHERICAL_POINTS_FINAL¶
Number of spherical points in final COSX grid.
Type: integer
Default: 110
COSX_SPHERICAL_POINTS_INITIAL¶
Number of spherical points in initial COSX grid.
Type: integer
Default: 50
snLinK Algorithm¶
SNLINK_INTS_TOLERANCE¶
Screening criteria for integrals and intermediates in snLinK
Type: conv double
Default: 1.0e-11
SNLINK_RADIAL_POINTS¶
Number of radial points in snLinK grid.
Type: integer
Default: 70
SNLINK_RADIAL_SCHEME¶
Radial Scheme for snLinK grid. MURA is default here as it matches the GauXC default option
Type: string
Possible Values: MURA, TREUTLER, EM
Default: MURA
SNLINK_SPHERICAL_POINTS¶
Number of spherical points in snLinK grid.
Type: integer
Default: 302
SNLINK_USE_GPU¶
Use GPU for GauXC?
Type: boolean
Default: false
SAD Guess Algorithm¶
SAD_D_CONVERGENCE¶
Convergence criterion for SCF density in the SAD guess, analogous to D_CONVERGENCE
Type: conv double
Default: 1e-5
SAD_E_CONVERGENCE¶
Convergence criterion for SCF energy in the SAD guess, analogous to E_CONVERGENCE
Type: conv double
Default: 1e-5
DFT¶
DFT_ALPHA¶
The DFT Exact-exchange parameter
Type: double
Default: 0.0
DFT_ALPHA_C¶
The DFT Correlation hybrid parameter
Type: double
Default: 0.0
DFT_BASIS_TOLERANCE¶
DFT basis cutoff.
Type: conv double
Default: 1.0e-12
DFT_BS_RADIUS_ALPHA¶
Factor for effective BS radius in radial grid.
Type: double
Default: 1.0
DFT_DISPERSION_PARAMETERS¶
Parameters defining the dispersion correction. See Table -D Functionals for default values and Table Dispersion Corrections for the order in which parameters are to be specified in this array option. Unused for functionals constructed by user.
Type: array
Default: No Default
DFT_GRAC_ALPHA¶
The gradient regularized asymptotic correction alpha value
Type: double
Default: 0.5
DFT_GRAC_BETA¶
The gradient regularized asymptotic correction beta value
Type: double
Default: 40.0
DFT_GRAC_SHIFT¶
The gradient regularized asymptotic correction shift value
Type: double
Default: 0.0
DFT_NUCLEAR_SCHEME¶
Nuclear Scheme.
Type: string
Possible Values: TREUTLER, BECKE, NAIVE, STRATMANN, SBECKE
Default: TREUTLER
DFT_OMEGA¶
The DFT Range-separation parameter
Type: double
Default: 0.0
DFT_OMEGA_C¶
The DFT Correlation Range-separation parameter
Type: double
Default: 0.0
DFT_PRUNING_SCHEME¶
Select approach for pruning. Options
ROBUST
andTREUTLER
prune based on regions (proximity to nucleus) whileFLAT
P_GAUSSIAN
D_GAUSSIAN
P_SLATER
D_SLATER
LOG_GAUSSIAN
LOG_SLATER
prune based on decaying functions (experts only!). The recommended scheme isROBUST
.
Type: string
Default: NONE
DFT_RADIAL_POINTS¶
Number of radial points.
Type: integer
Default: 75
DFT_RADIAL_SCHEME¶
Radial Scheme.
Type: string
Possible Values: TREUTLER, BECKE, MULTIEXP, EM, MURA
Default: TREUTLER
DFT_SPHERICAL_POINTS¶
Number of spherical points (A Lebedev Points number).
Type: integer
Default: 302
DFT_SPHERICAL_SCHEME¶
Spherical Scheme.
Type: string
Possible Values: LEBEDEV
Default: LEBEDEV
DFT_V2_RHO_CUTOFF¶
Minima spin-summed density cutoff for the second derivative. Defaults to the density tolerance.
Type: double
Default: -1.0
DFT_VV10_B¶
Define VV10 parameter b
Type: double
Default: 0.0
DFT_VV10_C¶
Define VV10 parameter C
Type: double
Default: 0.0
DFT_VV10_POSTSCF¶
post-scf VV10 correction
Type: boolean
Default: false
DFT_VV10_RADIAL_POINTS¶
Number of radial points for VV10 NL integration.
Type: integer
Default: 50
DFT_VV10_SPHERICAL_POINTS¶
Number of spherical points (A Lebedev Points number) for VV10 NL integration.
Type: integer
Default: 146
LOCAL_CONVERGENCE¶
The convergence on the orbital localization procedure
Type: conv double
Default: 1e-12
LOCAL_MAXITER¶
The maxiter on the orbital localization procedure
Type: integer
Default: 200
NL_DISPERSION_PARAMETERS¶
Parameters defining the -NL/-V dispersion correction. First b, then C
Type: array
Default: No Default
SAVE_UHF_NOS¶
Save the UHF NOs
Type: boolean
Default: false
UHF_NOONS¶
The number of NOONs to print in a UHF calc
Type: string
Default: 3
TDSCF¶
TDSCF_COEFF_CUTOFF¶
Cutoff for printing excitations and de-excitations contributing to each excited state
Type: double
Default: 0.1
TDSCF_GUESS¶
Guess type, only ‘denominators’ currently supported
Type: string
Default: DENOMINATORS
TDSCF_MAXITER¶
Maximum number of TDSCF solver iterations
Type: integer
Default: 60
TDSCF_PRINT¶
Verbosity level in TDSCF
Type: integer
Default: 1
TDSCF_R_CONVERGENCE¶
Convergence threshold for the norm of the residual vector. If unset, default based on D_CONVERGENCE
Type: conv double
Default: 1e-4
TDSCF_STATES¶
Number of roots (excited states) we should seek to converge. This can be either an integer (total number of states to seek) or a list (number of states per irrep). The latter is only valid if the system has symmetry. Furthermore, the total number of states will be redistributed among irreps when symmetry is used.
Type: array
Default: No Default
TDSCF_TDA¶
Run with Tamm-Dancoff approximation (TDA), uses random-phase approximation (RPA) when false
Type: boolean
Default: false
TDSCF_TDM_PRINT¶
Which transition dipole moments to print out: - E_TDM_LEN : electric transition dipole moments, length representation - E_TDM_VEL : electric transition dipole moments, velocity representation - M_TDM : magnetic transition dipole moments
Type: array
Default: No Default
TDSCF_TRIPLETS¶
Controls inclusion of triplet states, which is only valid for restricted references. Valid options: - none : No triplets computed (default) - also : lowest-energy triplets and singlets included, in 50-50 ratio. Note that singlets are privileged, i.e. if seeking to converge 5 states in total, 3 will be singlets and 2 will be triplets. - only : Only triplet states computed
Type: string
Possible Values: NONE, ALSO, ONLY
Default: NONE
WCOMBINE¶
combine omega exchange and Hartree–Fock exchange into one matrix for efficiency? Disabled until fixed.
Type: boolean
Default: false
Expert General Wavefunction Info¶
PK_ALL_NONSYM¶
All densities are considered non symmetric, debug only.
Type: boolean
Default: false
PK_MAX_BUCKETS¶
Maximum numbers of batches to read PK supermatrix.
Type: integer
Default: 500
SCF_SUBTYPE¶
For certain SCF_TYPE algorithms that have internal sub-algorithms depending on available memory or other hardware constraints, allow the best sub-algorithm for the molecule and conditions (
AUTO
; usual mode) or forcibly select a sub-algorithm (usually only for debugging or profiling). Presently,SCF_SUBTYPE=DF
,SCF_SUBTYPE=MEM_DF
, andSCF_SUBTYPE=DISK_DF
can haveINCORE
andOUT_OF_CORE
selected; andSCF_TYPE=PK
can haveINCORE
,OUT_OF_CORE
,YOSHIMINE_OUT_OF_CORE
, andREORDER_OUT_OF_CORE
selected.
Type: string
Possible Values: AUTO, INCORE, OUT_OF_CORE, YOSHIMINE_OUT_OF_CORE, REORDER_OUT_OF_CORE
Default: AUTO
WFN¶
Wavefunction type
Type: string
Possible Values: SCF
Default: SCF
Expert Convergence Control/Stabilization¶
FOLLOW_STEP_INCREMENT¶
When using STABILITY_ANALYSIS = FOLLOW, the increment to modify FOLLOW_STEP_SCALE value if we end up in the same SCF solution.
Type: double
Default: 0.2
FOLLOW_STEP_SCALE¶
When using STABILITY_ANALYSIS
FOLLOW
, how much to scale the step along the eigenvector by. A full step of \(pi/2\) corresponds to a value of 1.0.
Type: double
Default: 0.5
MAX_ATTEMPTS¶
When using STABILITY_ANALYSIS
FOLLOW
, maximum number of orbital optimization attempts to make the wavefunction stable.
Type: integer
Default: 1
Expert Parallel Runtime¶
DISTRIBUTED_MATRIX¶
The dimension sizes of the distributed matrix
Type: array
Default: No Default
PARALLEL¶
Do run in parallel?
Type: boolean
Default: false
PROCESS_GRID¶
The dimension sizes of the processor grid
Type: array
Default: No Default
TILE_SZ¶
The tile size for the distributed matrices
Type: integer
Default: 512
Expert Misc.¶
SAPT¶
Are going to do SAPT? If so, what part?
Type: string
Default: FALSE
Expert DFSCF Algorithm¶
DF_FITTING_CONDITION¶
Fitting Condition, i.e. eigenvalue threshold for RI basis. Analogous to S_TOLERANCE
Type: double
Default: 1.0e-10
DF_INTS_IO¶
IO caching for CP corrections, etc. Changing this selects Disk_DF over Mem_DF. Note that setting this forces DiskDFJK when SCF_TYPE=DF.
Type: string
Possible Values: NONE, SAVE, LOAD
Default: NONE
Expert COSX Algorithm¶
COSX_BASIS_TOLERANCE¶
Screening criteria for basis function values on COSX grids
Type: conv double
Default: 1.0e-10
COSX_DENSITY_TOLERANCE¶
Screening criteria for shell-pair densities in COSX
Type: conv double
Default: 1.0e-10
COSX_OVERLAP_FITTING¶
Do reduce numerical COSX errors with overlap fitting?
Type: boolean
Default: true
COSX_PRUNING_SCHEME¶
Pruning scheme for COSX grids
Type: string
Default: ROBUST
Expert snLinK Algorithm¶
SNLINK_BASIS_TOLERANCE¶
Screening criteria for basis function values on snLinK grids
Type: conv double
Default: 1.0e-10
SNLINK_DENSITY_TOLERANCE¶
Screening criteria for shell-pair densities in snLinK
Type: conv double
Default: 1.0e-10
SNLINK_FORCE_CARTESIAN¶
Force snLinK to use cartesian coordinates
Type: boolean
Default: false
SNLINK_GPU_MEM¶
Proportion (in %) of available GPU memory to allocate to snLinK.
Type: boolean
Default: 90
SNLINK_GRID_BATCH_SIZE¶
Maximum number of grid points per grid block for GauXC
Type: integer
Default: 2048
SNLINK_INTEGRATOR_KERNEL¶
Integrator execution kernel for snLinK GauXC also has SHELLBATCHED, but it is incompatible with Psi4 due to not being yet implemented with sn-LinK.
Type: string
Possible Values: DEFAULT, INCORE
Default: DEFAULT
SNLINK_LOAD_BALANCER_KERNEL¶
Load Balancer kernel for snLinK
Type: string
Possible Values: DEFAULT, REPLICATED, REPLICATED-PETITE, REPLICATED-FILLIN
Default: DEFAULT
SNLINK_LWD_KERNEL¶
Integrator local work driver kernel for snLinK GauXC also has SCHEME1-CUTLASS, but it is disabled in Psi4 for now due to compile-time issues and requiring very modern CUDA CCs (>=80)
Type: string
Possible Values: DEFAULT, REFERENCE, SCHEME1, SCHEME1-MAGMA
Default: DEFAULT
SNLINK_MOL_WEIGHTS_KERNEL¶
Molecular Weights kernel for snLinK
Type: string
Possible Values: DEFAULT
Default: DEFAULT
SNLINK_PRUNING_SCHEME¶
Pruning scheme for snLinK grids
Type: string
Possible Values: ROBUST, TREUTLER, NONE
Default: ROBUST
SNLINK_REDUCTION_KERNEL¶
Integrator reduction kernel for snLinK GauXC also has NCCL, but it is incompatible with Psi4 due to requiring MPI.
Type: string
Possible Values: DEFAULT, BASICMPI
Default: DEFAULT
SNLINK_USE_DEBUG_GRID¶
Overwrite sn-LinK grid options with debug grid matching GauXC’s Ultrafine grid spec
Type: integer
Default: false
Expert SAD Guess Algorithm¶
DF_BASIS_SAD¶
Density fitting basis used in SAD
Type: string
Possible Values: basis string
Default: SAD-FIT
SAD_CHOL_TOLERANCE¶
SAD guess density decomposition threshold
Type: conv double
Default: 1e-7
SAD_FRAC_OCC¶
Do force an even distribution of occupations across the last partially occupied orbital shell?
Type: boolean
Default: true
SAD_MAXITER¶
Maximum number of atomic SCF iterations within SAD
Type: integer
Default: 50
SAD_PRINT¶
The amount of SAD information to print to the output
Type: integer
Default: 0
SAD_SCF_TYPE¶
SCF type used for atomic calculations in SAD guess
Type: string
Possible Values: DIRECT, DF, MEM_DF, DISK_DF, PK, OUT_OF_CORE, CD, GTFOCK
Default: DF
SAD_SPIN_AVERAGE¶
Do use spin-averaged occupations instead of atomic ground spin state in fractional SAD?
Type: boolean
Default: true
Expert DFT¶
DFT_BLOCK_MAX_POINTS¶
The maximum number of grid points per evaluation block.
Type: integer
Default: 256
DFT_BLOCK_MAX_RADIUS¶
The maximum radius to terminate subdivision of an octree block [au].
Type: double
Default: 3.0
DFT_BLOCK_MIN_POINTS¶
The minimum number of grid points per evaluation block.
Type: integer
Default: 100
DFT_BLOCK_SCHEME¶
The blocking scheme for DFT.
Type: string
Possible Values: NAIVE, OCTREE, ATOMIC
Default: OCTREE
DFT_DENSITY_TOLERANCE¶
density cutoff for LibXC. A negative value turns the feature off and LibXC defaults are used.
Type: conv double
Default: -1.0
DFT_GRAC_C_FUNC¶
The gradient regularized asymptotic correction functional corr form.
Type: string
Default: XC_LDA_C_VWN
DFT_GRAC_X_FUNC¶
The gradient regularized asymptotic correction functional exch form.
Type: string
Default: XC_GGA_X_LB
DFT_GRID_NAME¶
The DFT grid specification, such as SG1.
Type: string
Possible Values: SG0, SG1
Default: No Default
DFT_PRUNING_ALPHA¶
Spread alpha for logarithmic pruning.
Type: double
Default: 1.0
DFT_REMOVE_DISTANT_POINTS¶
Remove points from the quadrature grid that exceed the spatial extend of the basis functions.
Type: boolean
Default: true
DFT_VV10_RHO_CUTOFF¶
Rho cutoff for VV10 NL integration.
Type: double
Default: 1.e-8
DFT_WEIGHTS_TOLERANCE¶
grid weight cutoff. Disable with -1.0.
Type: conv double
Default: 1.0e-15