# CFOUR¶

Interface to CFOUR program written by Stanton and Gauss. Keyword descriptions taken from the CFOUR Website and extended by interface comments.

## Psi4 Control of CFOUR¶

### TRANSLATE_PSI4¶

Do translate set Psi4 options to their cfour counterparts.

Type: booleanDefault: true

## CFOUR Internal¶

### CFOUR_ABCDTYPE¶

Specifies the way the \(\langle ab||cd \rangle\) molecular orbital integrals are handled in post-MP2 calculations. STANDARD (= 0) uses directly the corresponding MO integrals and thus results in an algorithm which in particular for large-scale calculations results in excessive use of disk space (storage of all \(\langle ab||cd\rangle\) integrals. AOBASIS (=2) uses an AO-based algorithm to evaluate all terms involving the \(\langle ab||cd\rangle\) integrals and significantly reduces the amount of disk storage. The use of ABCDTYPE=AOBASIS is strongly recommended for all CC calculations up to CCSD(T) and has been implemented for energy, gradient, second-derivative, and excitation energy calculations.

Type: stringPossible Values: STANDARD, AOBASISDefault: STANDARD

### CFOUR_ACTIVE_ORBI¶

Specifies the active orbitals used in a TCSCF calculation and has to be used in combination with the keyword CFOUR_CORE_ORBITALS The active orbitals are specified by either NIRREP or 2*NIRREP integers specifying the number of active orbitals of each symmetry type, where NIRREP is the number of irreducible representations in the computational point group. If there are no orbitals of a particular symmetry type a zero must be entered. For more information and an example see CFOUR_OCCUPATION .

Type: arrayDefault: No Default

### CFOUR_ANHARMONIC¶

Specifies treatment of anharmonic effects by calculating cubic and/or quartic force fields. VIBROT (=3) requests calculation of only those cubic constants of the form \(\phi_{nij}\), where n is a totally symmetric coordinate. These are sufficient to determine the vibration-rotation interaction constants needed to calculate vibrational corrections to rotational constants, but are

notsufficient to generate the corresponding cubic constants of isotopologs that have a lower point-group symmetry (i.e.HOD isotopolog of water). VPT2 (=1, note that the old value CUBIC can be still used and is equivalent to VPT2) generates all cubic constants and all quartic constants apart from those of the form \(\phi_{ijkl}\), which is enough for: 1) generation of cubic constants of isotopologs (see manual entries associated with anharmonic calculations for an example); 2) calculation of vibrational energy levels with VPT2. This keyword also directs the program to analyze resonances and calculate intensities of one- and two-quantum transitions. FULLQUARTIC (=2) (not part of the public release) is largely self-explanatory; it directs the program to calculate all quartic constants. This is sufficient (but this has not been implemented) to generate the full quartic force field of all isotopologs.

Type: stringPossible Values: CUBIC, VPT2, FULLQUARTIC, VIBROT, OFFDefault: OFF

### CFOUR_ANH_ALGORITHM¶

Specifies which algorithm is used for CFOUR_ANHARMONIC =VIBROT, VPT2, and FULLQUARTIC calculations. If STANDARD (=0) is chosen, then simply invoking

`xcfour`

will cause a complete job to be run with all second-derivative calculations being done in series. If PARALLEL (=1), then the job stops after the second-derivative calculation at the reference geometry and generates out all input geometries for the remaining calculation. These can be then processed in “parallel” (currently not recommended). Note that it is recommended to carry out all calculations with PARALLEL, even when the actual calculation is carried out in a sequential mode.

Type: stringPossible Values: STANDARD, PARALLELDefault: STANDARD

### CFOUR_ANH_DERIVATIVES¶

Specifies whether the anharmonic force field is calculated using analytic gradients (=FIRST) or analytic Hessians (=SECOND).

Type: stringPossible Values: FIRST, SECONDDefault: SECOND

### CFOUR_ANH_STEPSIZE¶

Controls the stepsize used in anharmonic force field calculations. The value is specified in reduced normal coordinates, which are dimensionless. The actual stepsize used in the calculation is \(\times 10^6\) the integer value specified.

Type: integerDefault: 50000

### CFOUR_ANH_SYMMETRY¶

Specifies whether non-abelian symmetry is to be exploited in determining displacements for CFOUR_ANHARMONIC =VIBROT or VPT2 calculations. If set to NONABELIAN (=0), maximum advantage will be taken of symmetry and the full set of cubic force constants will be generated from a skeleton set by application of the totally symmetric projection operator. If set to ABELIAN (=1), only the operations of the abelian subgroup will be exploited. Note: It is important to point out that the symmetrization currently works only for cubic constants. Therefore, if you require quartic force constants (for frequency calculations), you

mustuse the ABELIAN option. Moreover, the latter work for only asymmetric tops and linear molecules.

Type: stringPossible Values: ABELIAN, NONABELIANDefault: ABELIAN

### CFOUR_AO_LADDERS¶

Can be used to control the algorithm used by CFOUR when terms involving \(\langle ab||cd\rangle\) molecular orbital integrals are calculated in the atomic orbital basis (see CFOUR_ABCDTYPE . MULTIPASS (= 0) uses an approach where the AO integral file is read a number of times in order to ensure maximal vectorization and is usually the optimal strategy on supercomputers; SINGLEPASS (= 1) determines the contributions with only a single pass through the AO integrals, but at the cost of significantly reduced vectorization. In general, however, SINGLEPASS is definitely preferable on workstations with RISC architectures. (Default : MULTIPASS on all 64-bit machines (e.g., CRAY-YMP) ; SINGLEPASS on all 32-bit machines (e.g., IBM-RS6000, HP-735, SGI-Indigo, DEC alphastations)). SPARSE_AO (=2) uses a sparse matrix algorithm which first rearranges the integral matrix in order to get “well-occupied” and “very sparse” blocks. “Well-occupied” blocks will be multiplied by matrix multiplication while in “very sparse” blocks only the non-zero elements are considered. The computational time is further reduced using symmetrized and anti-symmetrized integral and amplitude matrices in the multiplication. Substantial saving is assumed if SPARSE_AO (=2) is used.

Type: stringPossible Values: MULTIPASS, SINGLEPASSDefault: SINGLEPASS

### CFOUR_AV_SCF¶

Experimental Use! ON (=1) requests and averaged SCF over two states. So far only implemented for degenerate doublet-Pi states and used in conjunction with SOPERT.

Type: booleanDefault: false

### CFOUR_BASIS¶

Specifies the AO basis used in the calculation. One can either specify a basis known to CFOUR or via BASIS=SPECIAL (=0) requests an arbitrary basis (see non-standard basis-set input). However, the latter must be available in the supplied GENBAS file. As standard basis sets, currently the following are available.

Psi4 Interface:Recommended to use instead BASIS for larger basis set selection and greater flexibility. When BASIS used, CFOUR_SPHERICAL is set appropriately.

Type: stringPossible Values: STO-3G, 3-21G, 4-31G, 6-31G, 6-31G*, 6-31G**, 6-311G, 6-311G*, 6-311G**, DZ, DZP, TZ, TZP, TZ2P, PVDZ, PVTZ, PVQZ, PV5Z, PV6Z, PCVDZ, PCVTZ, PCVQZ, PCV5Z, PCV6Z, AUG-PVDZ, AUG-PVTZ, AUG-PVTZ, AUG-PVQZ, AUG-PV5Z, AUG-PV6Z, D-AUG-PVDZ, D-AUG-PVTZ, D-AUG-PVQZ, D-AUG-PV5Z, D-AUG-PV6Z, cc-pVDZ, cc-pVTZ, cc-pVQZ, cc-pV5Z, cc-pV6Z, cc-pCVDZ, cc-pCVTZ, cc-pCVQZ, cc-pCV5Z, cc-pCV6Z, PWCVDZ, PWCVTZ, PWCVQZ, PWCV5Z, PWCV6Z, PwCVDZ, PwCVTZ, PwCVQZ, PwCV5Z, PwCV6Z, svp, dzp, tzp, tzp2p, qz2p, pz3d2f, 13s9p4d3f, WMR, ANO0, ANO1, ANO2, EVEN_TEMPERED, SPECIALDefault: SPECIAL

### CFOUR_BRUCK_CONV¶

experimental use

Type: integerDefault: 4

### CFOUR_BRUECKNER¶

Specifies whether Brueckner orbitals are to be determined for the specified CC method. OFF(=0) Brueckner orbitals are not to be determined, ON (=1) they are to be determined.

Type: booleanDefault: false

### CFOUR_CACHE_RECS¶

The number of records held in the i/o cache used by the post-SCF programs. The maximum number of records which can be held is 100.

Type: integerDefault: 10

### CFOUR_CALC_LEVEL¶

Defines the level of calculation to be performed.

Psi4 Interface:Keyword set from argument of computation command: CCSD if`energy('c4-ccsd')`

,etc.See Energy (CFOUR) and Gradient (CFOUR). for all available.

Type: stringPossible Values: SCF, HF, MBPT(2), MP2, MBPT(3), MP3, SDQ-MBPT(4), SDQ-MP4, MBPT(4), MP4, CCD, CCSD, CCSD(T), CCSDT-1, CCSDT-1b, CCSDT-2, CCSDT-3, CCSDT-4, CCSDT, CC2, CC3, QCISD, QCISD(T), CID, CISD, UCC(4), B-CCDDefault: SCF

### CFOUR_CC_CONV¶

Specifies the convergence criterion for the CC amplitude equations. The amplitudes are considered to be converged when the maximum of all (absolute) changes in the amplitudes is less than \(10^N\), where \(N\) is the value associated with the keyword.

Type: integerDefault: 7

### CFOUR_CC_EXPORDER¶

Specifies the maximum number of expansion vectors used in the iterative subspace to enhance convergence in the solution of the CC equations.

Type: integerDefault: 5

### CFOUR_CC_EXTRAPOLATION¶

Specifies the type of convergence acceleration used to solve the CC equations. RLE (=0) uses the RLE methods of Purvis and Bartlett, DIIS (=1) uses the DIIS approach by Pulay, NOJACOBI (=2) uses RLE with continuous extrapolation, OFF (=3) uses no convergence acceleration. In general, DIIS provides the best results and is recommended, while OFF often results in poor convergence and thus cannot be recommended.

Type: stringPossible Values: RLE, DIIS, NOJACOBI, OFFDefault: DIIS

### CFOUR_CC_MAXCYC¶

Specifies the maximum number of iterations in solving the CC amplitude equations.

Type: integerDefault: 50

### CFOUR_CC_PROGRAM¶

Specifies which CC program is used. The available options are VCC (=0), ECC (=1), MRCC (=2), and EXTERNAL (=3). The default for all calculations is currently VCC which requests usage of

`xvcc`

, but in many cases (e.g., for CCSD and CCSD(T)) ECC should be preferred due to the better performance of`xecc`

(available currently for CCSD, CCSD+T, CCSD(T), and closed-shell CCSDT-n, CC3, and CCSDT). MRCC and External are intended for CC programs outside the CFOUR suite, e.g., the general CC module mrcc written by M. Kallay (Budapest, Hungary). Default: VCC Note: Using the option ECC is not recommended for ROHF gradients. That is, if you are doing a geometry optimization with ROHF as your reference wave function then it is safe to use the option VCC.Psi4 Interface:Keyword set according to best practice for the computational method CFOUR_CALC_LEVEL reference CFOUR_REFERENCE (NYI) and derivative level CFOUR_DERIV_LEVEL according to Table Best Practices when method specified by argument to computation command (e.g., when`energy('c4-ccsd')`

requested but not when`energy('cfour')`

requested). Value can always be set explicitly.

Type: stringPossible Values: VCC, ECC, NCC, MRCC, EXTERNALDefault: VCC

### CFOUR_CHARGE¶

Specifies the molecular charge.

Psi4 Interface:Keyword set from active molecule.

Type: integerDefault: 0

### CFOUR_CIS_CONV¶

Specifies the convergence threshold as \(10^{-N}\) for CIS calculations.

Type: integerDefault: 5

### CFOUR_CONTINUUM¶

Signifies that one or more “continuum” orbitals should be added to the calculation. VIRTUAL and DVIRTUAL specify one or two orbital which should be initially unoccupied (in the SCF calculation), while OCCUPIED and DOCCUPIED specify one or two orbitals which should be initially occupied.

Type: stringPossible Values: NONE, VIRTUAL, DVIRTUAL, OCCUPIED, DOCCUPIEDDefault: NONE

### CFOUR_CONTRACTION¶

Specifies the contraction scheme used by the integral and integral derivative program. SEGMENTED (=0) uses a segmented contraction scheme; GENERAL (=1) uses a general contraction scheme, and UNCONTRACTED (=2) uses the corresponding uncontracted sets. Note that even for truly segmented basis sets, the integral programs run significantly faster in the GENERAL mode.

Type: stringPossible Values: SEGMENTED, GENERAL, UNCONTRACTEDDefault: GENERAL

### CFOUR_CONVERGENCE¶

Identical to CFOUR_GEO_CONV

Type: integerDefault: 4

### CFOUR_COORDINATES¶

Specifies the type of coordinates used in the input file ZMAT. Value INTERNAL (=0) means that the geometry is supplied in the usual Z-matrix format, while CARTESIAN (=1) means that the geometry is given in Cartesian coordinates. A third option is XYZINT (=2) for which a Z-matrix connectivity is defined, but with values of the internal coordinates defined implicitly by supplying Cartesian coordinates. Note that geometry optimizations are currently only possible for INTERNAL and XYZ2INT.

Psi4 Interface:Keyword set from active molecule, always CARTESIAN. Above restrictions on geometry optimizations no longer apply.

Type: stringPossible Values: INTERNAL, CARTESIAN, XYZINTDefault: INTERNAL

### CFOUR_CORE_ORBITALS¶

Specifies the core orbitals used in a TCSCF calculation and has to be used in combination with the keyword CFOUR_ACTIVE_ORBI The core orbitals are specified by either NIRREP or 2*NIRREP integers specifying the number of core orbitals of each symmetry type, where NIRREP is the number of irreducible representations in the computational point group. If there are no orbitals of a particular symmetry type a zero must be entered. For more information and an example see CFOUR_OCCUPATION

Type: arrayDefault: No Default

### CFOUR_CPHF_CONVER¶

Specifies the convergence criterion for the iterative solution of the CPHF and Z-vector equations. The solutions are considered to be converged when the residual norm of the error vector falls below \(10^N\).

Type: integerDefault: 12

### CFOUR_CPHF_MAXCYC¶

Specifies the maximum number of cycles allowed for the solution of the CPHF- and/or Z-vector equations.

Type: integerDefault: 64

### CFOUR_CURVILINEAR¶

Specifies whether or not Hessian matrix is transformed (nonlinearly) to curvilinear internal coordinates. A value of 0 (or OFF) turns the transformation off if the analytic force constants are not available, while it is always performed if CURVILINEAR=1 (or ON). Values higher than 1 (or NO) unconditionally turn the transformation off.(Default: ON if analytic Hessian is available, OFF otherwise).

Type: booleanDefault: true

### CFOUR_DBOC¶

Specifies whether the diagonal Born-Oppenheimer correction (DBOC) to the energy is evaluated (ON =1) or not (OFF =0). DBOC calculations are currently only available for HF-SCF and CCSD using RHF or UHF reference functions.

Type: booleanDefault: false

### CFOUR_DCT¶

Specifies whether the Dipole Coupling Tensor (DCT) is calculated (ON =1) or not (OFF =0).

Type: booleanDefault: false

### CFOUR_DERIV_LEVEL¶

Specifies whether or not energy derivatives are to be calculated and if so whether first or second derivatives are computed. ZERO (= 0) derivatives are not calculated, FIRST (=1) first derivatives are calculated, SECOND (=2) second derivatives are calculated. Note that this keyword usually needs not be set in any calculation since it is automatically set if the appropriate other options in the CFOUR namelist are turned on.

Psi4 Interface:Keyword set from type of computation command: ZERO if`energy()`

, FIRST if`gradient()`

or`optimization()`

,etc.

Type: stringPossible Values: ZERO, FIRST, SECONDDefault: ZERO

### CFOUR_DIFF_TYPE¶

Specifies whether orbital-relaxed (RELAXED =0) or orbital-unrelaxed (UNRELAXED =1) derivatives are computed in the CC calculation.

Type: stringPossible Values: RELAXED, UNRELAXEDDefault: RELAXED

### CFOUR_DROPMO¶

Specifies which molecular orbitals will be dropped from the post-SCF calculation. The orbitals are numbered in ascending order from the most stable (negative energy) to the most unstable (largest positive energy). Individual orbitals must be separated with a dash, while x>y means orbitals x through y inclusive. For example, the string

`1>10-55-58>64`

, would result in orbitals 1,2,3,4,5,6,7,8,9,10,55,58,59,60,61,62,63 and 64 being dropped. For UHF calculations, the appropriate orbitals are deleted for both spin cases. No dropped virtual MOs are currently allowed for gradient or property calculations.Psi4 Interface:The array above is specified in PSI as (white space tolerant) [1,2,3,4,5,6,7,8,9,10,55,58,59,60,61,62,63,64].

Type: arrayDefault: No Default

### CFOUR_ECP¶

Specifies whether effective core potentials (pseudopotentials) are used (ON, =1) or not (OFF, =0).

Type: booleanDefault: false

### CFOUR_EIGENVECTOR¶

Specifies which eigenvector of the totally symmetric part of the block-factored Hessian is to be followed uphill in a transition state search. Eigenvectors are indexed by their eigenvalues – the lowest eigenvalue is 1, the next lowest is 2, etc. The default is 1, which should always be used if you are not looking for a specific transition state which you know corresponds to motion along a different mode. In the future, relatively sophisticated generation of a guessed eigenvector will be implemented, but this is the way things are for now. Of course, this keyword has no meaning if CFOUR_METHOD is not set to TS.

Type: integerDefault: 1

### CFOUR_EL_ANHARM¶

Experimental use, ON = 1 requests the evaluation of electrical anharmonicities

Type: booleanDefault: false

### CFOUR_EOM_NONIT¶

Controls whether non-iterative triples corrections are applied after various types of EOM-CCSD calculation. Works with CFOUR_EXCITE set to EOMIP, might work with EOMEE, certainly doesn’t work with EOMEA. Use with great caution, preferably after having a few drinks.

Type: booleanDefault: false

### CFOUR_ESTATE_CONV¶

Specifies the threshold used in converging CC-LR/EOM-CC calculations. The iterative diagonalization is continued until the RMS residual falls below \(10^{-N}\) with \(N\) as the value specified with this keyword.

Type: integerDefault: 5

### CFOUR_ESTATE_MAXCYC¶

The maximum number of expansion vectors used in the solution of EOMCC equations (Default: 20, hard-coded to 4 in triples calculations)

Type: integerDefault: 20

### CFOUR_ESTATE_PROP¶

This keyword applies only to EOM-CC calculations and specifies whether any excited or ionized state one-electron properties are to be calculated. Proper use of this keyword requires a relatively advanced knowledge of quantum chemistry and the available options are discussed here. The options are: OFF (=0) [no properties or transition moments are calculated]; EXPECTATION (=1) [transition moments and dipole strengths are calculated along with selected one-electron properties which are evaluated as expectation values]; UNRELAXED (=2) [selected one-electron properties are calculated in an approximation that neglects relaxation of molecular orbitals]; RESPONSE (=3) [selected one-electron properties are calculated as analytic first derivatives of the energy]. Except for EOMCC calculations on two-electron systems (which are exact), properties obtained by the three approaches will not be equivalent. The default value for this keyword is slightly complicated. For TDA calculations, the default is EXPECTATION since the evaluation of transition moments involves only a negligible amount of additional computation relative to the evaluation of the excitation energies. For EOMCC, the default is OFF since evaluation of any transition moments or properties requires approximately twice the computational time. Transition moments and dipole strengths are evaluated by default for all values of ESTATE_PROP other than OFF.

Type: stringPossible Values: OFF, EXPECTATION, UNRELAXED, RESPONSEDefault: No Default

### CFOUR_ESTATE_SYM¶

Specifies the number of excited states which are to be determined in each irreducible representation of the computational subgroup. The program attempts to find all of the lowest roots, but this is not guaranteed because the eigenvalue problem is not solved by direct matrix diagonalization, but rather by an iterative (modified Davidson) algorithm. For excited state gradient calculations, only one root (clearly) is used. In such a case, one and only one non-zero entry in the string can be used, and this value is usually set to one (

i.e.0/1/0/0). (However sometimes one wants to calculate the gradient for, say, the second root of a given symmetry, and in such a case, one could use 0/2/0/0. What happens is that both roots are calculated, but only the second one is used in the subsequent density matrix and gradient calculation.) The format used for this keyword is identical to that used in CFOUR_OCCUPATION For example, for a computational subgroup having four symmetry species, the string 3/1/0/2 specifies that 6 total roots should be searched for, three in the first block, one in the second block, and two in the fourth block. It is also important to note that the`%excite*`

input, if present, takes precedence over this keyword. Default: All zeros.Psi4 Interface:The array above is specified in PSI as (white space tolerant) [3,1,0,2].

Type: arrayDefault: No Default

### CFOUR_ESTATE_TRANS¶

Specifies whether just the excitation energies (OFF, =0) or in addition transition moments (EXPECTATION, =1) are calculated. Note that this keyword should not be used in excited-state calculations involving analytic gradients and that transition moments are essentially only available for EOM-CCSD/CCSD-LR.

Type: stringPossible Values: OFF, EXPECTATIONDefault: OFF

### CFOUR_EVAL_HESS¶

Tells the program, in the course of a geometry optimization, to calculate the Hessian explicitly every N cycles. 0 means never calculated explicitly.

Psi4 Interface:Geometry optimizations run through PSI (except in sandwich mode) use PSI’s optimizer and so this keyword has no effect. Use optking keywords instead, particularly FULL_HESS_EVERY

Type: integerDefault: 0

### CFOUR_EXCITATION¶

Specifies in CC calculations using mrcc the excitation level if the calculation level has been chosen as CC(n), CI(n), or CCn(n).

Type: integerDefault: 0

### CFOUR_EXCITE¶

Specifies the type of EOM-CC/LR-CC treatment to be performed. Available options are NONE (=0), EOMEE (=3, the EOM-CC/CC-LR approach for the treatment of excited states), EOMIP (=4, the EOM-CC/CC-LR approach for the treatment of ionized states), EOMEA (=7, the EOM-CC/CC-LR approach for the treatment of electron-attached states).

Type: stringPossible Values: NONE, EOMEE, EOMIP, EOMEADefault: NONE

### CFOUR_FC_FIELD¶

Specifies the strength of a Fermi-Contact perturbation as required for finite-field calculations of spin densities and the FC contributions to indirect spin-spin coupling constants. The value must be specified as an integer and the FC strength used by the program will be the value of the keyword \(\times 10^{-6}\). The atom for which the FC perturbation is switched on is specified in the ZMAT file after the CFOUR command line and potential basis set input, as follows %spin density N with N as the number of atom (in (X5,I3) format) in the order they are written by JODA to the MOL file. Be aware that for some atoms, the calculation has to be run in lower symmetry or even without symmetry. (Default : 0)

Type: integerDefault: 0

### CFOUR_FD_CALCTYPE¶

Specifies the algorithm used to compute the harmonic force constants in finite-difference calculations.GRADONLY (=0) evaluates the force constants and dipole moment derivatives by numerical differentiation of analytic gradients; ENERONLY (=1) evaluates the force constants by second differences of energies (dipole moment derivatives are not evaluated); while MIXED (=2) evaluates 1x1 blocks of symmetry-blocked force constants by second differences pf energies and all other elements by first differences of gradients. the GRADONLY and MIXED approaches may, of course, only be used hwen using computational methods for which analytic gradients are available.

Type: stringPossible Values: GRADONLY, ENERONLY, MIXEDDefault: GRADONLY

### CFOUR_FD_IRREPS¶

Requests that only vibrational frequencies of certain symmetry types are evaluated in a VIBRATION=FINDIF calculation. The numbers of the irreducible representations for which vibrational analysis is to be performed are separated by slashes. For example, FD_IRREP=1/3/4 means compute the frequencies of modes transforming as the first, third, and fourth irreducible representations. If a symmetry is specified for which there are no vibrational modes, the program will terminate. The labels of the irreducible representations for this keyword are not usually the same as those used in the rest of the calculation. Moreover, for some point groups, for example, those of linear molecules, the two sets of labels refer to different subgroups. There is as yet no straightforward way to determine what they will be without starting a calculation. If one runs the

`xjoda`

and then the`xsymcor`

executables, the relevant irreducible representations will be listed. If all vibrational frequencies are desired, this keyword need not be included. Default : compute vibrational frequencies for all irreducible representations

Type: arrayDefault: No Default

### CFOUR_FD_PROJECT¶

Specifies whether or not rotational degrees of freedoms are projected out from the symmetry-adapted coordinates in a finite difference calculations. ON (=0) uses rotationally projected coordinates, while OFF (=1) retains the rotational degrees of freedom. At a stationary point on the potential energy surface, both options will give equivalent harmonic force fields, but OFF should be used at non-stationary points.

Type: stringPossible Values: ON, OFFDefault: ON

### CFOUR_FD_STEPSIZE¶

Specifies the step length in mass-weighted coordinates (in \(10^{-4} amu^{1/2} bohr\) ) used in generating the force constant matrix by finite difference of Cartesian gradients.

Type: integerDefault: 5

### CFOUR_FD_USEGROUP¶

In finite difference calculations using the FINDIF option, this keyword specifies the point group to be used in generating the symmetry-adapted vibrational coordinates. FULL (= 0) specifies the full molecular point group, COMP (= 1) specifies the Abelian subgroup used in the electronic structure calculation.

Type: stringPossible Values: FULL, COMPDefault: FULL

### CFOUR_FILE_RECSIZ¶

This specifies the physical length (in integer words) of the records used in the word-addressable direct access files used by CFOUR. This value should always be chosen as a multiple of 512 bytes, as your local system manager certainly understands.

Type: integerDefault: 2048

### CFOUR_FILE_STRIPE¶

This option allows the splitting of files. Input is required in the form N1/N2/N3/N4/N5, where N1, N2, N3, N4, and N5 specify the number of files in which

`MOINTS`

,`GAMLAM`

,`MOABCD`

,`DERINT`

, and`DERGAM`

are split, respectively.

Type: stringDefault: 0/0/0/0/0

### CFOUR_FINITE_PERTURBATION¶

Specifies the field strength for a perturbation (defined within a

`%perturbation`

section). The value must be given as an integer, and the field strength used by the program will be then the value of the keyword \(\times 10^{-6}\).

Type: integerDefault: 0

### CFOUR_FOCK¶

This option is used to control the algorithm used for construction of the Fock matrix in SCF calculations. PK (=0) uses the PK-supermatrix approach while AO (=1) constructs the matrix directly from the basis function integrals. In general, PK is somewhat faster, but results in considerable use of disk space when out-of-core algorithms are required. (Default: FOCK).

Type: stringPossible Values: PK, AODefault: No Default

### CFOUR_FREQ_ALGORITHM¶

FREQ_ALGORIT experimental use

Type: stringPossible Values: STANDARD, PARALLELDefault: STANDARD

### CFOUR_FROZEN_CORE¶

Specifies whether in the correlation treatment all electron (OFF =0) or only the valence electrons (ON =1) are considered. This keyword provides an alternative to the CFOUR_DROPMO keyword, as it allows frozen-core calculation without explicitly specifying the corresponding inner-shell orbitals.

Type: booleanDefault: false

### CFOUR_FROZEN_VIRT¶

Specifies whether in the correlation treatment all virtual orbitals (OFF =0) or only a subset of virtual orbitals (ON =1) are used. In the latter case, the threshold for deleting virtual orbitals based on the orbital energy needs to be specified in a

`%frozen_virt`

section.

Type: booleanDefault: false

### CFOUR_GAMMA_ABCD¶

Used to control the handling and storage of two-particle density matrix elements with four virtual indices \(\Gamma(abcd)\). DISK (=0) directs the program to calculate and store all elements of \(\Gamma(abcd)\), while DIRECT (=1) tells the program to use alternative algorithms in which \(\Gamma(abcd)\) is calculated and used “on the fly”. Note that this option might be not available for all type of calculations.

Type: stringPossible Values: DISK, DIRECTDefault: DISK

### CFOUR_GENBAS_1¶

This keyword applies only to Hydrogen and Helium atoms and specifies the number of contracted Gaussian functions per shell. There is usually no need to use this keyword, but it can be useful for using a subset of the functions in a particular entry in the

`GENBAS`

file, particularly for generally contracted WMR basis sets. For example, if entry H:BASIS in the`GENBAS`

file contains 7 contracted s functions, 4 p functions and a single d function, then setting GENBAS_1=730 would eliminate the last p function and the d function. Default: use the unaltered`GENBAS`

entry.

Type: stringDefault: No Default

### CFOUR_GENBAS_2¶

This keyword performs the same function as CFOUR_GENBAS_1 above, but applies to second-row atoms.

Type: stringDefault: No Default

### CFOUR_GENBAS_3¶

This keyword performs the same function as CFOUR_GENBAS_1 and CFOUR_GENBAS_2 , but applies to third-row atoms.

Type: stringDefault: No Default

### CFOUR_GENBAS_4¶

This keyword performs the same function as CFOUR_GENBAS_1 , CFOUR_GENBAS_2 , and CFOUR_GENBAS_3 , but applies to fourth-row atoms.

Type: stringDefault: No Default

### CFOUR_GEO_CONV¶

Specifies the convergence criterion for geometry optimization. The optimization terminates when the RMS gradient is below \(10^{-N}\) Hartree/bohr, where \(N\) is the specified value.

Psi4 Interface:Geometry optimizations run through PSI (except in sandwich mode) use PSI’s optimizer and so this keyword has no effect. Use optking keywords instead, particularly G_CONVERGENCE =CFOUR, which should be equivalent except for different internal coordinate definitions.

Type: integerDefault: 5

### CFOUR_GEO_MAXCYC¶

Specifies the maximum allowed number of geometry optimization cycles.

Psi4 Interface:Geometry optimizations run through PSI (except in sandwich mode) use PSI’s optimizer and so this keyword has no effect. Use optking keywords instead, particularly GEOM_MAXITER

Type: integerDefault: 50

### CFOUR_GEO_MAXSTEP¶

Specifies largest step (in millibohr) which is allowed in geometry optimizations.

Psi4 Interface:Geometry optimizations run through PSI (except in sandwich mode) use PSI’s optimizer and so this keyword has no effect. Use optking keywords instead, particularly INTRAFRAG_STEP_LIMIT

Type: integerDefault: 300

### CFOUR_GEO_METHOD¶

Specifies the used geometry optimization methods. The following values are permitted: NR (=0) — straightforward Newton-Raphson search for minimum; RFA (=1) — Rational Function Approximation search for minimum (this method can be used to find minima when the initial structure is in a region where the Hessian index is nonzero); TS (=2) Cerjan-Miller eigenvector following search for a transition state (can be started in a region where the Hessian index is not equal to unity); MANR (=3) — Morse-adjusted Newton-Raphson search for minimum (very efficient minimization scheme, particularly if the Hessian is available); SINGLE_POINT (=5) for a single-point energy calculation. ENERONLY (=6) requests a geometry optimization based on single-point energy calculations. Default: SINGLE-POINT (NR as soon as variables are marked to be optimized).

Type: stringPossible Values: NR, RFA, TS, MANR, SINGLE_POINT, ENERONLYDefault: SINGLE_POINT

### CFOUR_GIAO¶

Specifies whether gauge-including atomic orbitals are used (ON) or not (OFF). Default: ON for CFOUR_PROPS =NMR and =MAGNETIC, otherwise OFF

Type: stringPossible Values: ON, OFFDefault: No Default

### CFOUR_GRID¶

Keyword used to control type of grid calculation (see later section in this manual). Options are OFF (=0), no grid calculation; CARTESIAN (=1), steps are in Cartesian coordinates (which must be run with CFOUR_COORDINATES =CARTESIAN); INTERNAL (=2), steps are in Z-matrix internal coordinates; QUADRATURE (=3) steps are chosen for an integration based on Gauss-Hermite quadrature. (Default: OFF)

Type: stringPossible Values: OFF, CARTESIAN, INTERNAL, QUADRATUREDefault: OFF

### CFOUR_GUESS¶

Where the initial SCF eigenvectors are read from. MOREAD means to read from the disk (the

`JOBARC`

file) and CORE means to use a core Hamiltonian initial guess. If MOREAD is chosen but no disk file is present, the core Hamiltonian is used. (Default: MOREAD)

Type: stringPossible Values: MOREAD, COREDefault: MOREAD

### CFOUR_HBAR¶

This keyword determines which action is taken by the linear response program. ON (=1) the full effective Hamiltonian is calculated and written to disk; OFF (=0) the “lambda” linear response equations are solved.

Type: booleanDefault: false

### CFOUR_HFSTABILITY¶

Control analysis of the stability of RHF, ROHF and UHF wavefunctions, as well as a possible search for a lower SCF solution. There are three possible options for this keyword. OFF (=0) does nothing, while ON (=1) performs a stability analysis and returns the number of negative eigenvalues in the orbital rotation Hessian. A third option, FOLLOW (=2) performs the stability analysis and then proceeds to rotate the SCF orbitals in the direction of a particular negative eigenvalue of the orbital rotation Hessian (see the explanation of keyword CFOUR_ROT_EVEC , after which the SCF is rerun.

Type: stringPossible Values: OFF, ON, FOLLOWDefault: OFF

### CFOUR_INCORE¶

This keyword can be used to significantly reduce disk i/o, and should be implemented very soon. The following options are available: OFF (= 0), no special algorithms are used (the default case); ALL (=1) all quantities except the \(\langle ab\vert\vert cd\rangle\) molecular integral lists are held in core; PARTIAL (= 2), the T2 and T1 vectors are held in core throughout the calculation; (=4) all quantities except the \(\langle ab\vert\vert cd\rangle\) and \(\langle ab\vert\vert ci\rangle\) integrals are held in core; (=5) \(\langle ij\vert\vert kl\rangle\) and \(\langle ij\vert\vert ka\rangle\) and two-index quantities are held in core; (=6) all direct access files (

`MOINTS`

,`GAMLAM`

, etc.) are held in core. At present, these options have been implemented only in the energy code`xvcc`

and the excitation energy code`xvee`

. (Default: 0)

Type: stringPossible Values: OFF, ALL, PARTIALDefault: OFF

### CFOUR_INPUT_MRCC¶

Specifies whether an input for mrcc is written (ON, =0) or not (OFF, =1) if CFOUR_CC_PROGRAM =EXTERNAL has been specified.

Type: booleanDefault: true

### CFOUR_INTEGRALS¶

This keyword defines what type of integral input will be written by

`xjoda`

. VMOL (=1) has to be used with the programs of CFOUR. Using ARGOS (=0), input for Pitzer’s ARGOS integral program will be written. (Default: VMOL).

Type: stringPossible Values: VMOL, ARGOSDefault: VMOL

### CFOUR_JODA_PRINT¶

Controls amount of debug printing performed by

`xjoda`

. The higher the number, the more information is printed. Values of 25 or higher generally do not produce anything of interest to the general user. Do not set JODA_PRINT to 999 as this will cause the core vector to be dumped to disk.

Type: integerDefault: 0

### CFOUR_LINEQ_CONV¶

Convergence threshold for linear equations controlled by LINEQ_TYPE. Equations are iterated until smallest residual falls below \(10^{-N}\), where \(N\) is the value associated with this keyword.

Type: integerDefault: 7

### CFOUR_LINEQ_MAXCY¶

The maximum number of iterations in all linear CC equations.

Type: integerDefault: 50

### CFOUR_LINEQ_TYPE¶

Determines the algorithm used to solve linear equations ( \(\Lambda\) and derivative \(T\) and \(\Lambda\) ). POPLE (=0) uses Pople’s method of successively orthogonalized basis vectors, while DIIS (=1) uses Pulay’s DIIS method. The latter offers the practical advantage of requiring much less disk space, although it is not guaranteed to converge. Moreover, POPLE has not been tested for some time and should definitely be checked! (Default : DIIS)

Type: stringPossible Values: POPLE, DIISDefault: DIIS

### CFOUR_LOCK_ORBOCC¶

This keyword is used by the SCF program to determine if the orbital occupancy (by symmetry block) is allowed to change in the course of the calculation. ON (=1) locks the occupation to that set by the keyword CFOUR_OCCUPATION (or the initial guess if omitted); OFF (= 0) permits the occupation to change. (Default : 1 if the occupation is specified with CFOUR_OCCUPATION and for second and later steps of optimizations; 0 if CFOUR_OCCUPATION omitted.)

Type: booleanDefault: false

### CFOUR_MAXSTEP¶

Identical to CFOUR_GEO_MAXSTEP

Type: integerDefault: 300

### CFOUR_MEMORY_SIZE¶

Specifies the amount of core memory used in integer words (default) or in the units specified via the keyword CFOUR_MEM_UNIT Default: 100 000 000 (approximately 381 or 762 MB for 32 or 64 bit machines, respectively).

Psi4 Interface:Keyword set in MB from memory input command when given.

Type: integerDefault: 100000000

### CFOUR_MEM_UNIT¶

Specifies the units in which the amount of requested core memory is given. Possible choices are INTEGERWORDS (default), kB, MB, GB, and TB.

Psi4 Interface:Keyword set from memory input command when given, always MB.

Type: stringPossible Values: INTEGERWORDS, KB, MB, GB, TBDefault: INTEGERWORDS

### CFOUR_METHOD¶

Specifies the geometry optimization strategy. Four values are permitted: NR (=0) – Straightforward Newton-Raphson search for minimum; RFA (=1) – Rational Function Approximation search for minimum (this method can be used to find minima when the initial structure is in a region where the Hessian index is nonzero); TS (=2) Cerjan-Miller eigenvector following search for a transition state (can be started in a region where the Hessian index is not equal to unity); MANR (=3) – Morse-adjusted Newton-Raphson search for minimum (very efficient minimization scheme, particularly if the Hessian is available); 4 is currently unavailable; SINGLE_POINT (=5) is a single point calculation.

Psi4 Interface:Geometry optimizations run through PSI (except in sandwich mode) use PSI’s optimizer and so this keyword has no effect. Use optking keywords instead, particularly OPT_TYPE and STEP_TYPE

Type: stringPossible Values: NR, RFA, TS, MANR, SINGLE_POINTDefault: SINGLE_POINT

### CFOUR_MRCC¶

Specifies the type of MRCC calculation. MK performs a MR-CC calculation based on Mukherjee’s ansatz.

Type: booleanDefault: false

### CFOUR_MULTIPLICITY¶

Specifies the spin multiplicity.

Psi4 Interface:Keyword set from active molecule.

Type: integerDefault: 1

### CFOUR_NACOUPLING¶

Calculation of non-adiabatic coupling. In case of ON (=1) the method by Ichino, Gauss, Stanton is used to obtain the lambda coupling, while in case of LVC (=3) the lambda coupling is computed by means of the algorithm by Tajti and Szalay. Furthermore, NACV (=2) requests the computation of the full non-adiabatic coupling. Note that for calculations using LVC or NACV options the multiroot diagonalization has to be used, as requested via the keyword CFOUR_EOM_NSTATES (dne?) =MULTIROOT.

Type: stringPossible Values: ON, NACV, LVCDefault: OFF

### CFOUR_NEGEVAL¶

Specifies what to do if negative eigenvalues are encountered in the totally symmetric Hessian during an NR or MANR geometry-optimization search. If ABORT (=0), the job will terminate with an error message; if SWITCH (=1) the program will just switch the eigenvalue to its absolute value and keep plugging away (this is strongly discouraged!); and if RFA (=2), the keyword CFOUR_GEO_METHOD is switched to RFA internally and the optimization is continued.

Psi4 Interface:Geometry optimizations run through PSI (except in sandwich mode) use PSI’s optimizer and so this keyword has no effect. Use optking keywords instead.

Type: stringPossible Values: ABORT, SWITCH, RFADefault: ABORT

### CFOUR_NEWNORM¶

All components of spherical AO’s are normalized to 1. This feature can help with numerical convergence issues if AO integrals are involved. Currently only working for single-point energy calculations.

Type: booleanDefault: false

### CFOUR_NONHF¶

Specifies whether the reference function used in the correlation energy calculation satisfies the (spin-orbital) HF equations or not. Usually there is no need to set this parameter (OFF = 0 and ON =1), since standard non-HF reference functions (QRHF and ROHF) set this flag automatically.

Type: booleanDefault: false

### CFOUR_NTOP_TAMP¶

Specifies how many t amplitudes will be printed for each spin case and excitation level. For =N, The largest N amplitudes for each spin case and excitation level will be printed.

Type: integerDefault: 15

### CFOUR_OCCUPATION¶

Specifies the orbital occupancy of the reference function in terms of the occupation numbers of the orbitals and their irreducible representations. The occupancy is specified by either NIRREP or 2*NIRREP integers specifying the number of occupied orbitals of each symmetry type, where NIRREP is the number of irreducible representations in the computational point group. If there are no orbitals of a particular symmetry type a zero must be entered. If the reference function is for an open-shell system, two strings of NIRREP occupation numbers separated by a slash are input for the \(\alpha\) and \(\beta\) sets of orbitals. An example of the use of the OCCUPATION keyword for the water molecule would be OCCUPATION=3-1-1-0. For the \(^2A_1\) water cation, an open-shell system, the keyword would be specified by OCCUPATION=3-1-1-0/2-1-1-0. It should be noted that the

`xvmol`

integral program orders the irreducible representations in a strange way, which most users do not perceive to be a logical order. Hence, it is usually advisable initially to run just a single point integral and HF-SCF calculation in order to determine the number and ordering of the irreducible representations. The occupation keyword may be omitted, in which case an initial orbital occupancy is determined by diagonalization of the core Hamiltonian. In many cases, HF-SCF calculations run with the core Hamiltonian guess will usually converge to the lowest energy HF-SCF solution, but this should not be blindly assumed. (Default: The occupation is given by the core Hamiltonian initial guess).Psi4 Interface:The arrays above are specified in PSI as (white space tolerant) [3,1,1,0] and [[3,1,1,0],[3,0,1,0]].

Type: arrayDefault: No Default

### CFOUR_OPEN-SHELL¶

Specifies which kind of open-shell CC treatment is employed. The default is a spin-orbital CC treatment (SPIN-ORBITAL =1) which is the only possible choice for UHF-CC schemes anyways. For ROHF-CC treatments, the possible options are beside the standard spin-orbital scheme a spin-restricted CC approach (SR-CC=3), as well as a corresponding linear approximation (which in the literature usually is referred to as partially-spin-adapted CC scheme) (PSA-CC=1). SR-CC and PSA-CC are within the CCSD approximation restricted to excitations defined by the first-order interacting space arguments. With the keywords PSA-CC_FULL (=2) or SR-CC_FULL (=6) inclusion of the so called “pseudo-triples” beyond the first-order interacting space is also possible. The two-determinant CC method for open-shell singlet states can be activated by TD-CC (=8).

Type: stringPossible Values: SPIN-ORBITAL, SR-CC, PSA-CC_FULL, SR-CC_FULL, TD-CCDefault: SPIN-ORBITAL

### CFOUR_OPT_MAXCYC¶

Identical to CFOUR_GEO_MAXCYC

Type: integerDefault: 50

### CFOUR_ORBITALS¶

Specifies the type of molecular orbitals used in post-HF calculations. STANDARD (=0) requests usage of the orbitals (from a corresponding HF-SCF calculation) without any modification. These are in the case of RHF/UHF the usual canonical HF orbitals and in the case of ROHF calculations the standard ROHF-orbitals with equal spatial parts for both the \(\alpha\) and the \(\beta\) spin orbitals. SEMICANONICAL (=1) forces in ROHF type calculations a transformation to so-called semicanonical orbitals which diagonalize the occupied-occupied and virtual-virtual blocks of the usual Fock-matrices. The use of semicanonical orbitals is, for example, required for ROHF-CCSD(T) calculations and for those calculations also automatically set. LOCAL requests a localization of the HF orbitals and this is currently done according to the Pipek-Mezey localization criterion. Note that it is strongly recommended not to use this keyword unless you know what are you doing. Default: STANDARD except for ROHF-CCSD(T) and ROHF-MP4 calculations for which SEMICANONICAL is the default.

Type: stringPossible Values: STANDARD, SEMICANONICALDefault: STANDARD

### CFOUR_PERT_ORB¶

Specifies the type of perturbed orbitals used in energy derivative calculations. STANDARD means that the gradient formulation assumes that the perturbed orbitals are not those in which the (perturbed) Fock matrix is diagonal. CANONICAL means that the perturbed orbitals are assumed to be canonical. This keyword is set automatically to CANONICAL in derivative calculations with methods which include triple excitations (MBPT[4]/MP4, CCSD+T[CCSD], CCSD[T], QCISD[T] and all iterative schemes like CCSDT-n and CC3) apart from CCSDT. IJ_CANONICAL requests a canonical perturbed-orbital treatment only for the occupied-occupied block of the unperturbed density matrix in analytic derivative calculations. For testing purposes, it is possible to force the use standard perturbed orbitals even in case of iterative triple excitations via the option FORCE_STANDA (dne?). Note also that in case of unrelaxed derivatives standard orbitals must be used. Default : STANDARD for all methods without triples (except CCSDT), CANONICAL for all methods with triples in case of relaxed derivatives.

Type: stringPossible Values: STANDARD, CANONICAL, IJ_CANONICALDefault: No Default

### CFOUR_POINTS¶

Specifies either single (=1, or SINGLE) or double (=2, DOUBLE) sided numerical differentiation in the finite difference evaluation of the Hessian. Two-sided numerical differentiation is considerably more accurate than the single-sided method, and its use is strongly recommended for production work.

Type: stringPossible Values: SINGLE, DOUBLEDefault: DOUBLE

### CFOUR_PRINT¶

Controls the amount of printing in the energy and energy derivative calculation programs. Using a value of 1 will produce a modest amount of additional output over the default value of 0, which includes some useful information such as SCF eigenvectors, Fock matrix elements, etc.

Type: integerDefault: 0

### CFOUR_PROPS¶

Specifies whether and which molecular property is calculated. OFF (=0) means that no property is calculated, FIRST_ORDER (=1) requests computation of various one-electron first-order properties (e.g., dipole moment, quadrupole moment, electric field gradient, spin densities,etc.), SECOND_ORDER (=2, in the next release replaced by STAT_POL) computes static electric polarizabilities, DYNAMICAL (=7, in the next release replaced by DYN_POL) requests the calculation of frequency-dependent polarizabilities (note that here an additional input of the frequency is required), NMR (=5) requests the calculation of NMR chemical shifts/chemical shielding tensors (by default using GIAOs), J_FC requests the calculation of the Fermi-Contact contribution to indirect spin-spin coupling constants, J_SD the calculation of the corresponding spin-dipole contribution, and J_SO the calculation of the corresponding spin-orbit contribution to J; HYPERPOL (=22) invokes a calculation of static hyperpolarizabilities, DYN_HYP (=23) requests the calculation of frequency-dependent hyperpolarizabilities, SHG (=24) the calculation of hyperpolarizabilities related to the second-harmonic generation, OPT_REC (=25) the computation of hyperpolarizabilities related to optical rectification, VERDET (=26) the calculation of Verdet constants.

Type: stringPossible Values: OFF, FIRST_ORDER, SECOND_ORDER, NMR, HYPERPOL, DYN_HYP, SHG, OPT_REC, VERDETDefault: OFF

### CFOUR_PROP_INTEGRAL¶

Allows storage of property integrals computed in

`xvdint`

on internal files (e.g.,`MOINTS`

and`GAMLAM`

, default choice INTERNAL, =0) or on external files (EXTERNAL, =1).

Type: stringPossible Values: INTERNAL, EXTERNALDefault: INTERNAL

### CFOUR_QRHFGUES¶

If this keyword is set to ON (=1), then the QRHF orbitals specified by the CFOUR_QRHF_GENERAL CFOUR_QRHF_ORBITAL and CFOUR_QRHF_SPIN (nyi?) keywords are used as a starting guess for a restarted SCF procedure. This can be an extremely useful way to converge “difficult” SCF solutions, such as those that correspond to states that are not the lowest states of a given symmetry. Note that when this option is used, the calculation that is performed is not a QRHF-CC calculation; it is instead a UHF-based or ROHF-based calculation, depending on what type of reference is specified by the CFOUR_REFERENCE keyword. The QRHF aspect of the calculation is used simply as a device to converge the orbitals.

Type: booleanDefault: false

### CFOUR_QRHF_GENERAL¶

The presence of this keyword specifies that a QRHF based CC calculation, or alternatively, an SCF calculation that uses the CFOUR_QRHFGUES option, is to be performed.

Type: arrayDefault: No Default

### CFOUR_QRHF_ORBITAL¶

By default, in QRHF calculations, electrons are removed from the highest occupied orbital in a symmetry block (symmetry block HOMO), while electrons are added to the lowest unoccupied orbital within a symmetry block (symmetry block LUMO). The purpose of the QRHF_ORBITAL keyword is to allow additional flexibility in choosing which orbitals will have their occupation numbers altered. The value of this keyword gives the offset with respect to the default orbital for the orbital which will be depopulated (or populated) in QRHF-CC calculations. For calculations involving more than one removal or addition of electrons, values are separated by commas and correspond to the CFOUR_QRHF_GENERAL input on a one-to-one basis. For example, specifying CFOUR_QRHF_GENERAL =2/-4, QRHF_ORBITAL=3/2 means that an electron will be added to the third lowest virtual in symmetry block 2 and another will be removed from the second highest occupied orbital in symmetry block 4. Examples given later in this manual further illustrate the QRHF input options and may help to clarify any confusion resulting from this documentation. (Default : 1)

Type: arrayDefault: No Default

### CFOUR_RAMAN_INT¶

ON (=1) requests a calculation of Raman intensities based on the geometrical derivatives of the static polarizability tensor, while DYN (=2) requests a calculation of Raman intensities based on the derivatives of the dynamical polarizability tensor.

Type: stringPossible Values: ON, DYN, OFFDefault: OFF

### CFOUR_RAMAN_ORB¶

Specifies whether Raman intensities are calculated with orbital relaxation with respect to the electric field perturbation (RELAXED, = 1) or without orbital relaxation (UNRELAXED, = 0).

Type: stringPossible Values: RELAXED, UNRELAXEDDefault: UNRELAXED

### CFOUR_RDO¶

Specifies whether or not relaxed density natural orbitals are to be computed. This option only has meaning for a correlated calculation. For =0, Do not compute. For =1, compute.

Type: booleanDefault: true

### CFOUR_REFERENCE¶

Specifies the type of SCF calculation to be performed. RHF (= 0) requests a restricted Hartree-Fock reference; UHF (= 1) an unrestricted Hartree-Fock reference; ROHF (= 2) a restricted open-shell Hartree-Fock calculation; TCSCF (=3) a two-configurational SCF calculation, and CASSCF (=4) a complete-active space SCF calculations (currently not implemented).

Psi4 Interface:Keyword subject to translation from value of REFERENCE unless set explicitly.

Type: stringPossible Values: RHF, UHF, ROHF, TCSCF, CASSCFDefault: RHF

### CFOUR_RELATIVISTIC¶

Specifies the treatment of relativistic effects. The default is a non-relativistic treatment (OFF), while perturbational treatments are invoked via MVD1 (mass-velocity and 1-electron Darwin contribution), MVD2 (mass-velocity and 1- and 2-electron Darwin contribution), DPT2 (second-order direct perturbation theory approach), SF-DPT4 (scalar-relativistic part of fourth-order direct perturbation theory, DPT4 (full fourth-order DPT including spin-orbit corrections), SF-DPT6 (scalar-relativistic part of sixth-order direct perturbation theory), SFREE (spin-free treatment), X2C1E (spin-free X2C-1e treatment), or DPT (synonym with DPT2).

Type: stringPossible Values: OFF, MVD1, MVd2, DPT2, SF-DPT4, DPT4, SF-DPT6, SFREE, X2C1E, DPTDefault: OFF

### CFOUR_RELAX_DENS¶

Specifies whether the relaxed density matrix is computed for correlated wave functions. OFF (= 0) The relaxed density will not be computed, ON (= 1) it will be computed.

Type: booleanDefault: false

### CFOUR_RESTART_CC¶

Offers the possibility to restart a CC calculation which stopped for various reasons, e.g. time limit, in the correlation part. However, note that a restart which is specified by ON (= 1) needs the following files of the previous unfinished calculation:

`JOBARC`

,`JAINDX`

,`MOINTS`

, and`MOABCD`

.

Type: booleanDefault: false

### CFOUR_RES_RAMAN¶

This option can be used to convert an analytically calculated gradient vector to a particular normal coordinate representation. A useful application is to calculate the gradient of an electronically excited state in the normal coordinate representation of the ground electronic state, as this provides a first approximation to resonance Raman intensities (hence the name of the keyword). Calculations that use the this option require the externally supplied force constant matrix

`FCMFINAL`

, which is written to disk during the course of both analytic and finite-difference vibrational frequency calculations. No such transformation is performed if OFF (=0); while ON (=1) directs the program to evaluate the gradient and transform it to the chosen set of normal coordinates. A warning message is printed if the force constant matrix is unavailable.

Type: booleanDefault: false

### CFOUR_ROT_EVEC¶

Specifies which eigenvector of the orbital rotation Hessian is to be used to rotate the original SCF orbitals. By default, it will use that associated with the lowest eigenvalue of the totally symmetric part of the block-factored Hessian, as this choice often leads to the lowest energy SCF solution. For RHF stability checks, only those instabilities which correspond to RHF solutions will be considered. It is important to understand that following non-symmetric eigenvectors lowers the symmetry of the wavefunction and that following RHF –> UHF stabilities leads to a UHF solution. To converge the SCF roots associated with such instabilities, one must run the calculation in reduced symmetry and as a closed-shell UHF case, respectively. Value

ndirects the program to follow the vector associated with thenth lowest eigenvalue having the proper symmetry (totally symmetric) and spin (RHF–>RHF or UHF–>UHF) properties. 0 means use the lowest eigenvalue.

Type: integerDefault: 0

### CFOUR_SAVE_INTS¶

Tells CFOUR whether to delete large files (AO integrals and

`MOINTS`

file for now) when they are no longer needed. OFF (=0) They will not be saved, ON (=1) they will be saved.

Type: booleanDefault: false

### CFOUR_SCALE_ON¶

Controls whether step scaling is based on the absolute step length (1-norm) (=0 or MAG(S)) or the largest individual step in the internal coordinate space (=1 or MAX(S)).

Type: stringPossible Values: MAG(S), MAX(S)Default: MAG(S)

### CFOUR_SCF_CONV¶

Specifies the convergence criterion for the HF-SCF equations. Equations are considered converged when the maximum change in density matrix elements is less than \(10^{-N}\).

Psi4 Interface:Keyword subject to translation from value of D_CONVERGENCE unless set explicitly.

Type: integerDefault: 7

### CFOUR_SCF_DAMPING¶

Controls the damping (in the first iterations (specified by CFOUR_SCF_EXPSTART via \(D_{new} = D_{old} + X/1000 * (D_{new} - D_{old})\) with \(X\) as the value specified by the keyword. The default value is currently 1000 (no damping), but a value of 500 is recommended in particular for transition metal compounds where the SCF convergence is often troublesome.

Psi4 Interface:Keyword subject to translation from value of DAMPING_PERCENTAGE unless set explicitly.

Type: integerDefault: 1000

### CFOUR_SCF_EXPORDER¶

Specifies the number of density matrices to be used in the DIIS convergence acceleration procedure.

Type: integerDefault: 6

### CFOUR_SCF_EXPSTART¶

Specifies the first iteration in which the DIIS convergence acceleration procedure is applied.

Type: integerDefault: 8

### CFOUR_SCF_EXTRAPOLATION¶

Specifies whether or not the DIIS extrapolation is used to accelerate convergence of the SCF procedure. OFF (=0) means do not use DIIS, ON (=1) means use DIIS.

Type: booleanDefault: true

### CFOUR_SCF_MAXCYC¶

Specifies the maximum number of SCF iterations.

Psi4 Interface:Keyword subject to translation from value of MAXITER unless set explicitly.

Type: integerDefault: 150

### CFOUR_SD_FIELD¶

Specifies the strength of a spin-dipole perturbation as required for finite-field calculations of the SD contributions to indirect spin-spin coupling constants. The value must be specified as an integer and the SD strength used by the program will be the value of the keyword \(\times 10^{-6}\). (Default : 0, currently not implemented)

Type: integerDefault: 0

### CFOUR_SPHERICAL¶

Specifies whether spherical harmonic (5d, 7f, 9g, etc.) or Cartesian (6d, 10f, 15g, etc.) basis functions are to be used. ON (= 1) uses spherical harmonics, OFF (= 0) uses Cartesians.

Psi4 Interface:Keyword set according to basis design when BASIS is used instead of CFOUR_BASIS Keyword subject to translation from value of PUREAM unless set explicitly.

Type: booleanDefault: true

### CFOUR_SPINROTATION¶

Specifies whether nuclear spin-rotation tensors are computed within a NMR chemical shift calculation (ON, =1) or not (OFF, =9). In the case of electronic g-tensor calculations for open-shell molecules this keyword controls the calculation of the electronic spin-rotation tensor.

Type: booleanDefault: false

### CFOUR_SPIN_FLIP¶

Controls whether excitation energy calculations allow for a “spin flip” which changes the \(M_s\) quantum number. Such calculations have some advantages for biradicals and are currently implemented (together with gradients) for CIS and CIS(D) calculations. Options are OFF and ON.

Type: booleanDefault: false

### CFOUR_SPIN_ORBIT¶

Experimental Use! ON (=1) requests calculation of one-electron spin-orbit integrals. MEANSO additionally gives a mean-field treatment of the two-electron terms (spin-orbit mean field treatment as described Mol. Phys. 98, 1823-1833 (2000)).

Type: stringPossible Values: ON, MEANSO, OFFDefault: OFF

### CFOUR_SPIN_SCAL¶

ON (=1) requests the spin-component scaled variant of the MP2 approach. This keyword has only an effect when CFOUR_CALC_LEVEL =MP2 is specified and must be used together with CFOUR_REFERENCE =UHF.

Type: booleanDefault: false

### CFOUR_SUBGROUP¶

Specifies an Abelian subgroup to be used in a calculation. Acceptable arguments are DEFAULT (=0); C1 (= 1); C2 (= 2); CS (= 3); CI (= 4); C2V (= 5); C2H (= 6); D2 (= 7) and D2H (= 8). Use of C1 is of course equivalent to setting CFOUR_SYMMETRY =OFF in the input. The DEFAULT option (which is the default) uses the highest order Abelian subgroup.

Type: stringPossible Values: DEFAULT, C1, C2, CS, CI, C2V, C2H, D2, D2H, OFFDefault: DEFAULT

### CFOUR_SYMMETRY¶

Specifies what subgroup of the full point group is to be used in the energy and/or gradient calculation (the computational point group). OFF (=1) forces a no symmetry run (in \(C_1\) ) and ON (=0) runs the calculation in the largest self-adjoint subgroup ( \(D_{2h}\) and its subgroups).

Type: booleanDefault: true

### CFOUR_SYM_CHECK¶

In principle can be used to force the SCF to converge a solution for which the density matrix transforms as the totally symmetric representation of the point group (i.e. no broken symmetry solutions). The code seems to work in most cases, but has currently been implemented for point groups with E type representation and not for those with triply-, quadruply- or pentuply-degenerate representations. Extending the code to those cases is probably straightforward, and the reader is encouraged to do so if (s)he is so inclined. SYM_CHECK=0 “forces” the high-symmetry solution. SYM_CHECK=OVERRIDE (=1) doesn’t. The latter is the default.

Type: booleanDefault: true

### CFOUR_T3_EXTRAPOL¶

Specifies whether the T3 amplitudes are included ON (=1) or not included OFF (=0) in the DIIS convergence acceleration during CCSDT calculations. Inclusion of T3 speeds up convergence and allows tight convergence, but on the other hand it increases disk space requirements. Note that this keyword is only available with module

`xecc`

.

Type: booleanDefault: false

### CFOUR_TAMP_SUM¶

Specifies how often the largest \(t\) amplitudes are to be printed. For =0, amplitudes are printed at the beginning and end of the run. For =1, amplitudes are printed every iteration. For =2, amplitudes are printed every other iteration, etc.

Type: integerDefault: 5

### CFOUR_THERMOCHEMISTRY¶

Specifies whether to calculate finite-temperature thermodynamic corrections after a frequency calculation. OFF (=0) skips this; ON (=1) gives abbreviated output; and VERBOSE (=2) gives elaborate output that is separated by translation, rotation and vibration. Default: ON (currently not available in public version)

Type: stringPossible Values: OFF, ON, VERBOSEDefault: ON

### CFOUR_TRANS_INV¶

Specifies whether or not translational invariance is exploited in geometrical derivative calculations. USE(=0) specifies that translational invariance is exploited, while IGNORE (=1) turns it off.

Type: stringPossible Values: USE, IGNOREDefault: USE

### CFOUR_TREAT_PERT¶

Specifies whether in a correlated NMR chemical shift calculations all perturbations are treated at once or sequentially. Available option are SIMULTANEOUS (=0) and SEQUENTIAL (=1). The latter is at least preferred for large-scale calculations, as it has less demands on the available disk space.

Type: stringPossible Values: SIMULTANEOUS, SEQUENTIALDefault: SIMULTANEOUS

### CFOUR_UIJ_THRESHOLD¶

Specifies the threshold value (given as an integer) for the treatment of CPHF coefficients in second derivative calculations using perturbed canonical orbitals. If a CPHF coefficient is above the threshold, the corresponding orbital rotation is treated (at the expense of additional CPU cost) using the standard non-canonical procedures, while orbital pairs corresponding to CPHF coefficients below the threshold are treated using perturbed canonical representation. Default: 25 (Default: 1 in the developer version)

Type: integerDefault: 25

### CFOUR_UNITS¶

Specifies the units used for molecular geometry input. ANGSTROM (= 0) uses Angstrom units, BOHR (= 1) specifies atomic units.

Psi4 Interface:Keyword set from active molecule, always ANGSTROM.

Type: stringPossible Values: ANGSTROM, BOHRDefault: ANGSTROM

### CFOUR_UPDATE_HESSIAN¶

Specifies whether or not the Hessian update is carried out. OFF (= 0) uses the initial Hessian (however supplied, either the default guess or a

`FCMINT`

file), ON (= 1) updates it during subsequent optimization cycles. (not in current public version).

Type: booleanDefault: true

### CFOUR_VIBRATION¶

Specifies whether (harmonic) vibrational frequencies are calculated or not. If the default NO (=0) is specified then no frequencies are calculated. For ANALYTIC, vibrational frequencies are determined from analytically computed second derivatives, and for FINDIF (=2) vibrational frequencies are calculated from a force field obtained by numerical differentiation of analytically evaluated gradients (or even single-point energies) using symmetry-adapted mass-weighted Cartesian coordinates. If vibrational frequencies are calculated, a normal mode analysis using the computed force-constant matrix is performed, rotationally projected frequencies are computed, infrared intensities are determined, and zero-point energies (ZPE) are evaluated.

Type: stringPossible Values: NO, ANALYTIC, FINDIFDefault: NO

### CFOUR_VTRAN¶

This keyword defines what type of integral transformation is to be performed in the program

`xvtran`

. FULL/PARTIAL (=0) allows the transformation program to choose the appropriate type of transformation, while FULL (=1) requires a full integral transformation and PARTIAL (=2) means a MBPT(2)-specific transformation where the \((ab \vert cd)\) integrals are not formed.

Type: stringPossible Values: FULL/PARTIAL, FULL, PARTIALDefault: FULL/PARTIAL

### CFOUR_XFIELD¶

Specifies the X-component of an external electric field. The value must be specified as an integer and the field used by the program will be the value of the keyword \(\times 10^{-6}\). This allows field strengths \(|\varepsilon| > 10^{-6}\) to be used.

Type: integerDefault: 0

### CFOUR_XFORM_TOL¶

The tolerance for storing transformed integrals. Integrals less than \(10^{-N}\) are neglected and not stored on disk.

Type: integerDefault: 11

### CFOUR_YFIELD¶

Specifies the Y-component of an external electric field. The value must be specified as an integer and the field used by the program will be the value of the keyword \(\times 10^{-6}\). This allows field strengths \(|\varepsilon| > 10^{-6}\) to be used.

Type: integerDefault: 0

### CFOUR_ZFIELD¶

Specifies the Z-component of an external electric field. The value must be specified as an integer and the field used by the program will be the value of the keyword \(\times 10^{-6}\). This allows field strengths \(|\varepsilon| > 10^{-6}\) to be used.

Type: integerDefault: 0

*Expert* Psi4 Control of CFOUR¶

### CFOUR_OMP_NUM_THREADS¶

Sets the OMP_NUM_THREADS environment variable before calling CFOUR. If the environment variable

`OMP_NUM_THREADS`

is set prior to calling Psi4 then that value is used. When set, this option overrides everything. Be aware the`-n`

command-line option described in section Threading does not affect CFOUR.

Type: integerDefault: 1