Input variable list
Mandatory input variables
scf basis
Physical system setup input variables
nspin charge magnetization temperature length_unit natom xyz_file nghost
General input variables
scf move_nuclei nstep tolforce basis auxil_basis incore gaussian_type ecp_type ecp_elements ecp_quality ecp_basis ecp_auxil_basis
Self-consistency input variables
alpha_hybrid beta_hybrid gamma_hybrid grid_quality integral_quality partition_scheme nscf alpha_mixing mixing_scheme diis_switch level_shifting_energy init_hamiltonian tolscf density_matrix_damping kerker_k0 min_overlap npulay_hist
Correlation and excited states post-treatment input variables
postscf tddft_grid_quality nstep_gw tda triplet nexcitation toldav nstep_dav frozencore ncoreg ncorew nvirtualg nvirtualw nvirtualspa nomega_imag selfenergy_state_min selfenergy_state_max selfenergy_state_range nomega_sigma step_sigma pt3_a_diagrams eta scissor alpha_cohsex beta_cohsex dft_core gamma_cohsex delta_cohsex epsilon_cohsex virtual_fno rcut_mbpt gwgamma_tddft ci_greens_function ci_type ci_nstate ci_nstate_self ci_spin_multiplicity
IO input variables
basis_path read_restart ignore_bigrestart force_energy_qp print_hartree print_exchange print_eri print_wfn print_cube read_fchk print_w print_restart print_bigrestart print_sigma print_pdos print_multipole
Hardware input variables
grid_memory scalapack_block_min scalapack_nprow scalapack_npcol mpi_nproc_ortho
Complete list of input variables
- scf
Mandatory
Default: None
Contains the self-consistent scheme name.
Try LDA, PBE, HSE06, or HF for instance
- postscf
Optional
Default: None
Contains the post-processing scheme name.
TD stands for TD-DFT or TD-HF.
BSE stands for Bethe-Salpeter.
GW stands for perturbative G0W0.
GnW0 stands for GW with eigenvalue self-consistentcy on G.
GnWn stands for GW with eigenvalue self-consistentcy on both G and W.
MP2 stands for guess what.
GWGAMMA (EXPERIMENTAL) stands for vertex corrections.
- move_nuclei
Optional
Default: no
Tells the code to move or not the position of the nuclei. Available options are 'no' or 'relax'.
- nstep
Optional
Default: 50
Sets the number of steps when moving the nuclei.
- tolforce
Optional
Default: 1e-05
Sets the target threshold for the maximum force component after nuclei relaxation.
- alpha_hybrid
Optional
Default: 0.0
Only works for Range-Separated hybrid functionals scf='rsh' Sets the amount of range-independent exact-exchange
- beta_hybrid
Optional
Default: 0.0
Only works for Range-Separated hybrid functionals scf='rsh' Sets the amount of long-range exact-exchange
- gamma_hybrid
Optional
Default: 1000000.0
Only works for Range-Separated hybrid functionals scf='rsh' Sets the separation between long-range and short-range. It is input in bohr^-1.
- basis
Mandatory
Default: None
Sets the basis set For Pople sets, use 6-31G for instance or 6-31pGs, where p stands for + and s for *. For Dunning sets, use aug-cc-pVTZ for instance. Note that Pople sets are to be used with gaussian_type='cart' One may use ones own basis sets provided that the files are labeled X_mybasisset where X is the element.
- auxil_basis
Optional
Default: None
Sets the auxiliary basis set. For instance, cc-pVDZ-RI for a Weigend basis set. If present, the auxiliary basis will be used for both the scf cycles and the postscf calculations (TD-DFT, BSE, or GW).
- incore
Optional
Default: yes
Specify if the 4-center integrals are all calculated at once and stored or if they are calculated on-the-fly.
- basis_path
Optional
Default: None
Sets the path pointing to the basis functions files. If not specified, then the basis set files will be searched in folder ~molgw/basis/.
- small_basis
EXPERIMENTAL
Optional
Default: None
Calls for a smaller basis set used to represent the virtual orbital space with fewer functions. Only meaningful for GW.
- ecp_small_basis
EXPERIMENTAL
Optional
Default: None
Calls for a smaller basis set used to represent the virtual orbital space with fewer functions. This is the small basis set used for elements with an effective core potential. Only meaningful for GW.
- gaussian_type
Optional
Default: pure
Asks for pure or spherical Gaussian type orbitals with 'pure' or for Cartesian Gaussian orbital with 'cart'.
- nspin
Optional
Default: 1
Sets the number of spin channels. 1 enforces spin-restricted calculations. 2 means spin-unrestricted.
- charge
Optional
Default: 0.0
Sets the total charge of the system. 0 is a neutral system. -2 is a doubly charged anion etc.
- magnetization
Optional
Default: 0.0
Sets the number of unpaired electrons. In other words, this is the difference between the spin up and spin down occupation. For instance, a spin-doublet calculation is obtained with magnetization=1.0. Only meaningful when nspin=2.
- temperature
Optional
Default: 0.0
Sets the electronic temperature in the Fermi-Dirac functions. Helps the convergence for some systems. The value is input in Hartree atomic units.
- grid_quality
Optional
Default: high
Sets the number of grid points use to evaluate the exchange-correlation integrals in real space for the DFT potential and energy. Possible values are 'low', 'medium', 'high', 'very high', 'insane'. It could be abbreviated in 'l', 'm', 'h', 'vh', 'i'. 'high' is usually fine. 'insane' is only meant for debugging since it is overdoing a lot.
- tddft_grid_quality
Optional
Default: high
Sets the number of grid points use to evaluate the exchange-correlation integrals in real space for the TDDFT kernel. Possible values are 'low', 'medium', 'high', 'very high', 'insane'. It could be abbreviated in 'l', 'm', 'h', 'vh', 'i'. 'high' is usually fine. 'insane' is only meant for debugging since it is overdoing a lot.
- integral_quality
Optional
Default: high
Sets the tolerance value for the screening of the negligible integrals. Possible values are 'low', 'medium', 'high', 'very high', 'insane'. It could be abbreviated in 'l', 'm', 'h', 'vh', 'i'. 'high' is usually fine. 'insane' is only meant for debugging since it is overdoing a lot.
- partition_scheme
Optional
Default: ssf
Sets the partition scheme for the xc quadrature. Possible choices are 'becke' or 'ssf' (Stratmann-Scuseria-Frisch).
- nscf
Optional
Default: 50
Sets the maximum number of SCF cycles
- alpha_mixing
Optional
Default: 0.7
Sets the amount of output density-matrix for the next iteration. When the SCF cycles have difficulties to converge, one may try to lower this value.
- mixing_scheme
Optional
Default: pulay
Sets the density-matrix update method for SCF cycles. Possible choices are 'pulay' for Pulay DIIS method, 'adiis' for Hu-Yang method, or 'simple' for a simple linear mixing between input and output density-matrices.
- diis_switch
Optional
Default: 0.05
When running ADIIS, sets the residue value below which the DIIS method is used to finalize the convergence.
- level_shifting_energy
Optional
Default: 0.0
Sets the energy shift up of the unoccupied states. Should help the convergence in the case of small HOMO-LUMO gaps.
- init_hamiltonian
Optional
Default: guess
Selects how to initiate the first hamiltonian for SCF cycles. Today, two options are available: 'guess' for an educated guess based on approximate atomic densities or 'core' for the core hamiltonian.
- tolscf
Optional
Default: 1e-07
Sets the residual norm target for the density matrix for the SCF cycles.
- density_matrix_damping
Optional
Default: 0.0
Adds an additional linear mixing on the density matrix in combination with the Hamiltonian mixing in order to damp out the charge oscillations. Especially useful for metallic systems.
- kerker_k0
EXPERIMENTAL
Optional
Default: 0.0
Analog to k0 in Kerker preconditioning for metallic systems. Helps to damp charge oscillations to ensure better SCF convergence.
- min_overlap
Optional
Default: 1e-05
Sets the minimal eigenvalue of the overlap matrix S. Small eigenvalues imply overcompleteness of the basis set.
- npulay_hist
Optional
Default: 6
Sets the history record length for Pulay DIIS.
- nstep_gw
Optional
Default: 1
Sets the number of GW iterations for eigenvalue self-consistent GW calculations (GnWn or GnW0).
- tda
Optional
Default: no
Triggers the use of Tamm-Dancoff approximation in TD-DFT or BSE.
- triplet
Optional
Default: no
Triggers the calculation of the triplet final state in TD-DFT or BSE.
- nexcitation
Optional
Default: 0
Sets the number of neutral excitations to be calculated in TD-DFT or BSE. 0 stands for all the states and triggers the full diagonalization.
- toldav
Optional
Default: 0.0001
Sets the tolerance criterium for the maximum norm of the residual in the Davidson diagonalization of TD-DFT, BSE, and full CI.
- nstep_dav
Optional
Default: 15
Sets the maximum number of Davidson partial diagonalization steps. Used for TD-DFT, BSE, and full CI.
- frozencore
Optional
Default: no
Triggers the neglect of core states in GW. H, He, Li, Be have no core states. B-Na have the 1s. Al-Ca have the 1s2s2p. Manual tuning could be achieved with ncoreg, ncorew.
- ncoreg
Optional
Default: 0
Sets the number of frozen core states in the Green's function G.
- ncorew
Optional
Default: 0
Sets the number of frozen core states in the screened Coulomb interaction W, in TD-DFT, and in BSE.
- nvirtualg
Optional
Default: 100000
Sets the starting state beyond which states are excluded from the sum in the Green's function G.
- nvirtualw
Optional
Default: 100000
Sets the starting state beyond which states are excluded from the sum in the screened Coulomb interaction W, in TD-DFT, and in BSE.
- nvirtualspa
Optional
Default: 100000
Sets the starting state beyond which states are accounted for with a Single Pole Approximation for the screened Coulomb interaction W for GW.
- nomega_imag
Optional
Default: 0
Sets the number of frequencies used to perform the integral on the imaginary axis
- selfenergy_state_min
Optional
Default: 1
Sets the starting states for the range of the self-energy evaluation
- selfenergy_state_max
Optional
Default: 100000
Sets the final states for the range of the self-energy evaluation
- selfenergy_state_range
Optional
Default: 100000
Sets the range of states around the HOMO level for the self-energy evaluation. For instance, selfenergy_state_range=0 will trigger the calculation of the HOMO only. selfenergy_state_range=1 will trigger the evaluation of the HOMO-1, HOMO, HOMO+1. etc.
- nomega_sigma
Optional
Default: 51
Sets the number of frequencies used to solve the quasiparticle equation in the GW self-energy.
- step_sigma
Optional
Default: 0.01
Sets the spacing between frequencies in the GW self-energy evaluation.
- pt3_a_diagrams
Optional
Default: yes
Switch whether to calculate the A diagrams family in PT3. A diagrams are the self-consistent diagrams (PT2 inclusions in the Green's function).
- read_restart
Optional
Default: no
Read the RESTART file and restart from it.
- ignore_bigrestart
Optional
Default: no
Considers a big RESTART as if it was a small RESTART.
- force_energy_qp
Optional
Default: no
Force the reading of the ENERGY_QP file whatever the postscf choice.
- print_hartree
Optional
Default: no
Prints the Hartree potential expectation value on eigenstates.
- print_exchange
Optional
Default: no
Prints the Exchange operator expectation value on eigenstates.
- print_eri
Optional
Default: no
Dumps the Electron Repulsion Integral on a file.
- print_wfn
Optional
Default: no
Prints some wavefunctions along some selected lines.
- print_cube
Optional
Default: no
Prints some wavefunctions in a 3D volumetric file with cube format
- read_fchk
Optional
Default: no
Triggers the reading of an external Gaussian formatted checkpoint file (named gaussian.fchk) that contains density matrices. Basis sets have to be precisely the same in MOLGW and in Gaussian, which requires a manual input of the basis set in both codes. Options are 'no' (no reading), 'SCF' (for self-consistent field), 'CC' (for coupled-cluster), or 'MP2' (for MP2). Today, only works for Cartesian Gaussian and for spin restricted calculations.
- print_w
Optional
Default: no
Dumps the spectral function of the screened Coulomb W. This is necessary for a subsequent BSE run.
- print_restart
Optional
Default: yes
Prints a small RESTART file at each SCF cycle. There are two kinds of RESTART files: the small RESTART and the big RESTART. The former contains only the information about the occupied wavefunctions. This is a very small file and the writing should not hit too much on performance.
- print_bigrestart
Optional
Default: yes
Prints the big RESTART file at the end of the SCF loop. There are two kinds of RESTART files: the small RESTART and the big RESTART. The latter is written only when self-consistency has been reached. It contains all the states and the Hamiltonian and allows one to completely skip the scf loop or to start over with another basis set.
- print_sigma
Optional
Default: no
Prints the value of the GW self-energy on the sampling frequencies in files.
- print_pdos
Optional
Default: no
Prints the Mulliken weight of each eigenvector on a given atom or a given series of atoms.
- print_multipole
Optional
Default: no
Prints the electric multipole expansion for the electronic density and the nuclei.
- length_unit
Optional
Default: angstrom
Chooses the units of the atomic coordinates. Can be 'angstrom' or 'bohr'. Could be abbreviated in 'A' or 'au'.
- natom
Optional
Default: 0
Sets the number of atoms in the molecule. This is the number of lines to be read in the following section of the input file if no xyz file is provided.
- xyz_file
Optional
Default: None
Specifies the location of the xyz file that contains the atomic positions. It can be used as an alternate route to set atomic coordinate.
- nghost
Optional
Default: 0
Sets the number of ghost atoms in the molecule. Used to place basis function where there is no atom. Useful for Basis Set Superposition Error
- eta
Optional
Default: 0.001
Is a the tiny imaginary part used in the denominator of the Green's function to shift the pole off the axis, so to avoid divergences.This is an energy in Hartree. It should be set to the lowest value possible in theory. However, in practice, a too low value of eta would induce huge and unstable GW corrections. The default value is usually very accurate and there is no need to use a lower value. But for states apart from the band gap, a large value of eta may be beneficial for stability. eta=0.01 is already much more stable. Note that for QSGW increasing eta is most often unavoidable.
- scissor
Optional
Default: 0.0
Sets a rigid energy shift of the unoccupied states, so to mimick a GW calculation without actually doing it.
- grid_memory
Optional
Default: 400.0
Sets the maximum memory usage in Mb allowed to store the wavefunctions on the quadrature points for XC integrals.
- scalapack_block_min
Optional
Default: 400
Sets the minimum block size to distribute a non-distributed matrix with SCALAPACK. If scalapack_block_min=400, then a 900x900 matrix will be distributed on a 2x2 processor grid. If scalapack_block_min=500, then a 900x900 matrix will no be distributed.
- scalapack_nprow
Optional
Default: 1
Sets number of row processors for SCALAPACK distribution of the SCF matrices. nprow X npcol should be lower or equal to the number of processors.
- scalapack_npcol
Optional
Default: 1
Sets number of column processors for SCALAPACK distribution of the SCF matrices. nprow X npcol should be lower or equal to the number of processors.
- mpi_nproc_ortho
Optional
Default: 1
Sets the number of processors left to parallelize on other directions. The main direction (auxiliary basis or DFT grid points) is obtained by mpi_nproc / mpi_nproc_ortho, which must be an integer.
- alpha_cohsex
Optional
Default: 1.0
Sets the amount of static Screened EXchange in the self-energy. Only works with scf='COHSEX' or with postscf='COHSEX'.
- beta_cohsex
Optional
Default: 1.0
Sets the amount of static COulomb Hole in the self-energy. Only works with scf='COHSEX' or with postscf='COHSEX'.
- dft_core
Optional
Default: 0
Sets the number of states considered as core in <Σ_{x}-v_{xc}>. This options is meant to mimic the pseudopotential approximation.
- gamma_cohsex
EXPERIMENTAL
Optional
Default: 0.0
EXPERIMENTAL
- delta_cohsex
EXPERIMENTAL
Optional
Default: 0.0
EXPERIMENTAL
- epsilon_cohsex
EXPERIMENTAL
Optional
Default: 0.0
EXPERIMENTAL
- virtual_fno
Optional
Default: no
Activates the Frozen Natural Orbitals technique to span the virtual orbitals subspace with fewer orbitals. The dimension of the space is set up with the input keyword nvirtualg or nvirtualw. Actually the virtual orbital space is determined by the minimum MIN(nvirtualg,nvirtualw).
- rcut_mbpt
EXPERIMENTAL
Optional
Default: 1.0
EXPERIMENTAL
- gwgamma_tddft
EXPERIMENTAL
Optional
Default: no
EXPERIMENTAL. Calculates the vertex using the DFT flavor specified in the ground-state calculation.
- ecp_type
Optional
Default: None
Name of the Effective Core Potential. For instance, Gold using the cc-pVDZ-PP basis set should have ecp_type='PP', so that MOLGW looks for the file Au_PP in the basis_path folder.
- ecp_elements
Optional
Default: None
Contains the list of elements (separated by spaces) that should be treated with an Effective Core Potential.
- ecp_quality
Optional
Default: high
Sets the number of grid points use to evaluate the Effective Core Potential integrals in real space. Possible values are 'low', 'medium', 'high', 'very high', 'insane'. It could be abbreviated in 'l', 'm', 'h', 'vh', 'i'. 'high' is usually fine. 'insane' is only meant for debugging since it is overdoing a lot.
- ecp_basis
Optional
Default: None
Name of the basis set to be used for elements specified in list ecp_elements.
- ecp_auxil_basis
Optional
Default: None
Name of the auxiliary basis set to be used for elements specified in list ecp_elements.
- ci_greens_function
EXPERIMENTAL
Optional
Default: holes
EXPERIMENTAL. Selects which part of the Green's function is to be calculated: holes, electrons, or both.
- ci_type
Optional
Default: all
Selects which excitations will be included in the CI expansion. Valid choices are 'all', 'CISD', 'CISDT', 'CISDTQ'.
- ci_nstate
Optional
Default: 1
Selects how many CI states should be calculated in the diagonalization. If ci_nstate is lower than the number of configuration, a Davidson partial diagonalization is performed, else a full (SCA)LAPACK diagonalization is triggered.
- ci_nstate_self
Optional
Default: 1
Selects how many CI states in the N+1 or N-1 electron calculations. If ci_nstate_self is lower than the number of configuration, a Davidson partial diagonalization is performed, else a full (SCA)LAPACK diagonalization is triggered.
- ci_spin_multiplicity
Optional
Default: 1
Spin multiplicity in CI calculations.
Back to the top of the page
Generated by input_parameter.py on 25 May 2018