| name | default | description |
| CVG_ENR | 1.0e-5 | Convergence criterium for energy (when IDCVG ≥ 2) |
| CVG_GRD | 1.0e-4 | Convergence criterium for maximum component of gradient; depending on the value of IDELOCAL, either the delocalized or Cartesian gradient is checked |
| CVG_STP | 1.0e-4 | Convergence criterium for maximum component of step (when IDCVG ≥ 2) |
| DIFSTEP | 1.0e-5 | Stepsize for numerical differentiation (with numerical gradients/Hessian)) |
| I_ADD_DUMMIES | 1 | Index to do (1) or do not (0) add dummy atoms for avoiding (nearly-)linear angles |
| ICREATE | 7 | Index which method to use for generating the primitive coordinates |
| IDCVG | 1 | Index how to signal convergence: 1) check nr. of negative Hessian eigenvalues is correct and max. component and rms value of gradient are less than the convergence criterium (see CVG_GRD) 3) same as 1, but both max. component of step and change in energy should be less than their respective convergence criteria (see CVG_STP and CVG_ENR) 2) same as 3, but only of the additional criteria has to be fulfilled |
| IDELOCAL | 1 | Kind of coordinates to use in the geometry optimization: 1) adapted delocalized coordinates 0) Cartesian coordinates |
| IDIIS | 3 | Kind of GDIIS equations to use: 0) original GDIIS 1) same as 0, but with Farkas-Schlegel rules applied 2) use gradient as error vector 3) same as 2, but with Farkas-Schlegel rules applied 4) use 'energy' vector as error vector 5) same as 4, but with Farkas-Schlegel rules applied |
| IDSTEP | 5 | Step to take: 1) RSO for minimizations, RFO (Baker) for TransitionStates 3) RFO (Baker) always 5) Generalized RSO (Swart) using image-function for TransitionStates |
| IEXCST | 1 | Number of excited state to use for numerical gradients By default for singlet excited state; triplet excited state can be used by adding ONLYTRIP keyword to EXCITATIONS block on input |
| IHOPT | 3 | Index for force constants method to use for initial Hessian: 0) Baker (0.5 bonds, 0.2 angles, 0.1 dihedrals) 1) Thomas Fischer 2) simplification of Lindh 3) Swart-Bickelhaupt scheme 7) Swart generalized scheme (works well for close to minima) |
| IHUPD | -1 cq. 4 | Index for Hessian update scheme: 1 BFGS for Hessian (-1 BFGS for inverse Hessian) 2 Powell-symmetric-Broyden, PSB (for Transition States) 3 Murtagh-Sargent (Symmetric Rank-One, SR1) 4 Bofill weighted combi of PSB and SR1 (for Transition State) 5 Farkas-Schlegel weighted combi of BFGS and SR1 6 Bakken-Halgaker combi of BFGS and SR1 |
| IQUILD_OUTPUT | 1 | Amount of output requested, debug output ≥2 |
| IRESTART | 0 | Index if ADF/ORCA jobs should restart from t21.files from previous geometry < 0 ORCA uses restart, ADF not > 0 both ORCA and ADF use restart |
| ITRUST | 0 | Index if dynamic trust radius should be used (1) or not (0) |
| MXDIIS | 5 | Maximum number of GDIIS vectors to use |
| MXGEO | 50 | Maximum number of geometry cycles (overrides value read from ITERATIONS in GEOMETRY block) |
| NR_REGIONS | 1 | Number of different regions for multi-level approach |
| NRLT | 0 | Number of LinearTransit steps |
| RTRUST | 0.20 | Trust radius value |
| SMETAGGA | - | String for functional from METAGGA post-SCF scheme to use for numerical gradients, should be given exactly as on METAGGA output |
| TRUST_ALFA | 1.20 | Factor to increase trust radius with if Δ energy agrees with model prediction |
| TRUST_BETA | 0.70 | Factor to decrease trust radius with if Δ energy does not agree with model |
| TRUST_GOOD | 0.80 | Lower threshold for increasing trust radius |
| TRUST_RMIN | 0.40 | Upper threshold for increasing trust radius |
The other keywords that are printed in the output are for debug purposes, under development, or of technical nature. More information about them can be obtained (if needed) from SCM or M. Swart.
