ORCA 6 Changes — ORCA 6.0 Manual (2024)

LeanSCF: reduced memory, more robust convergence

Electric field optimizations

General ROHF implementation (SCF/Gradient) with all approximations

General CSF ROHF

New density functionals

Delta-SCF

UHF STEOM-CCSD

UHF-IP-EOM-CCSD

UHF-EA-EOM-CCSD

Regularized OOMP2

Solvation in OOMP2

Improved stability analysis featuring all approximations, solvation etc

MixGuess to converge to biradicaloid open shell singlet type broken symmetry solutions

Approximate Spin Projection Method for broken symmetry calculations (SCF/Gradient)

TRAH, AVAS, MCRPA

Linear response CASSCF

Vastly improved Recursive CI coupling coefficient generation

MPn

CCSD(T) gradients

CCSDT

X2C

New and consistent DKH infrastructure

DLPNO-CCSD(T) PTES Approach

SMD analytical Hessian

Dynamically adjusted radii: DRACO

Improved surface grids

Interface to openCOSMO-RS

Explicit Solvator

Molecule Docker

FMM implementation for embedding

CIM implementation works with DLPNO-CC, DLPNO-MP2

More robust optimizer (fewer cycles, fewer cases with negative frequencies)

GOAT global optimizer and conformer generator

Basis set limit extrapolated optimizations through compound scripts

Extrapolation with counterpoise correction through compound scripts

Group parallelization

Performance improvements

Analytical gradient for meta-GGA functionals

VCD implementation at the SCF level

MCD with vibronic structure

General spin ROCIS

Higher order moments and exact field matter coupling

Spin rotation constants

Local dipole moments and polarizabilities

Frequency dependent electric properties

VPT2 enhancements

Restructured NMR simulation program orca_nmrspectrum

MBIS charges

Property file: Machine readable, Human readable summary of ORCA run

Compound: vastly improved Syntax, features, optimizations, …

orca_2json: generate integrals, property file, run backwards to get MOs into ORCA

Citation tool for helping find the right references

Significant improvements to the SOSCF solver to make it more robust, preventing huge steps that break the SCF. Overall improvements on the DIIS solvers.

Due to the SCF updates, the AutoTRAH is now not so often needed and will start now only from above 50 cycles (AutoTRAHIter).

Improvements to the memory handling of TD-DFT, CP-SCF and the Hessian

def-TZVP and ma-def-TZVP pseudo-potential basis sets for the actinides (Z = 89, Ac - 103, Lr)

Lehtola’s hydrogenic gaussian basis set family (HGBS) including polarized (HGBSP) and augmented (AHGBS, AHGBSP) variants for all elements up to Oganesson (Z = 118)

def2-SVPD, def2-TZVPD, def2-TZVPPD, def2-QZVPD, def2-QZVPPD basis sets for lanthanoids

!MINIX now correctly activates the corresponding ECP

Added user-specified L-limit to AutoAux AutoAuxLLimit

Fixed segfault in dhf-ECP

Fix for DelECP in %coords

Added ReadFragBasis keywords read fragment-specific basis sets from a file

New charge correction / compensation algorithm (corrected charges printed in an additional file)

C-PCM/B scheme for QM/MM calculations

DDCOSMO and CPCM/X available for XTB calculations and QM/MM calculations

Generalization of names within all solvation models (C-PCM/SMD/ALPB/DDCOSMO/CPCM-X)

New discretization scheme for the cavity (C-PCM) based on a constant number of charges per unit of area

Allow LibXC functional customization via external parameters

Simple input keywords added for some LibXC functionals

Added wB97M(2) functional parameters: must be used with wB97M-V orbitals in a two-step job (compound script available)

Bugfixes for LibXC combined *_xc_* functionals

Fixed crash for D4 + ghost atoms

Analytical gradient for meta-GGA functionals

Small bugfix to spin-adapted triplets and NACMEs.

The FolllowIRoot for excited state optimization uses now a much more robust algorithm.

Enabled NumGrad with relativistic methods

Second order DKH picture-change correction of contact density

Minor fixes in DKH picture-change corrections of magnetic properties

Picture change corrections are activated automatically

Reading PDB files for 10k+ atoms with HETATMs now possible

Enabled correct FlipSpin behavior with QMMM

More efficient MM Module

Implemented wall potential

Implemented energy ordering for PNO generation

Added semicore treatment for DLPNO

Enable DLPNO-CCSD(T) calculations to run DLPNO-CCSD unrelaxed densities

Corrected memory estimates and batching in response and gradient

Removed the slow and limited analytic (RI-)MP2 Hessian code

Removed non-default Gamma-in-core option for RI-MP2 response

Disabled single-precision calculations

Disabled SemiDirect option in AO-MP2

Enabled range-separated DHDFT gradients with RIJDX

Improved IDPP initial path

More efficient GFN-xTB runs for NEB

Improvements to numerical integration grids, both for DFT and COSX

Faster grid step

Improved performance and accuracy in COSX, also for the gradient and Hessian

NMR spin-spin coupling:

• Added SpinSpinElemPairs and SpinSpinAtomPairs keywords to limit which couplings are computed

• Reduced the number of CP-SCF perturbations necessary via a stochastic selection

• DSO term was transposed.

• Off-diagonal PSO elements had the wrong sign

• Efficiency improvement: solve SD/FC CP-SCF equations in restricted mode for RHF, instead of always using UHF

Optimized numeric integration for HFC gauge correction

Removed RITRAFO option for CP-SCF

Switched to tau=Dobson as default handling of the kinetic energy density in meta-GGA magnetic properties with GIAOs

Improvements to the Hessian to avoid accumulation on numerical noise and reduce the number of spurious negative frequencies.

Several improvements to the geometry optimization, making is much more stable. Complete redesign of the Cartesian optimizer (!COPT), making it quick enough to be used together with faster methods.

Fallbacks in the geometry optimization in case something fails, e.g. if the internal coordinates are unacceptable.

Arbitrary spherical, ellipsoidal or box-like wall potentials can be added, which will reflect on the energy and gradients and can be used during geometry optimization.

CHELPG charges that reproduce the ESP together with the molecular dipole moment

Fixed issues with constraints in multi-step jobs

Molden output: store ECP info in [Pseudo] block, set point charge atomic number to 0, handling of ghost atoms

Made the ExtOpt interface easier to use

Store energy from NEB and IRC in the XYZ file