Python interface of DIRAC with Openfermion (Bruno Senjean) to perform relativistic quantum chemistry calculations simulated on a quantum computer .
Nuclear Spin-Rotation tensors. Contributors: I. Agustin Aucar and Trond Saue.
Reference: I. A. Aucar, S. S. Gómez, M. C. Ruiz de Azúa, and C. G. Giribet Theoretical study of the nuclear spin-molecular rotation coupling for relativistic electrons and non-relativistic nuclei.J. Chem. Phys. 136 (2012) 204119
Nuclear Magnetic-Quadrupole-Moment interaction constant in KRCI (Malaya K. Nayak)
Reference: T. Fleig, M. K. Nayak and M. G. Kozlov TaN, a molecular system for probing P,T-violating hadron physics.Phys. Rev. A 93 (2016) 012505
Improvements
Improved root tracking for EOM-CC (Luuk)
Use Kramers conjugation on doubly degenerate CI vectors in GASCIP code (cuts time for CI in half for ESR doublets) (Hans Jørgen)
Bugfixes
DFT magnetizatibilities with LAOS and symmetry (Gosia Oejniczak and Trond Saue)
Resolved runtime issues in KRCI property modules (Malaya K. Nayak)
Change of defaults
Upgrade to python3
New features in DIRAC18
DFT magnetizabilities. Contributors: M. Olejniczak and Trond Saue.
Limitations: Magnetizabilities and NMR shieldings calculated at the DFT level are so far restricted to C1 symmetry, but we expect to fix this soon.
Enhancements to the frozen density embedding (FDE) functionality
FDE contributions to magnetic properties (NMR shieldings, indirect spin-spin coupling constants, magnetizabilities), see ".FDE" and "*FDE" entries of the manual for details. Contributors: M. Olejniczak, R. Bast, A. S. P. Gomes
References:
M. Olejniczak, R. Bast, A. S. P. Gomes On the calculation of second-order magnetic properties using subsystem approaches in a relativistic framework.Phys. Chem. Chem. Phys. 19 (2017) 8400
FDE interaction energies with CCSD, MP2 and mean-field densities. Contributors: M. Olejniczak, A. Shee, R. Bast, A. S. P. Gomes
Equation of motion coupled cluster
Energies for electronic excitations (EE), electron attachment (EA) and electron detachment (IP), see ".EOMCC", "*EOMCC" and "*CCDIAG" entries of the manual for details. Contributiors: A. Shee, T. Saue, L. Visscher, A. S. P. Gomes
References:
A. Shee, T. Saue, L. Visscher, A. S. P. Gomes Equation-of-motion coupled-cluster theory based on the 4-component Dirac-Coulomb(-Gaunt) Hamiltonian. Energies for single electron detachment, attachment, and electronically excited states.J. Chem. Phys. 149 (2018) 174113
Improvements
Polarized embedding can be done with xyz-files. Contributor: Trond Saue
Improved quaternion diagonalization Contributor: H. J. Aa. Jensen
Improvements in the visualization module (**VISUAL) Contributors: M. Olejniczak and T. Saue.
the possibility to calculate the NMR shielding tensor in a selected point in space, see ".NICS"
the possibility to visualize various densities on an imported 3D grid, see ".3D_IMP"
the possibility to calculate magnetic properties densities using the imported magnetically-induced current density, see ".READJB"
the possibility to scale densities by Cartesian products x^{i}y^{j}z^{k}, see ".CARPOW"
the possibility to generate and visualize radial distributions, see ".RADIAL"
Change of defaults
New convergence criterium for CC amplitude equation: The convergence criterium for the amplitude equations that determine the CC energy has been revised and made consistent with the criterium used in the lambda equations used for molecular properties. In both cases we now take the norm of the differences between amplitudes of subsequent iterations. In practice this typically means the program will use a few iterations less. For normal calculations this is of no consequence as the default is still to converge very tightly, but if extremely high precision is required one may need to check the achieved convergence.
Change in the reorthonormalization terms in the calculation of magnetic properties with London atomic orbitals: the reorthonormalization and response contributions involve the same orbital pairs, for instance if all rotations between occupied and virtual orbitals are present in response equations, the reorthonormalization terms are also constructed from all orbital blocks; the keywords .DOEPRN and .NOEPRN under *NMR are depreciated;
New features in DIRAC17
Kramers-restricted Polarization Propagator in the ADC framework for electronic excitations, activated with ”.POLPRP”.
References:
M. Pernpointner. The relativistic polarization propagator for the calculation of electronic excitations in heavy systems.J. Chem. Phys. 140, 084108 (2014)
M. Pernpointner, L. Visscher and A. B. Trofimov. Four-component Polarization Propagator Calculations of Electron Excitations: Spectroscopic Implications of Spin-Orbit Coupling Effects.J. Chem. Theory Comput. 14, 1510 (2017).
Reference: see T. Fleig and M. K. Nayak. Electron electric-dipole-moment interaction constant for HfF^{+} from relativistic correlated all-electron theory. Phys. Rev. A, 88:032514, 2013
Reference: M. Denis, M. Nørby, H. J. Aa . Jensen, A. S. P. Gomes, M. K. Nayak, S. Knecht, and T. Fleig. Theoretical study on ThF^{+}, a prospective system in search of time-reversal violation.New J. Phys., 17:043005, 2015
New ”.MVOFAC” option in *KRMC input section for Modified Virtual Orbitals in MCSCF. Contributor: H. J. Aa. Jensen.
New and numerically stable procedure for elimination/freezing of orbitals at SCF level. Contributor: T. Saue.
Support for use of DIRAC in PyADF and QMFlows workflow engines. Contributors: Lucas Visscher, Andre Gomes and Christoph Jacob
New easier options for point charges in the .mol file: “LARGE POINTCHARGE” or “LARGE NOBASIS” (the two choices are equivalent), see here
Provided memory counter for RelCC calculations, suitable for memory consuming large scale Coupled Cluster calculations, see here for details. Contributor: Miroslav Iliaš
Write out effective Hamiltonian in Fock space coupled cluster to a file for post processing. Can be used with external code of Andrei Zaitsevskii (St. Petersburg).
Restart for RELCCSD. Contributor: Andre Gomes. See the keyword .RESTART and the section *CCRESTART
Performance Improvements
Restored integral screening. Contributor: Hans Jørgen Aagaard Jensen
POLPRP module + Davidson diagonalizer now parallel. Contributor: Markus Pernpointner
Corrections
Fixed errors for quaternion symmetries in 2-electron MO integrals used in CI calculations with GASCIP. It is now possible to do CI calculations with GASCIP for C1 symmetry (i.e. no symmetry).
Fixed error for parallel complex CI or MCSCF with GASCIP
Fixed compilation of XCFun on Mac OS X High Sierra.
Change of defaults
Change of final (open shell) orbital energies + SCF cycle modification. Contributors: Hans Jørgen Aagaard Jensen and Trond Saue
.SKIPEP is now default for KR-MCSCF, new keyword .WITHEP to include e-p rotations
Basis set news
Added the RPF-4Z and aug-RPF-4Z basis sets for f-elements to the already existing files with sets for s, p and d elements. Deleted the aug-RPF-3Z set as that was not an official set.
Fixed the p exponents for Na in the dyall 4z basis sets to match the archive. The changes are small so should not significantly affect results.
Updated basis_dalton/ with basis set updates in the Dalton distribution:
fix of errors in Ahlrichs-pVDZ (several diffuse exponents were a factor 10 too big)
fix of errors for 2. row atoms in aug-cc-pCV5Z
added many atoms to aug-cc-PVTZ_J
added many Frank Jensen “pc” type basis sets
added Turbomole “def2” type basis sets
New features in DIRAC16
RELCCSD expectation values. For more information, see J. Chem. Phys. 145 (2016) 184107 as well as test/cc_gradient for an example.
Improved start potential for SCF: sum of atomic LDA potentials, generated by GRASP.
Change of defaults
Negative denominators (e.g. appearing in core ionized systems) accepted in RELCCSD
AOFOCK is now default if at least 25 MPI nodes (parallelizes better than SOFOCK). And .AOFOCK documented.
Corrections
Error corrections and updates in isotope properties for the following atoms:
Br isotope 2: quadrupole moment .2620 → .2615
Ag isotope 2: magnetic moment .130563 → -.130691 (note sign change)
In isotope 2: quadrupole moment .790 → .799
Nd magnetic moments of isotopes 4 and 5 were interchanged: -0.065 → -1.065 and -1.065 → -0.065
Gd: quadrupole moments of isotopes 4 and 5 updated: 1.36 → 1.35 and 1.30 → 1.27
Ho isotope 1: quadrupole moment updated 3.49 → 3.58
Lu isotope 2: quadrupole moment updtaed 4.92 → 4.97
Hf isotope 1: mass was real*4, not real*8, thus 7 digits instead of 179.9465457D0 (i.e. approx 179.9465)
Fixed Carbon basis set (wrong contraction coefficients, see [MOLCAS ANO-RCC](http://www.molcas.org/ANO/).
Modified the 3 Th h-functions by replacing them with the 3 Ac h-functions to Th.
Fixed reading of ANO-RCC and ANO-DK3 basis sets from the included basis set library.
New defaults
For open-shell SCF calculations, .OPENFAC = 0.5 by default, as this seems to improve convergence. Final orbital energies are recalculated with .OPENFAC 1.0, for IP interpretation.
Intrinsic Atomic Orbitals (IAOs), as formulated by Gerald Knizia, have been implemented to eliminate the polarization contribution in projection analysis.
The Polarizable Continuum Model (PCM) is available for the inclusion of solvent effects. For more details, see this paper
As a byproduct of the PCM implementation, molecular electrostatic potential (MEP) maps are available for 4-component electronic-structure calculations, see this paper
+Q corrections (size-consistency corrections) for KR-CI calculations
Functional derivatives using automatic differentiation (XCFun)
New visualization options
RKBIMP: MO-coefficients generated using restricted kinetic balance (RKB) can be extended by their unrestricted kinetic balance (UKB) complement, thus providing magnetic balance for response calculations involving external magnetic fields
New and improved 2c Hamiltonian schemes
New build system and infrastructure
New compilation scheme: configure replaced by CMake mechanism
New pam script (python)
Alternative launcher: wrapper.py (python)
New testing framework based on python (runscript)
Many static allocation calls replaced by dynamic allocation; in practice this means that you may need less WORK array memory and/or more space for dynamic allocation compared to DIRAC10.