qcscine/utilities: Release 4.0.0

• Main Authors: Bosia, Francesco, Brunken, Christoph, Csizi, Katja-Sophia, Grimmel, Stephanie A., Gugler, Stefan, Haag, Moritz P., Heuer, Michael A., Müller, Charlotte H., Polonius, Severin, Simm, Gregor N., Sobez, Jan-Grimo, Steiner, Miguel, Türtscher, Paul L., Unsleber, Jan P., Vaucher, Alain C., Weymuth, Thomas, Reiher, Markus
• Format: Software
• Language: unknown
• Published: 2021
• Subjects: Bofill; Davidson; Orca
• Online Access: https://zenodo.org/record/5782808; https://doi.org/10.5281/zenodo.5782808

Changes: Improve GDIIS: numerical stability Improve BFGS: add automatic damping Improve EVF and Bofill: allows to select mode and follows mode independent of order Add periodic boundary conditions Add support for CP2K Improve Gaussian interface: allows to reuse SCF results as guesses in subsequent calculations and to retrieve molecular orbital coefficients Add support for Turbomole Improve MD: Fix a bug regarding the time step size, check gradient calculations for SCF convergence, add the option to use bias potentials and add a stochastic dynamics integrator Add Python bindings for CalculatorWithReference Add Log accessor to Python bindings of Calculator and CalculatorWithReference Add Python bindings for Davidson diagonalizer with possibility of having custom sigma vector evaluators/preconditioners Add functions to get all closest atoms within a certain distance and to build an atom pair-list Distinguish now between true internal, true Cartesian, and Cartesian with removed translation and rotation coordinate systems Add Python bindings for ThermochemistryCalculator to calculate thermodynamic properties from a Hessian in Python Add Python bindings for SettingsNames Add support for the SMD solvation model in ORCA Add the option to obtain gradients from a CalculatorWithReference in MD simulations Refactoring of GeometryUtilities into sub-namespaces Add data structures needed for downstream methods that are general to linear response methods Remove Logger option for downstream LcaoMethods as it can be accessed through the calculator interface Refactor Davidson diagonalizers: Create IterativeDiagonalizer interface Create KrylovDiagonalizer interface, inheriting from IterativeDiagonalizer Create the 2 versions, NonOrthogonalDavidson and OrthogonalDavidson Add Python bindings for OrthogonalDavidson and NonOrthogonalDavidson, tested in Python and added an example on how to extend the SigmaVectorEvaluator to customize the Davidson directly in Python