New features in Dalton 1.2

Dalton 1.2 includes several large modules that have been added in order to increase the possibilities for user applications of the code. In addition, several improvements in the general performance of the code has been made relative to Dalton 1.1. The most important new features and improvements are presented here.

1. Coupled-cluster theory: The most important and largest improvement is the addition of a complete integral-direct coupled cluster module, capable of calculating all common coupled cluster models, most of them in an integral-direct manner. Frequency-dependent response functions have also been implemented up to the cubic response functions for the CCS, CC2, and CCSD models. Several models for calculating molecular properties of excited states are also available. Geometry optimization using analytical gradients for electronic ground states using first-order optimization methods is also available.

2. Non-equilibrium solvation Non-equilibrium solvation linear, quadratic, and cubic response functions have been implemented, and this makes for instance solvent shifts on excitation energies directly accessible in a one-step calculation. Solvent effects on non-linear optical processes can also be calculated with this approach.

3. SOPPA The Second-Order Polarization Propagator Approach (SOPPA) has been introduced in the ABACUS module, allowing for convenient and easy calculations of second-order properties such as polarizabilities, indirect spin-spin coupling constants, nuclear shieldings and magnetizabilities, the latter two only without London orbitals.

4. CSF Configurations State Functions can now be used also in property calculations in both the RESPONSE and ABACUS modules, allowing for better control of the spin-state of the reference wave function also in property calculations.

5. AMFI An Atomic Mean-Field Approximation of the spin-orbit integrals have been implemented. Even though this mean-field spin-orbit approximation only involves one-electron integrals, it has proven to be a very good approximation to the full one- and two-electron Breit-Pauli spin-orbit operator.

6. ECP Effective Core Potentials can now be used for calculations of energies and response functions in which the basis set does not depend on the applied perturbations. That is, molecular gradients and Hessians, as well as London orbital-based magnetic properties, can not be calculated with ECPs.

7. Douglas-Kroll The Douglas-Kroll scalar relativistic Hamiltonian has been implemented to second order. It can be used to account for scalar relativistic effects in molecular energies and response functions, and although molecular gradients and Hessians can be calculated, they are not strictly correct, as they use the non-relativistic derivatives of the one-electron Hamiltonian.

8. Linear coupling model The linear coupling model can be used for estimating Franck-Condon factors.

9. Vibrational averaging A new approach for calculating zero-point [14,15] or temperature-dependent [16] vibrational averaged molecular properties. For general properties, this have to be in a two-step procedure in which the first step determines the vibrational averaged geometry, and the second calculates the average of the molecular property over a harmonic oscillator approximation for the different vibrational modes.

10. Magnetic Circular Dichroism The ${\cal{B}}(0\rightarrow f)$ term contributing to magnetic circular dichroism (MCD) has been implemented [17].

11. Two-photon absorption Although possible to calculate with Dalton 1.1, the input required for calculating two-photon absorption has been significantly improved.

12. Solvent geometry optimizations Geometry optimizations using the spherical cavity model can now be done using symmetry. However, only the second-order geometry optimization routines in the *WALK module will be able to do geometry optimizations with the solvent model.

13. File splitting For file systems where the maximum file length is 2 Gb, DALTON will automatically split large files so that calculations can be performed on these systems even though some files may be longer than 2 Gb. Currently the implementation limits the maximum file size to 22 Gb.

14. Restart features in RESPONSE Assuming that the files RESULTS.RSP and RSPVEC are available, the RESPONSE program can now be restarted at any point of a linear, quadratic or cubic response calculation. At the most a few micro-iterations of a solution of a linear set of equations will be lost. One can now also reuse the converged response vectors of a quadratic response calculation in a cubic response vector, making it significantly more computationally efficient to determine both linear, quadratic and cubic response functions using SCF or MCSCF wave functions.