Strong electron correlation can be captured with multireference wave function methods, but an accurate description of the electronic structure requires accounting for the dynamic correlation, which they miss. In this work, a new approach for the correlation energy based on the adiabatic connection (AC) is proposed. The AC n method accounts for terms up to order n in the coupling constant, and it is size-consistent and free from instabilities. It employs the multireference random phase approximation and the Cholesky decomposition technique, leading to a computational cost growing with the fifth power of the system size. Because of the dependence on only one- and two-electron reduced density matrices, AC n is more efficient than existing ab initio multireference dynamic correlation methods. AC n affords excellent results for singlet–triplet gaps of challenging organic biradicals. The development presented in this work opens new perspectives for accurate calculations of systems with dozens of strongly correlated electrons.

The adiabatic connection (AC) theory offers an alternative to the perturbation theory methods for computing correlation energy in the multireference wavefunction framework. We show that the AC correlation energy formula can be expressed in terms of the density linear response function as a sum of components related to positive and negative parts of the transition energy spectrum. Consequently, generalization of the adiabatic connection fluctuation-dissipation theory to electronically excited states is obtained. The component of the linear response function related to the negative-transition energy enters the correlation energy expression with an opposite sign to that of the positive-transition part and is non-negligible in the description of excited states. To illustrate this, we analyze the approximate AC model in which the linear response function is obtained in the extended random phase approximation (ERPA). We demonstrate that AC can be successfully combined with the ERPA for excited states, provided that the negative-excitation component of the response function is rigorously accounted for. The resulting AC0D model, an extension of the AC0 scheme introduced in our earlier works, is applied to a benchmark set of singlet excitation energies of organic molecules. AC0D constitutes a significant improvement over AC0 by bringing the excitation energies of the lowest excited states to a satisfactory agreement with theoretical best estimates, which parallels or even exceeds the accuracy of the n-electron valence state perturbation theory method. For higher excitations, AC0D is less reliable due to the gradual deterioration of the underlying ERPA linear response.

Diarylethenes (DAEs), promising photochromic molecular switches, undergo pericyclic reactions upon ultraviolet or visible light illumination. For this reason, most studies on DAEs employ UV–vis spectroscopies. However, also their infrared (IR) spectra are valuable, in particular, for understanding the vibrational dynamics which accompanies the relevant photoreactions. An accurate assignment of IR bands to molecular modes can be achieved through a comparison between experimental and computed theoretical spectra. Even though more sophisticated computational methods are available, the density functional theory (DFT) is usually employed for this task, because of its modest cost and versatility. Here, we have tested the ability of several DFT functionals to reproduce the wide-range, 400–3200 cm−1, IR spectra of open and closed isomers of four representative DAE molecules. We find that global and range-separated, corrected for anharmonicity by scaling factors, hybrid DFT functionals are able to reproduce the IR spectra of DAEs, however, instead of the popular B3LYP functional we propose the use of the dispersion-corrected PBE0 functional. The paper also proposes a semi-automatic method of band assignment.