Please post any questions related to Q-Chem Webinar 87 (Improvements in the OO-DFT user interface), presented by Juan Arias Martinez on Apr 23 2026, in this thread.
Abstract: Orbital-optimized density functional theory (OO-DFT) has emerged as an appealing alternative to time-dependent (TD)DFT for describing charge-transfer (CT) excited states at a tractable computational cost. Indeed, the performance of OO-DFT for CT excitations is often superior to the much more complex and expensive equation-of-motion coupled-cluster singles and doubles (EOM-CCSD). However, as any user is likely to tell you, OO-DFT has been notoriously difficult to work with. Aside from electronic structure challenges — namely an often-turbulent SCF convergence — this is due to the lack of a robust user-facing infrastructure capable of conveniently exploiting all the OO-DFT capabilities available in QChem.
In this Webinar, I will primarily introduce you to a new driver that provides a simplified way to run and analyze OO-DFT calculations; in addition, I will share how suppressing the mixing between the singly-occupied molecular orbitals (SOMOs) often dramatically improves the convergence behavior of restricted open-shell Kohn-Sham (ROKS) theory — a form of OO-DFT. The new driver can perform open-shell singlet, doublet, and triplet calculations to be carried out with a single input file, allowing for independent customization for each of the SCF guess orbitals, algorithms, and convergence thresholds, while providing sensible defaults depending on the desired type of excitation. Crucially, the driver computes measures of similarity between the SCF guess and the relaxed orbitals to reveal whether the optimization converged to the state associated with the guess, as well as transition dipole moments with the ground state to compute spectra. Finally, it allows for printing of cube files of the key (open-shell) orbitals involved in the excitation for visualization. To wrap up I will share how, armed with much smoother ROKS convergences and an improved user interface for OO-DFT calculations, I have taken on computing organic donor-acceptor CT excitations in a previously inaccessible scale of throughput. These advances hope to connect the accuracy of OO-DFT with the blossoming use of machine learning in computational chemistry.
About the presenter: Juan E. Arias Martinez received his Ph.D in Chemistry from UC Berkeley in 2023, working under Prof. Martin Head-Gordon. The core of Juan’s graduate work revolved around exploring orbital-optimized SCF references as a starting point for accurate computations of X-ray absorption spectroscopy (XAS) signals. An exposure to the advances in experimental XAS techniques inspired his keystone project: developing a model capable of predicting the XAS signatures of the excited-states sought after in femtosecond time-resolved XAS experiments. Currently, he is postdoctoral researcher in Laura McCaslin’s group at Sandia National Laboratories, working on designing new optoelectronic materials based on organic donor-acceptor charge-transfer complexes.