We investigate a simple approach to compute a non-adiabatic thermal rate constant using the fewest switches surface hopping (FSSH) dynamics. We study the effects of both decoherence (using our augmented-FSSH (A-FSSH) algorithm) and forbidden hops over a large range of parameters, including high and low friction regimes, and weak and strong electronic coupling regimes. Furthermore, when possible, we benchmark our results against exact hierarchy equations of motion results, where we usually find a maximum error of roughly a factor of two (at reasonably large temperatures). In agreement with Hammes-Schiffer and Tully, we find that a merger of transition state theory and surface hopping can be both accurate and efficient when performed correctly. We further show that detailed balance is followed approximately by A-FSSH dynamics.

We have benchmarked the surface hopping method to capture nuclear quantum effects in the spin-Boson model in the deep tunneling regime. The thermal populations and the rate constants calculated using the surface hopping method are compared with those calculated using Boltzmann theory and Fermi's golden rule, respectively. Additionally, we have proposed a simple kinetic model that partially includes nuclear quantum effects within Marcus theory, and the results of the surface hopping method are analyzed under the framework of this simple kinetic model. A broad range of parameters are investigated to identify the regimes for the successes and failures of the surface hopping method. This work shows that with the accurate treatment of decoherence and velocity reversal, surface hopping can generally capture the nuclear quantum effects in the deep tunneling and weak diabatic coupling regime.

We employ surface hopping trajectories to model the short-time dynamics of gas-phase and partially solvated 4-(N,N-dimethylamino)benzonitrile (DMABN), a dual fluorescent molecule that is known to undergo a nonadiabatic transition through a conical intersection. To compare theory vs time-resolved fluorescence measurements, we calculate the mixed quantum-classical density matrix and the ensemble averaged transition dipole moment. We introduce a diabatization scheme based on the oscillator strength to convert the TDDFT adiabatic states into diabatic states of L and L character. Somewhat surprisingly, we find that the rate of relaxation reported by emission to the ground state is almost 50% slower than the adiabatic population relaxation. Although our calculated adiabatic rates are largely consistent with previous theoretical calculations and no obvious effects of decoherence are seen, the diabatization procedure introduced here enables an explicit picture of dynamics in the branching plane, raising tantalizing questions about geometric phase effects in systems with dozens of atoms.

We benchmark and improve the independent electron surface-hopping (IESH) method developed by J. C. Tully’s group for nonadiabatic simulations near metal surfaces. We have incorporated decoherence within the IESH method as well as implemented a scheme for the accurate calculation of diabatic populations. We benchmark the original IESH method with the above inclusions for a model system to calculate rate constants and long-time populations. The original IESH method fails to capture the detailed balance for some of the parameters, which is corrected with the inclusion of decoherence and accurate calculation of diabatic populations. Total rate constants are well captured both within the original IESH method as well as within our modified IESH.

We report the implementation of a hierarchical equations of motion (HEOM) module within the open-source Libra software. It includes the standard and scaled HEOM algorithms for computing the dynamics of open quantum systems interacting with a harmonic bath. The module allows the computing of the evolution of the reduced density matrix, as well as spectral lineshapes. The truncation, filtering, and "update list" schemes, as well as OpenMP parallelization, allow for further computational saving. The package is written in a mix of C++ and Python languages, delivering the best compromise between user friendliness and efficiency. The Python layer of the package takes advantage of standard Python libraries, such as h5py, which allows efficient storage and retrieval of the generated results. The package can be seamlessly used within Jupyter notebooks; its careful design shall provide the maximal convenience and intuitiveness to its users.