Extended software development workshop in quantum dynamics
Location: Durham University
Organisers
Quantum molecular dynamics simulations are pivotal to understanding and predicting the microscopic details of molecules, and strongly rely on a combined theoretical and computational effort. When considering molecular systems, the complexity of the underlying equations is such that approximations have to be devised, and the resulting theories need to be translated into algorithms and computer programs for numerical simulations. In the last decades, the joint effort of theoretical physicists and quantum chemists around the challenges of quantum dynamics made it possible to investigate the quantum dynamics of complex molecular systems, with applications ranging from energy conversion, energy storage, organic electronics, light-emitting devices, biofluorescent molecules, or photocatalysis, to name a few.
Two different strategies have been successfully applied to perform quantum molecular dynamics: wavepacket propagation or trajectories. The first family of methods includes all quantum nuclear effects, but their computational cost hampers the simulation of systems with moderate number of more than 10-12 degrees of freedom. The method coined multi-configuration time-dependent Hartree (MCTDH) constitutes one of the most successful developments in this field and is often considered as a gold standard for quantum dynamics [1]. Other strategies for wavepacket propagation try to identify procedures to optimize the “space” where the wavefunction information is computed, such that Cartesian grids can be replaced with Smolyak grids [2]. The second family of methods introduces the idea of trajectories as a way to approximate the nuclear subsystem, either classically or semiclassically, and is exemplified by methods like the trajectory surface hopping and Ehrenfest schemes [3], or the more accurate methods coupled-trajectory mixed quantum-classical (CT-MQC) [4] and quantum-classical Liouville equation (QCLE) [5].
From a computational perspective, both families of methods require extensive electronic structure calculations, as the nuclei move under the effect of the electronic subsystem, either “statically” occupying its ground state or “dynamically” switching between excited states. Solving the quantum nuclear dynamics equations also becomes in itself very expensive in the case of wavepacket propagation methods. Contrary to other, more consolidated, areas of modeling, quantum dynamics simulations do not benefit from established community packages and most of the progress occurs based on in-house codes, difficult to maintain and with limits in optimization and portability. One of the core actions of E-CAM has been to seed a change in this situation, by promoting systematic developments of software, providing a repository to host and share code, and fostering collaborations on adding functionalities and improving the performance of common software scaffolds for wavepacket (Quantics) and trajectory-based (PaPIM) packages. Collaborations on developments on other codes have also been initiated. This workshop aims at continuing and extending these activities based on input from the community.
References
Federica Agostini (Université Paris-Sud) - Organiser
United Kingdom
Basile Curchod (University of Bristol) - Organiser
Graham Worth (University College London) - Organiser