Hi all,
Tuesday at 2pm Alexander Miessen will present his master thesis on "Variational Quantum Algorithms for Real Space Wave Packet Dynamics", which he carried out with Ivano Tavernelli at IBM Research. See below for the abstract. The zoom link is https://ethz.zoom.us/j/362994444.
Note that at 3pm Andru Gheorghiu will deliver the ITS Fellows' Seminar on "Quantum computation: hardness, verification and protocols for near-term devices". See https://eth-its.ethz.ch/activities/its-fellows--seminar/Alexandru-Gheorghiu.... for more details. The zoom link is https://ethz.zoom.us/j/99922681517.
Best,
Joe
Abstract: Variational quantum algorithms (VQAs) have become the new paradigm in applied quantum computing, enabling calculations on current-day quantum hardware that would be far out of reach otherwise. The triumph of hybrid quantum-classical schemes over purely-quantum methods has led to numerous further-developments and applications to stationary problems in recent years, yet the most widely used method for quantum dynamics simulation remains the Trotterization of the unitary time-evolution operator. However, it relies on deep quantum circuits and is therefore hampered by the strong limitations of available quantum technology which prohibits its use for practical tasks in the near future. The development of a variational time-evolution (VTE) algorithm was therefore an important and necessary s tep for quantum dynamics to gain more foothold in present-day quantum computing. Nonetheless, VTE faces the difficulty of finding an ansatz for expressing the wavefunction which is able to maintain satisfactory accuracy throughout an evolution without diverging from the true state trajectory. While heuristic ansätze enjoy popularity, findings suggest that they require exponentially many parameters to offer enough flexibility, jeopardizing any potential quantum advantage. One hypothesized way out of this dilemma is the use of a physically motivated ansatz, the quantum circuit analog of an educated guess, which samples from a reduced variational space by taking into consideration model-specific symmetries. In this work, we aim to explore this possibility by studying the dynamics of molecular vibrations as well as a spin-boson model. We analyze existing physically motivated ansätze, developed for stationary VQAs, and propose modifications thereof to make them better suited for VTE. Subsequently, they are tested in the simulation of various system's dynamics. Our results indicate that it is indeed possible to find a compact yet accurate variational ansatz by leveraging symmetries of the system under study, requiring far less than exponentially many parameters.