Quantum 2025 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
THU: Thursday Contributed Sessions
THU 6: Quantum Computing and Communication: Contributed Session II (Concepts)
THU 6.2: Vortrag
Donnerstag, 11. September 2025, 14:30–14:45, ZHG007
On the convergence of the variational quantum eigensolver and quantum optimal control — •Marco Wiedmann1, Daniel Burgarth1, Gunther Dirr2, Thomas Schulte-Herbrüggen3,4, Emanuel Malvetti3,4, and Christian Arenz5 — 1Friedrich-Alexander-Universität Erlangen-Nürnberg, 91058 Erlangen, Germany — 2Julius-Maximilians-Universität Würzburg, 97074 Würzburg, Germany — 3Technische Universität München, 85748 Garching, Germany — 4Munich Center for Quantum Science and Technology (MCQST) & Munich Quantum Valley (MQV), 80799 München, Germany — 5Arizona State University, Tempe, AZ 85281, USA
Variational algorithms have gained a lot of attention in the recent years as a potential application of quantum computers. In broad terms, a parameterized unitary is implemented on a quantum computer, which is then used to measure some objective function that should be minimized by a classical optimization routine.
Gradient based optimizers can however get stuck at singular points of the parameterization, which resembles a gimbal lock like effect. We show that some popular parameterizations do indeed admit these singular points and propose alternatives which are globally regular. Finally, we use these parameterizations to prove that if the Variational Quantum Eigensolver does not run off to infinity, it almost always converges to a true ground state of the problem Hamiltonian.
Keywords: Variational Quantum Algorithms; Differential Geometry; Convergence of gradient algorithms; Optimization; Singular points