SKM 2023 – wissenschaftliches Programm
DY 18.5: Vortrag
Dienstag, 28. März 2023, 11:15–11:30, ZEU 148
On the Correlation of Functionality and Lyapunov Stability in Oscillator-based Ising machines — •Bakr Al Beattie1, Maximiliane Noll2, Hermann Kohlstedt2, and Karlheinz Ochs1 — 1Ruhr-Universität Bochum, Lehrstuhl für digitale Kommunikationssysteme, Bochum, Deutschland — 2Christian-Albrechts-Universität zu Kiel, Lehrstuhl für Nanoelektronik, Kiel, Deutschland
Oscillator-based Ising machines are a promising analog approach for dealing with combinatorial optimization problems that are classified as NP (nondeterministic polynomial). The idea is to mimic the Ising model by coupling electrical oscillators that behave like the spins of the Ising model. Here, the coupling should somehow map the Ising Hamiltonian onto the energy of electrical system. With this contribution, we demonstrate numerical evidence demonstrating the correlation between the Ising machine*s functionality and stability. We make use of the well-known Kuramoto model to describe a coupled oscillator network and show stability to be the key property that makes an Ising machine solve optimization problems. Furthermore, we give an answer to the question: when has an Ising machine finished solving a mapped problem?