SKM 2023 – scientific programme

Parts | Days | Selection | Search | Updates | Downloads | Help

DY: Fachverband Dynamik und Statistische Physik

DY 9: Quantum Dynamics, Decoherence and Quantum Information

DY 9.10: Talk

Monday, March 27, 2023, 16:30–16:45, MOL 213

The Influence of Dynamical Phases on a Quantum Processor based Reservoir Computer — •Brecht Donvil, Mirko Rossini, Dominik Maile, and Joachim Ankerhold — ICQ and IQST, University of Ulm, Ulm, Germany

Reservoir computing is subbranch of machine learning where a physical system, the reservoir, is used to perform computational tasks instead of a large neural network which has to be trained. In the last years, quantum systems systems are being explored as potential reservoirs[1,2,3]. While most of the research has focussed on Ising spin systems, recently it was shown that also quantum processors can serve as reservoirs. The authors of [3] illustrated this fact by successfully implementing reservoir computing on an IMBQ quantum processor.

The dynamical phase of the reservoir can influence its performance on certain information processing tasks. For example, the authors of [2] found that the transverse-field Ising model performs best on examples of memory related tasks on the edge of the ergodic phase. The work I present here concerns quantum processor based reservoir computing as proposed in [3]. I consider simple circuit layout which is know to exhibit a dynamical phase transition between a localised and ergodic phase. I show the influence of the dynamical phase of the circuit on its information processing capacity and its performance as a readout device.

[1] K. Fujii and K. Nakajimna, Phys. Rev. Applied 8, 024030 (2017) [2] R. Martínez-Peña et al., Phys. Rev. Lett. 127, 100502 (2021)

[3] J. Chen et al., Phys. Rev. Applied 14, 024065 (2020)