RESERVOIR COMPUTING AS A GENERAL FRAMEWORK FOR A COMPARATIVE STUDY OF CLASSICAL AND QUANTUM INFORMATION PROCESSING
Abstract
The overarching goal of this proposal is to develop a new paradigm for quantum information processing and re-examine computing with quantum hardware in the context of an analog implementation. Towards this end, we will examine analog information processing by a network of coupled non-linear oscillators, implementable with existing Josephson junction based technology, where the computational task is encoded in its continuous dynamical evolution. The computational task we will consider will be a hardware-based implementation of an appropriately modified version of a deep learning framework based on Recurrent Neural Networks, called Reservoir Computing. A well-defined signal processing task together with a dynamical system that can be tuned across the classical-to-quantum boundary by the choice of operating parameters will provide the basis to undertake a comparative study of information processing by a dynamical system operated in the classical or quantum regime. Results obtained from this project will reveal key aspects of the quantum dynamics of driven-dissipative many body systems across the classical to quantum crossover, and generally how complex information processing capacity arises under conditions far from equilibrium. The results and the conceptual information theoric framework established has the potential to pave the path to the quantification of the full computational potential of existing quantum hardware operated in the analog mode. The outcomes have the potential to drive new technologies for sensing, communication and computing, of key interest to DoD missions, using fast and robust analog quantum hardware.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 12, 2021
- Source ID
- FA95502010177
Entities
People
- Hakan E. Türeci
Organizations
- Air Force Office of Scientific Research
- Trustees of Princeton University
- United States Air Force