We discuss the physical nature of quantum information, in particular focussing on tasks that are achievable by some physical realizations of qubits but not by others.
We survey progress in understanding quantum information in terms of equivalences, reducibilities, and asymptotically achievable rates for transformations among nonlocal resources such as classical communication, entanglement, and particular quantum states or channels. In some areas, eg source coding, there are straightforward parallels to classical information theory; in others eg entanglement-assisted communication, new effects and tradeoffs appear that are beginning to be fairly well understood, or the remaining uncertainty has become focussed on a few simple open questions, such as conjectured additivity of the Holevo capacity. In still other areas, e.g. the role of the back communication and the classification of tripartite entanglement, much remains unknown, and it appears unlikely that an adequate description exists in terms of a finite number of resources.
Quantum reservoir computing is a type of machine-learning scheme that exploits the nonlinearity of natural quantum dynamics for temporal information processing. In a recent study of quantum reservoir computing, an artificially-engineered reservoir is introduced to model the system, which is implemented on noisy gate-base quantum computer to test its performance. However, the proposed system has difficulty in physical implementation from a practical perspective. In this study, we propose a new quantum reservoir computing scheme based on natural physical reservoir on a superconducting quantum device; this system makes full use of the realistic (un-modeled) noise that can couple neighboring qubits. Here, we utilize the IBM superconducting quantum processors to demonstrate the performance for a sequential data processing task; an emulation of the Non-Linear Auto-Regressive Moving Average dynamics (NARMA task). As a result, we observed our proposed model outperformed a simple classical linear regression. This result indicates that a natural physical quantum reservoir realized on a noisy quantum computer is a promising candidate for practical temporal information processing.
Scalable quantum information systems would store, manipulate, and transmit quantum information locally and across a quantum network, but no single qubit technology is currently robust enough to perform all necessary tasks. Defect centers in solid-state materials have emerged as potential intermediaries between other physical manifestations of qubits, such as superconducting qubits and photonic qubits, to leverage their complementary advantages. It remains an open question, however, how to design and to control quantum interfaces to defect centers. Such interfaces would enable quantum information to be moved seamlessly between different physical systems. Understanding and constructing the required interfaces would, therefore, unlock the next big steps in quantum computing, sensing, and communications. In this Perspective, we highlight promising coupling mechanisms, including dipole-, phonon-, and magnon-mediated interactions, and discuss how contributions from nanotechnologists will be paramount in realizing quantum information processors in the near-term.