An Optical Interconnect for Modular Quantum Computers (2412.09299v2)

Abstract: Much like classical supercomputers, scaling up quantum computers requires an optical interconnect. However, signal attenuation leads to irreversible qubit loss, making quantum interconnect design guidelines and metrics different from conventional computing. Inspired by the classical Dragonfly topology, we propose a multi-group structure where the group switch routes photons emitted by computational end nodes to the group's shared pool of Bell state analyzers (which conduct the entanglement swapping that creates end-to-end entanglement) or across a low-diameter path to another group. We present a full-stack analysis of system performance, a combination of distributed and centralized protocols, and a resource scheduler that plans qubit placement and communications for large-scale, fault-tolerant systems. We implement a prototype three-node switched interconnect and create two-hop entanglement with fidelities of at least 0.6. Our design emphasizes reducing network hops and optical components to simplify system stabilization while flexibly adjusting optical path lengths. Based on evaluated loss and infidelity budgets, we find that moderate-radix switches enable systems meeting expected near-term needs, and large systems are feasible. Our design is expected to be effective for a variety of quantum computing technologies, including ion traps and superconducting qubits with appropriate wavelength transduction.

• The paper introduces the Q-Fly architecture, a hierarchical, group-based network that minimizes optical losses and errors in quantum systems.
• It evaluates three network configurations achieving entanglement fidelities between 0.60 and 0.64, demonstrating viable photon-based quantum interconnects.
• The architecture shows promising scalability with performance projections closely matching optimal monolithic systems using quantum Fourier transform applications.

An Optical Interconnect for Modular Quantum Computers: A Detailed Exploration

This paper introduces a novel approach to optical interconnect networks specifically tailored for modular quantum computers, drawing inspiration from classical Dragonfly network topologies. The authors propose a system referred to as "Q-Fly," which aims to enhance scalability and efficiency for quantum computing systems projected to require millions of qubits. Building on the inherent challenges presented by the no-cloning theorem and the probabilistic nature of entanglement operations, the paper provides a comprehensive design and evaluation of a scalable interconnect architecture for quantum systems.

Key Contributions of the Paper

1. Q-Fly Architecture Design: The paper presents the Q-Fly topology as a viable network structure for large-scale quantum computers, with a focus on minimizing optical losses and errors while providing low-latency communication between distributed quantum nodes. The architecture leverages a group-based hierarchical structure, avoiding the substantial hop counts characteristic of more traditional tree-based architectures like fat trees.
2. Network Variants and Analysis: Three distinct Q-Fly network configurations are proposed: single-path quasi-half duplex, dual-path quasi-half duplex, and dual-path quasi-full duplex. Each configuration balances different priorities such as implementation complexity, switch radix, and overall network fidelity. The designs allow for full intra-group communication and flexible inter-group connectivity, with experimental results demonstrating acceptable fidelity levels for end-to-end entanglement.
3. Experimental Demonstration: Implementing a small-scale prototype comprising three end nodes and a switchable Bell state analyzer (BSA), the authors achieve entanglement fidelities between 0.60 and 0.64. This result supports the practicality of their architecture for photon-based quantum networks, albeit with further improvements required for scaling up.
4. Performance Projections: The paper estimates the scalability of Q-Fly using the quantum Fourier transform (QFT) as a test application underpinned by the surface code error correction model. Comparative analyses suggest that Q-Fly networks could deliver execution times substantially higher than a 2D lattice topology and within a factor of 4.5 of theoretically optimal monolithic quantum computers.

Theoretical and Practical Implications

The Q-Fly architecture extends the potential for distributed and modular quantum computing, addressing significant barriers to the construction of networks with realistic levels of interconnect loss and qubit fidelity. From a theoretical perspective, the architecture underscores the necessity of re-evaluating conventional topology designs to meet the unique requirements of quantum information systems. Practically, it moves the quantum computing community closer to realizing robust multi-node systems capable of executing complex quantum algorithms distributed across many physical devices.

Future Directions

Considering potential future developments, the paper lays the groundwork for further exploration into the interplay between network architecture and quantum error correction. Subsequent research may focus on integrating the Q-Fly topology with emerging quantum memory technologies to prolong qubit coherence. Additionally, the design of higher-layer protocols that leverage the unique capabilities of optical interconnects could lead to enhanced performance in executing non-trivial quantum circuits.

Overall, this paper provides a valuable contribution to both the understanding and practical engineering of next-generation quantum networks, inviting further research into optimizing and implementing scalable interconnects for comprehensive quantum computing platforms.