Circuit Design for A Measurement-Based Quantum Carry-Lookahead Adder

TL;DR

This paper designs and evaluates a measurement-based quantum carry-lookahead adder (MBQCLA), demonstrating it is significantly faster than ripple-carry adders for large registers by leveraging MBQC's rapid state transfer, despite higher resource costs.

Contribution

It introduces a measurement-based quantum carry-lookahead adder (MBQCLA) that achieves logarithmic depth, surpassing linear-depth ripple adders, and analyzes its resource requirements and optimization strategies.

Findings

MBQCLA has logarithmic depth, faster than ripple-carry adders for large registers.

MBQCLA reduces addition latency beyond nearest neighbor constraints.

Resource analysis shows trade-offs between speed and cluster state size.

Abstract

We present the design and evaluation of a quantum carry-lookahead adder (QCLA) using measurement-based quantum computation (MBQC), called MBQCLA. QCLA was originally designed for an abstract, concurrent architecture supporting long-distance communication, but most realistic architectures heavily constrain communication distances. The quantum carry-lookahead adder is faster than a quantum ripple-carry adder; QCLA has logarithmic depth while ripple adders have linear depth. MBQCLA utilizes MBQC's ability to transfer quantum states in unit time to accelerate addition. MBQCLA breaks the latency limit of addition circuits in nearest neighbor-only architectures : compared to the $\Theta(n)$ limit on circuit depth for linear nearest-neighbor architectures, it can reach $\Theta(log n)$ depth. MBQCLA is an order of magnitude faster than a ripple-carry adder when adding registers longer than 100 qubits, but requires a cluster state that is an order of magnitude larger. The cluster state resources can be classified as computation and communication; for the unoptimized form, $\approx$ 88 % of the resources are used for communication. Hand optimization of horizontal communication costs results in a $\approx$ 12% reduction in spatial resources for the in-place MBQCLA circuit. For comparison, a graph state quantum carry-lookahead adder (GSQCLA) uses only $\approx$ 9 % of the spatial resources of the MBQCLA.