An Abstraction Hierarchy Toward Productive Quantum Programming

TL;DR

This paper proposes an abstraction hierarchy for quantum programming to improve software engineering, analyzing current overlaps in models and demonstrating its application to eigenvalue estimation problems.

Contribution

It introduces a structured abstraction hierarchy for quantum software, highlighting overlaps in models and guiding future development of scalable quantum programming tools.

Findings

Identifies overlaps in quantum programming models

Demonstrates hierarchy application to eigenvalue estimation

Highlights conceptual challenges in quantum software engineering

Abstract

Experience from seven decades of classical computing suggests that a sustainable computer industry depends on a community of software engineers writing programs to address a wide variety of specific end-user needs, achieving both performance and utility in the process. Quantum computing is an emerging technology, and we do not yet have the insight to understand what quantum software tools and practices will best support researchers, software engineers, or applications specialists. Developers for today's quantum computers are grappling with the low-level details of the hardware, and progress towards scalable devices does not yet suggest what higher-level abstractions may look like. In this paper, we analyze and reframe the current state of the quantum software stack using the language of programming models. We propose an abstraction hierarchy to support quantum software engineering and discuss the consequences of overlaps across the programming, execution, and hardware models found in current technologies. We exercise this hierarchy for solving the eigenvalue estimation problem in two ways (a variational algorithm with error mitigation, and phase estimation with error correction) and pinpoint key differences in these approaches in terms of these layered models and their overlaps. While our work points to concrete conceptual challenges and gaps in quantum programming and proposes some specific steps forward, our primary thesis is that progress hinges on thinking about the abstraction hierarchy holistically, and not just about its components.