An HPC-Inspired Blueprint for a Technology-Agnostic Quantum Middle Layer

TL;DR

The paper addresses the fragmentation of quantum software stacks by introducing a backend-agnostic, HPC-inspired quantum middle layer composed of four orthogonal artifact classes: quantum data type descriptors, quantum operator descriptors, context descriptors, and algorithmic libraries. These artifacts enable late binding to diverse backends (e.g., gate-based and annealing) and support explicit cost hints and decoding schemas, demonstrated via a Max-Cut benchmark implemented with both QAOA on Qiskit Aer and Ising/BQM on D-Wave Ocean. A key result is that the same typed problem can be realized across heterogeneous hardware by varying only the operator formulation and the execution context, with the Ising energy expressed as $E(s)=\sum_i h_i s_i + \sum_{i<j} J_{ij} s_i s_j$ guiding annealing. The approach promises portability, composability, and extensibility, and points toward production paths using MLIR dialects or protobuf-based IRs to support broader hardware modalities while preserving a minimal core.

Abstract

We present a blueprint for a quantum middle layer that supports applications across various quantum technologies. Inspired by concepts and abstractions from HPC libraries and middleware, our design is backend-neutral and context-aware. A program only needs to specify its intent once as typed data and operator descriptors. It declares what the quantum registers mean and which logical transformations are required, without committing to gates, pulses, continuous-variable routines, or anneal backend. Such execution details are carried separately in a context descriptor and can change per backend without modifying the intent artifacts. We develop a proof of concept implementation that uses JSON files for the descriptors and two backends: a gate-model path realized with IBM Qiskit Aer simulator and an annealing path realized with D-Wave Ocean's simulated annealer. On a Max-Cut problem instance, the same typed problem runs on both backends by varying only the operator formulation (Quantum Approximated Optimization Algorithm formulation vs. Ising Hamiltonian formulation) and the context. The proposed middle layer concepts are characterized by portability, composability, and its minimal core can evolve with hardware capabilities.

Figures (3)

• Figure 1: A diagram of a backend-neutral context-aware quantum middle layer.
• Figure 2: Diagram of the workflow for solving the Max-Cut on IBM Qiskit Aer state vector simulator.
• Figure 3: Diagram of the workflow for solving the Max-Cut on D-Wave Ocean Neil simulator.