The Munich Quantum Software Stack: Connecting End Users, Integrating Diverse Quantum Technologies, Accelerating HPC

TL;DR

The paper addresses the need for a unified, modular software stack that can bridge diverse quantum hardware with classical HPC to enable practical hybrid quantum-classical workflows. It introduces the Munich Quantum Software Stack (MQSS), an open-source, multi-layer platform featuring a QDMI back-end interface, a QRM&CI runtime with an MLIR-based compiler, FoMaC libraries for hardware-aware optimization, and front-end adapters connected to popular quantum frameworks. Production deployment at LRZ demonstrates tight, on-premise HPCQC integration and a end-to-end workflow from problem specification to quantum execution and results, validated by an early 13-qubit QM/MM study achieving near-chemical accuracy. Collectively, MQSS offers a hardware-agnostic, community-driven path toward scalable, production-ready QC acceleration within classical HPC, with a clear roadmap toward fault-tolerant architectures and ongoing ecosystem growth.

Abstract

Quantum computing is advancing rapidly in hardware and algorithms, but broad accessibility demands a comprehensive, efficient, unified software stack. Such a stack must flexibly span diverse hardware and evolving algorithms, expose usable programming models for experts and non-experts, manage resources dynamically, and integrate seamlessly with classical High-Performance Computing (HPC). As quantum systems increasingly act as accelerators in hybrid workflows -- ranging from loosely to tightly coupled -- few full-featured implementations exist despite many proposals. We introduce the Munich Quantum Software Stack (MQSS), a modular, open-source, community-driven ecosystem for hybrid quantum-classical applications. MQSS's multi-layer architecture executes high-level applications on heterogeneous quantum back ends and coordinates their coupling with classical workloads. Core elements include front-end adapters for popular frameworks and new programming approaches, an HPC-integrated scheduler, a powerful MLIR-based compiler, and a standardized hardware abstraction layer, the Quantum Device Management Interface (QDMI). While under active development, MQSS already provides mature concepts and open-source components that form the basis of a robust quantum computing software stack, with a forward-looking design that anticipates fault-tolerant quantum computing, including varied qubit encodings and mid-circuit measurements.

Figures (8)

• Figure 1: a) Traditional view: many users, many systems, many stacks. b) Vision going forward: decoupling front-end and back-end in a single stack via shared IRs.
• Figure 2: Overview of the MQSS and its components, connecting end-users and their high-level problem descriptions (left) to a diverse set of quantum hardware back-ends, as well as simulators (right).
• Figure 3: Structure of the MQSS back-end, featuring the implementation of the QDMI interface as well as a series of system device plugins.
• Figure 4: The three main components of the QDMI: Session Management, Job Submission, and Query Interface.
• Figure 5: The structure of the MQSS middle-end, featuring the QRM&CI and its internal components implementing the compiler.