Weiterentwicklung der Hybrid Data Management Architecture zur plattformübergreifenden Nutzung von Quantencomputing Cloudservices und Simulatoren

Abstract

There is a broad consensus within the quantum computing community that quantum applications will primarily be hybrid, with classical computers handling general purpose computations efficiently while specialized operations are executed on quantum hardware. Against this background, the present work aims to further develop the existing Hybrid Data Management Architecture (HDMA) into a platform independent service architecture, enabling the execution of quantum circuits both on cloud-based quantum hardware and on local simulators. A key focus is on reducing the migration effort for new cloud service providers (CSPs). The theoretical framework is based on concepts from Service-Oriented Architecture (SOA) as well as established design patterns in quantum software engineering, which are analyzed through a systematic literature review and evaluated using an Implementation Maturity Level (IML) model. Accordingly, existing concepts and reference architectures (e.g. Qunicorn, Qubernetes, Quantum API Gateway) in the field of hybrid quantum applications are first examined. Based on this analysis, a modular architectural concept is developed, integrating provider-specific backend services, an asynchronous message broker (RabbitMQ) for inter-service communication, and a central monitor service for status tracking. Data from monitored quantum jobs are persistently stored in MongoDB. Local simulators are deployed in dedicated, scalable service containers. The implementation is containerbased using Docker Compose and allows dynamic configuration of the backends via YAML files. The prototype is validated using the graph coloring problem, executed on cloud backends from IBM Quantum and Azure Quantum, a local Azure Sparse Simulator and a local Aer Simulator. The results demonstrate that the HDMA reliably supports execution across all integrated backends and additionally enables the parallel execution of quantum jobs on multiple quantum backends. In conclusion, the study demonstrates that a combination of modular service isolation, a generic backend API, and event-driven communication can realize a scalable and extensible platform for hybrid quantum applications. The work provides practical insights for the development of platformindependent quantum computing infrastructures.