A Practical Open-Source Software Stack for a Cloud-Based Quantum Computing System

This slide deck summarizes our QCE2025 presentation on OQTOPUS (Open Quantum Toolchain for Operators and Users), an open-source, end-to-end software stack for cloud-based quantum computing. OQTOPUS integrates unified transpilation, backend execution, calibration, monitoring, and job orchestration to support scalable and reproducible operation of superconducting quantum processors. The system design, key capabilities, and deployment experience are covered, demonstrating how open infrastructure can accelerate practical quantum cloud services.

Published work: QCE2025 IEEE paper — https://ieeexplore.ieee.org/document/11250353

Transcript

1. https://github.com/oqtopus-team A Practical Open-Source Software Stack for a Cloud-Based Quantum Computing System Kakuko*, Gokita*, Masumoto†, Matsumoto‡, Miyaji†, Miyanaga†, Mori†, Nakayama‡, Sasada§, Takamiya§, Tsukano†, Uchida‡, Yamaguchi* IEEE QCE2025/QSYS315 September 5, 2025 *Fujitsu LTD. , †The University of Osaka, ‡Systems Engineering Consultants Co.,LTD., §TIS Inc. Check it out our software! 1 © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc.
2. https://github.com/oqtopus-team Outline ⚫Introduction ⚫Background & Motivation ⚫System Overview ⚫Frontend, Cloud, Backend, Operation ⚫Conclusion ⚫Summary & Outlook 2 © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc.
3. https://github.com/oqtopus-team ⚫ Quantum computers(QCs) have attracted attention from academia and industry for many years due to their ability beyond classical computers. ⚫ IBM launched the first cloud-based quantum computer in 2016. ⚫ AWS Braket followed in 2019, and Azure Quantum in 2020. Background QCs have become commercially available, and the number of services that users can run remotely via the cloud is increasing. 3 Materials Drug Discovery Finance, etc. © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc.
4. https://github.com/oqtopus-team Background ⚫ In Japan, RIKEN, Fujitsu, and the University of Osaka(UOsaka) have taken the lead in releasing 64-qubit superconducting QCs as cloud services. ⚫ RIKEN and Fujitsu develop world-leading 256-qubit QC in Apr. 2025. RIKEN(Sep. 2023) Fujitsu(Oct. 2023) UOsaka (Dec. 2023) RIKEN and Fujitsu (Apr. 2025) Growing availability of such services accelerates research on QC. 4 © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc. ©RQC ©RQC ©RQC ©UOsaka
5. https://github.com/oqtopus-team Motivation ⚫ Implementation of open-source software(OSS), especially in the most important area close to QC, remains limited. ⚫ Barrier to entry for the QC field is high. ⚫ Without OSS, the development of practical cloud-based QC systems' scale is unclear, making standardization challenging. We release Open Quantum Toolchain for OPerators and USers (OQTOPUS), one of the largest OSS, offering one-stop access to quantum systems developed by: 5 © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc.
6. https://github.com/oqtopus-team Key Features of Cloud-based QC Systems for Users ⚫ The most fundamental feature in QC is sampling, and conventional systems and OQTOPUS use similar features. ⚫ We assume that users desire feature enhancement allowing for efficient and effective use of QC. Prioritize them based on: 6 A) What will be more useful to users from the perspective of being essential for many quantum algorithms. ⇒ Server-side transpiler, Server-side execution, Estimation, Circuit cutting, ・・・ B) Improving the calculation accuracy of QCs suffering from errors. ⇒ Error Mitigation, Error Correction, ・・・ C) Enhancing convenience for users to use QCs. ⇒Multi-programming, Composer, Online code editor, ・・・ © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc.
7. https://github.com/oqtopus-team Key Features of Cloud-based QC Systems for Users All essential features—except composer—are implemented close to quantum hardware and are fully open-sourced in OQTOPUS, unlike many closed features in other cloud systems. Key features IBM Quantum Amazon Braket Azure Quantum Qibo qBraid OQTOPUS Server-side transpiler ✓(P)a ✓(N) ✓(N) ✓(P) ✗ ✓(P) Server-side executionb ✓(N) ✓(N) ✓(N) ✗ ✗ ✓(P) Multi-programmingb ✗ ✗ ✗ ✗ ✗ ✓(P) Error Mitigation ✓(P) ✓(N) ✓(N) ✓(P) ✓(N) ✓(P) Estimation ✓(P) ✓(N) ✗ ✓(P) ✗ ✓(P) Composerb ✓(N) ✗ ✗ ✗ ✓(N) ✓(P) Systems 7 b: Although these features already exist, they are not publicly available in these popular conventional systems. a: (P) and (N) mean public and non-public, respectively. © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc.
8. https://github.com/oqtopus-team Overview of OQTOPUS ⚫ OQTOPUS is full-stack quantum computing system including discussed features, operating across three layers (https://github.com/oqtopus-team). ⚫ OQTOPUS with superconducting QC already in operation at UOsaka. 8 © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc.
9. https://github.com/oqtopus-team Overview of OQTOPUS ⚫ Frontend layer, where computation is performed on user’s computer, sends quantum jobs written in Python submitted by user to cloud layer. ⚫ Quantum circuit is converted to a code written in OpenQASM 3. 9 © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc.
10. https://github.com/oqtopus-team Overview of OQTOPUS ⚫ In cloud layer receiving quantum jobs, OQTOPUS Cloud is responsible for job and user managements. ⚫ User can check jobs and hardware via web UI of OQTOPUS Frontend. 10 © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc.
11. https://github.com/oqtopus-team Overview of OQTOPUS ⚫ OQTOPUS Engine in backend layer retrieves jobs and executes quantum programs, working with Tranqu Server providing transpilers and Device Gateway serving as interface for connecting to pulse controller. 11 © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc.
12. https://github.com/oqtopus-team Overview of OQTOPUS ⚫ Operation ensures consistent performance with DevOps for Quantum. ⚫ OQTOPUS Admin facilitates user and device management via web UI. ⚫ QDash performs calibration and visualizes results for system operation. 12 © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc.
13. https://github.com/oqtopus-team Frontend & Cloud Layer 13
14. https://github.com/oqtopus-team Frontend layer - QURI Parts OQTOPUS ⚫ Adopting the quantum programming library QURI Parts written in Python. ✓ Many quantum circuit libraries, including Qiskit are written in Python. ⇒ It’s easy to learn. ✓ Includes the stable versions of popular libraries (NumPy, SciPy, etc.). ⚫ Converting to OpenQASM 3 code. ✓ Most widely used intermediate representation. ⇒ highly versatile. Code example of QURI Parts OQTOPUS 14 The Frontend layer allows users to write quantum programs in Python using QURI Parts OQTOPUS. © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc. https://github.com/oqtopus-team/quri-parts-oqtopus
15. https://github.com/oqtopus-team Cloud layer - OQTOPUS Cloud ⚫ Adopting a serverless architecture with Amazon Web Services (AWS). ✓ Scale flexibly in response to varying workloads while reducing operational costs. ✓ Reducing cost associated with cyberattack risk by leveraging AWS services. ⚫ Defining public interfaces in OpenAPI Specification. Overview of OQTOPUS Cloud 15 Cognito User manager API Gateway API Gateway Frontend layer Backend layer Lambda Server program Lambda Server program RDS Database The Cloud layer handles job submission and user authentication, offering a scalable and secure bridge between users and the backend system. © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc. https://github.com/oqtopus-team/oqtopus-cloud
16. https://github.com/oqtopus-team Cloud layer - OQTOPUS Frontend ⚫ Serving as GUI for writing and managing quantum jobs. ✓ User can check job status and device info (chip topology, T1/T2, readout fidelity). ✓ Providing composer allowing user to design, visualize, and run quantum circuits. ⚫ Issuing API key to interact with OQTOPUS from other frontend services. Job status Composer Device info 16 The Cloud Frontend provides users with a web interface to manage quantum jobs and devices, offering visualization, monitoring, and seamless API access. © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc. https://github.com/oqtopus-team/oqtopus-frontend
17. https://github.com/oqtopus-team Backend Layer – Base Features • Engine • Device Gateway • Error Mitigation 17 © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc.
18. https://github.com/oqtopus-team Backend layer - OQTOPUS Engine ⚫ Scheduling execution of quantum jobs. ✓ Getting jobs from the cloud layer every few seconds. ✓ Communicating with the Device Gateway to perform quantum computation. ✓ Following a first-in-first-out queue, it executes jobs sequentially. ⚫ Handling various processes for quantum computing, including key features, through protocols such as gRPC. ✓ Server-side execution ✓ Multi-programming ✓ Error mitigation ✓ Estimation OQTOPUS Engine with Microservices Job process • Synthesize circuit • Mapping to physical qubits Pre-Process Execution quantum circuit Process Error mitigation Post-Process Some Plugin Tranqu Device- Gateway REST API REST API … Mitigator Some Plugin Job queue 18 gRPC From: Cloud To: Cloud The Backend layer schedules and executes quantum jobs, connecting the cloud with quantum devices through queued workflows and gRPC services. © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc. Microservices Engine https://github.com/oqtopus-team/oqtopus-engine
19. https://github.com/oqtopus-team Backend layer – Tranqu ⚫ Unified API for multiple transpilers (e.g., Qiskit, staq) ⚫ Seamless switching without code changes ⚫ Server-side transpilation via Tranqu Server 19 © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc. Tranqu provides a unified interface for quantum circuit transpilation, enabling easy switching between transpilers and efficient server-side execution. Qiskit User Without vs With Tranqu Without Tranqu Mapping & Converting One unified IF for all transpilers - Seamless switching Staq Qiskit Transpiler Staq Tranqu User With Tranqu Unified IF ! Different IF ! https://github.com/oqtopus-team/tranqu
20. https://github.com/oqtopus-team Backend layer – Device Gateway ⚫ OpenQASM3 → pulse sequence conversion ⚫ Pluggable backend architecture (select at startup) ⚫ Current backends: Qubex※1 (hardware) and Qulacs ※2 (simulator) 20 Main task Pulse sequence (Output) Transpiled OpenQASM 3 (Input) Device Gateway © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc. Device Gateway converts OpenQASM 3 circuits into pulse commands and connects the OQTOPUS Engine with quantum control systems. Flexible integration with diverse quantum control platforms Device Gateway Qulacs Qubex Handler Pluggable Architecture ※1: https://github.com/amachino/qubex ※2: https://github.com/qulacs/qulacs https://github.com/oqtopus-team/device-gateway
21. https://github.com/oqtopus-team Backend layer - Error Mitigation 21 Error mitigation corrects noisy measurement data into more accurate signals using error matrices. ⚫ Noisy measurement results are mapped to ideal outcomes using an error matrix E. ⚫ Correction is performed by inverting E to recover optimal signals. ⚫ Tensor product method enables scalability to many qubits. ⚫ Other error mitigation techniques (e.g., ZNE, PEC) are planned for future implementation. 𝐸 = 𝑃(0|0) 𝑃(0|1) 𝑃(1|0) 𝑃(1|1) ⊗ 𝑃(0|0) 𝑃(0|1) 𝑃(1|0) 𝑃(1|1) 𝐶𝑛𝑜𝑖𝑠𝑦 = 𝐸𝐶𝑖𝑑𝑒𝑎𝑙 → 𝐶𝑖𝑑𝑒𝑎𝑙 = 𝐸−1𝐶𝑛𝑜𝑖𝑠𝑦 →Tensor products of each qubit Error Matrix Representation © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc. Mitigator Noisy Data Corrected Data Desired output. Readout Error Mitigation Process
22. https://github.com/oqtopus-team Backend Layer – Key Features • Server-Side Execution • Multi-Programming • Estimation 22 © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc.
23. https://github.com/oqtopus-team Server-side Execution: Fast Hybrid Loops with Low Latency ⚫ Conventional: Re-queuing through the cloud → high latency ⚫ Server-side: Local container near QPU → low latency, fast hybrid loops 23 Job queue Job Job Job Conventional vs Server-Side Frontend Cloud Backend QPU Server-side execution runs jobs near the QPU, avoiding cloud re-queuing and reducing latency. © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc. Re-queuing through cloud. Job queue Job Job Container Container executes close to QPU. Frontend Cloud Backend QPU
24. https://github.com/oqtopus-team Multi-Programming: Efficient Execution of Multiple Circuits ⚫ Manual mode (implemented): User specifies circuit combining ⚫ Automatic mode (planned): Circuits from different users automatically combined 24 Multi-Programming (higher utilization) Conventional (low utilization) © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc. Multi-programming improves QPU utilization by combining multiple circuits into a single job for parallel execution. Job queue Job Job Job Frontend Cloud Backend QPU Job queue Frontend Cloud Backend QPU combining Combined Job Job Job Job
25. https://github.com/oqtopus-team Estimation: Efficient expectation value calculation in one job submission ⚫ Direct estimation of ⟨𝐻⟩ on the engine. ⚫ Built-in error mitigation ⚫ No repeated re-queuing is required, improving efficiency 25 © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc. With Engine-side Estimation, the engine handles basis changes, job decomposition, and error mitigation in one job, returning the final expectation value efficiently. Job queue Job Job Job Conventional vs Engine-Side Estimation Frontend Cloud Backend QPU Submit a job for each term. Job queue Job Job Estimator Frontend Cloud Backend QPU Estimator automatically creates each term job.
26. https://github.com/oqtopus-team Operation • Admin • QDash 26 © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc.
27. https://github.com/oqtopus-team Operation - Admin 27 ⚫ User Management: View, suspend, delete accounts ⚫ Device Management: Register, update, remove QPUs & simulators ⚫ Simple Interface: Web-based GUI ⚫ Future Extension: Device Access Control A unified web-based GUI for managing users and quantum devices, designed for scalability and future device access control. © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc. User A User B QPU Simulator Device Access Control Example: User A can access the simulator. User B can access the QPU and the simulator. Secure and scalable management for quantum cloud operations https://github.com/oqtopus-team/oqtopus-admin
28. https://github.com/oqtopus-team Operation - Calibration 28 ⚫ Qubit properties (frequency, relaxation time T₁, coherence time T₂) change over time. ⚫ Regular calibration ensures high gate fidelity and stable measurements. ⚫ Calibration involves experiments such as spectroscopy, Rabi oscillations, and Ramsey sequences. ⚫ These experiments form the calibration process (see right). Calibration is essential for maintaining reliable quantum operations, compensating for noise and device drift in superconducting qubits. Calibration Process © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc. Spectroscopy Chevron pattern Rabi Ramsey T1 DRAG Readout RB HPI Pulse
29. https://github.com/oqtopus-team Operation - QDash 29 ⚫ Automates calibration steps (spectroscopy, Rabi, T₁, T₂, etc.). ⚫ Tracks qubit properties and gate fidelity over time. ⚫ Provides dashboards for monitoring and comparison. ⚫ Reduces reliance on individual expertise, ensuring consistent and reproducible calibration. QDash automates and visualizes calibration workflows, reducing manual effort and ensuring consistent qubit performance. GUI of QDash © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc. https://github.com/oqtopus-team/qdash
30. https://github.com/oqtopus-team Conclusion ⚫ OQTOPUS is a fully working, open-source stack for quantum cloud systems. Backend-level capabilities are implemented and public. ⚫ Our system is equipped with all basic operation software required to provide it as a cloud service. ⚫ We aim to build a foundation for QC that’s open, reproducible, and accessible. 30 GitHub: https://github.com/oqtopus-team Docs: https://oqtopus-team.github.io Check it out our software! © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc.
31. © 2025 Fujitsu LTD., UOsaka, SEC LTD., TIS Inc. 31 Thank you! 31 GitHub: https://github.com/oqtopus-team Docs: https://oqtopus-team.github.io