Periodic Reporting for period 1 - HPCQS (High Performance Computer and Quantum Simulator hybrid)
Período documentado: 2021-12-01 hasta 2023-05-31
Quantum computing (QC) enjoys great attention at the highest levels of research and economic policy on a national, European and international level. It promises unprecedented potential for important computing tasks such as simulations in chemistry or materials science, optimization and machine learning. With this potential, QC is increasingly attracting interest from industry and scientific groups that use high performance computing (HPC) for their applications. These pilot users are primarily interested in testing whether available quantum computers today or in the foreseeable future are suitable for simulating increasingly complex systems, analyzing large data sets using machine learning methods or performing the hardest optimization task.
Potential applications are materials design for batteries, drug design, portfolio optimization in finances, risk optimization for insurances, flight and train scheduling, hotel reservation, traffic flow prediction, power trading and scheduling, image-based medical diagnostics, and many more. Hence, QC can considerably change science, industry, economy and our everyday life. Although QC still lags behind classical computing for most practical applications, its completely novel computational approach could revolutionize scientific computing and enable discoveries across a diverse range of fields beyond the reach of classical methods.
The research in applications of QC to real-world problems, however, is still in its infancy. Therefore, the early entry into the practical application of QC is of utmost urgency in order to evaluate QC as a new computer technology. The prerequisite for this is early access to quantum computers at the forefront of development, taking into account the different technical approaches. Another prerequisite is the integration of quantum computers into HPC infrastructures in order to enable the execution of quantum-classical hybrid computing models on the integrated HPC-QC infrastructure.
A first argument for hybrid quantum-classical computations is that most applications in QC are today realized algorithmically with a high degree of hybridity. In hybrid computations, classical algorithms are combined with quantum algorithms (e.g. quantum optimizers for machine learning or variational quantum algorithms). Another argument for hybrid computing arises from the side of HPC when considering the energy consumption of modern supercomputers and the potential energy saving through the use of quantum computers.
The HPCQS ("High Performance Computer and Quantum Simulator hybrid") project is building a European pilot infrastructure that will provide tightly coupled systems of quantum simulators (QSs), which are analog quantum computers, and supercomputers to enable quantum HPC hybrid computing. Two such modular quantum HPC systems with identical QSs from Pasqal, each comprising more than 100 qubits, are being installed and federated at CEA/TGCC and FZJ/JSC (see Fig. 1). The programming and access platform is based on Atos' QLM system and uses the Modular Supercomputing Architecture concept developed by JSC and ParTec for tight integration with the HPC system. The infrastructure is developed in a co-design process together with selected use cases. The goal of HPCQS is to prepare European research and industry for the federated use of QSs to address the most demanding computational challenges. For this purpose, use cases are developed to demonstrate the relevance of the HPCQS approach for scientific and industrial applications. Courses for both scientific and industrial user needs are offered to support access and use of the HPCQS infrastructure, as well as training materials, and user manuals.
The deep integration of QSs into high-end HPC systems is unique in the world.
Potential applications are materials design for batteries, drug design, portfolio optimization in finances, risk optimization for insurances, flight and train scheduling, hotel reservation, traffic flow prediction, power trading and scheduling, image-based medical diagnostics, and many more. Hence, QC can considerably change science, industry, economy and our everyday life. Although QC still lags behind classical computing for most practical applications, its completely novel computational approach could revolutionize scientific computing and enable discoveries across a diverse range of fields beyond the reach of classical methods.
The research in applications of QC to real-world problems, however, is still in its infancy. Therefore, the early entry into the practical application of QC is of utmost urgency in order to evaluate QC as a new computer technology. The prerequisite for this is early access to quantum computers at the forefront of development, taking into account the different technical approaches. Another prerequisite is the integration of quantum computers into HPC infrastructures in order to enable the execution of quantum-classical hybrid computing models on the integrated HPC-QC infrastructure.
A first argument for hybrid quantum-classical computations is that most applications in QC are today realized algorithmically with a high degree of hybridity. In hybrid computations, classical algorithms are combined with quantum algorithms (e.g. quantum optimizers for machine learning or variational quantum algorithms). Another argument for hybrid computing arises from the side of HPC when considering the energy consumption of modern supercomputers and the potential energy saving through the use of quantum computers.
The HPCQS ("High Performance Computer and Quantum Simulator hybrid") project is building a European pilot infrastructure that will provide tightly coupled systems of quantum simulators (QSs), which are analog quantum computers, and supercomputers to enable quantum HPC hybrid computing. Two such modular quantum HPC systems with identical QSs from Pasqal, each comprising more than 100 qubits, are being installed and federated at CEA/TGCC and FZJ/JSC (see Fig. 1). The programming and access platform is based on Atos' QLM system and uses the Modular Supercomputing Architecture concept developed by JSC and ParTec for tight integration with the HPC system. The infrastructure is developed in a co-design process together with selected use cases. The goal of HPCQS is to prepare European research and industry for the federated use of QSs to address the most demanding computational challenges. For this purpose, use cases are developed to demonstrate the relevance of the HPCQS approach for scientific and industrial applications. Courses for both scientific and industrial user needs are offered to support access and use of the HPCQS infrastructure, as well as training materials, and user manuals.
The deep integration of QSs into high-end HPC systems is unique in the world.
A joint procurement procedure between GENCI and FZJ of the PPI – Public Procurement of Innovative Solutions – type was carried out with the support of the European High-Performance Computing Joint Undertaking for the acquisition of two QSs capable of controlling at least 100 qubits for tight integration with HPC systems at CEA/TGCC and FZJ/JSC. This PPI resulted with the choice of the Fresnel analog QSs provided by Pasqal, a French startup. The QSs are based on the technology of cold neutral atoms (rubidium) arranged in 2D/3D arrays and excited using laser-based optical tweezers reaching Rydberg states. Currently, both CEA/TGCC and FZJ/JSC are carrying out infrastructure work to install the QSs at the end of 2023.
With a view to future access to promising QC devices as prototypes in the HPCQS hybrid classical quantum infrastructure, HPCQS members have been monitoring the QC market and summarising the information gathered in a first public report evaluating emerging and promising hardware and software for QC. Preparatory work for a second report has already begun.
In order to achieve a sound and aligned design for the QS integration with both the cloud-based portal and the HPC system, the preliminary design for the resource management interface and the definition of the architecture for accessing the QS via the JupyterHub cloud-based portal were defined.
A first programming environment for the QS and the Pulser library provided by Pasqal for composing, simulating and executing pulse sequences for Pasqal devices are installed in the FZJ/JSC and CEA/TGCC supercomputing facilities. A generic programming language for the input to the QS and a first preliminary noisy emulator of the Pasqal HW for a small number of Rydberg atoms (up to 40 qubits) have been developed.
An assessment of the technical requirements for the development of use cases was carried out. This resulted in detailed descriptions of each use case, the expected outcomes and the set-up of the tools and platforms required for developing, integrating, testing and publishing the use cases.
A public website and social media accounts were set up for outreach and visibility of the project. Relevant stakeholders were identified and divided into interested "stakeholder groups", "user groups" or both.
To effectively support HPCQS project management, an intranet, a password-protected, web-based project management platform, was established at the beginning of the project. This platform is used for all kinds of project activities and documents, including reports, results, dissemination activities and publications. Mailing lists and a management guide describing the workflow for reporting have been created. A financial monitoring spreadsheet has been created and a data management plan is in place. Several project meetings have been organised.
With a view to future access to promising QC devices as prototypes in the HPCQS hybrid classical quantum infrastructure, HPCQS members have been monitoring the QC market and summarising the information gathered in a first public report evaluating emerging and promising hardware and software for QC. Preparatory work for a second report has already begun.
In order to achieve a sound and aligned design for the QS integration with both the cloud-based portal and the HPC system, the preliminary design for the resource management interface and the definition of the architecture for accessing the QS via the JupyterHub cloud-based portal were defined.
A first programming environment for the QS and the Pulser library provided by Pasqal for composing, simulating and executing pulse sequences for Pasqal devices are installed in the FZJ/JSC and CEA/TGCC supercomputing facilities. A generic programming language for the input to the QS and a first preliminary noisy emulator of the Pasqal HW for a small number of Rydberg atoms (up to 40 qubits) have been developed.
An assessment of the technical requirements for the development of use cases was carried out. This resulted in detailed descriptions of each use case, the expected outcomes and the set-up of the tools and platforms required for developing, integrating, testing and publishing the use cases.
A public website and social media accounts were set up for outreach and visibility of the project. Relevant stakeholders were identified and divided into interested "stakeholder groups", "user groups" or both.
To effectively support HPCQS project management, an intranet, a password-protected, web-based project management platform, was established at the beginning of the project. This platform is used for all kinds of project activities and documents, including reports, results, dissemination activities and publications. Mailing lists and a management guide describing the workflow for reporting have been created. A financial monitoring spreadsheet has been created and a data management plan is in place. Several project meetings have been organised.
HPCQS creates the core of the first pan-European quantum-HPC hybrid infrastructure providing Europe a visible position in quantum computing alongside the USA, China and Japan. It contributes to the development of a first ecosystem of hybrid HPC and quantum programming facilities and practical applications related to complex simulations and optimization problems. In addition, HPCQS contributes already now to the next generation of modular HPC systems by the integration of the quantum computers with the supercomputers through the modular supercomputer architecture. Moreover, HPCQS has an impact on research projects and market opportunities for companies that are involved in HPCQS, i.e. Pasqal, ATOS, FlySight , ParTec and ParityQC.