Accelerator Simulation Framework for Virtual Test Stands

STATEMENT OF THE PROBLEM Computer modeling is essential to furthering particle accelerator design and operation. Enhancements to the scale and sophistication of accelerator facilities have merited commensurate advances in modeling complexity. State-of-the-art modeling tools now permit the inclusion of multi-scale modeling techniques, machine learning enhanced algorithms, and execution across heterogeneous architectures up to and including exascale supercomputers. Proper use of these tools requires specialized expertise in the constituent models as well as computing environment and architecture constraints. GENERAL STATEMENT OF HOW THE PROBLEM IS BEING ADDRESSED We propose to develop an extensive graphical user interface for comprehensive particle accelerator simulations. The resulting interface will support the coordination, execution, and evaluation of robust simulations requiring multiple codes, multi-faceted models, and heterogeneous computing resources. Our approach is to develop standardized descriptions of beams and beamlines, building upon existing conventions and community standards to provide holistic support for end-to-end accelerator tracking. These shared descriptions will inform simulation inputs across modeling archetypes, ranging from matrix-based tracking to fully explicit 3D particle-in-cell approaches. We will complement these simulation descriptions with dedicated visualization and analysis tools, leveraging existing I/O and visualization frameworks. WHAT IS TO BE DONE IN PHASE I? During Phase I we will build an initial user interface and define shared representations for coordinating high fidelity accelerator simulations. We will leverage these representations in tandem with an application framework to coordinate execution and analysis of simulations integrating multiple constituent codes. We will build these tools within a containerized environment to support execution across heterogeneous computing architectures. We will identify multiple prototype modeling exercises based on existing accelerator facilities to serve as touchstones for the interface and associated applications. COMMERCIAL APPLICATIONS AND OTHER BENEFITS Scientific and industrial efforts increasingly rely on high performance computing to design, optimize, and analyze experiments. The open-source scientific application framework developed during this work will enable sophisticated simulation support and consulting services for targeted industry and research partners in burgeoning areas of energy production, medicine, and basic science. Integrated visualization features will circumvent bottlenecks associated with data storage and retrieval, thereby reducing costs for customers. By leveraging a suite of state-of-the-art libraries, the platform will be extensible to additional domain- specific applications in the future.