Integrated Workflow for Linear Accelerator Simulation

The use of high-performance computing systems for advanced numerical simulation often involves considerable effort on the part of scientists and engineers to prepare and organize simulation input and output data. In many cases, research productivity is limited by efforts to configure and work with a variety of disconnected software tools, and manually carry out data management, often by ad hoc methods. To reduce this logistical workload, an advanced software framework is proposed that integrates diverse software tools for simulation pre- and post-processing with a workflow management system. Based on the technical feasibility demonstrated in the first project phase, the Phase II effort will build out and deploy a production system at full scale, for use by research institutions in government, academia, and industry. This will include major development in four areas: (i) workflow management software will reduce the manual effort required to prepare and conduct complex, large-scale projects that combine multiple interdependent simulations, (ii) remote visualization software will eliminate many of the current limitations of desktop storage and I/O; in situ visualization will provide real-time feedback during long-duration simulation jobs in addition to reduced I/O, resulting in improved HPC utilization and increased fidelity, (iii) cloud computing will provide an agile alternative to large, fixed computing resources, which is often preferred by corporate enterprises, (iv) software maturation efforts will build on the feasibility prototypes developed in Phase I and incorporate these features into the product code base. Commercial applications for this technology can be found in organizations that use advanced numerical computing for physics-based modeling and simulation. This includes major research and development laboratories and manufacturers, where the proposed software can significantly improve technical workforce efficiency.