HPO-NAVI: Sustainable Infrastructures for Archiving and Publishing High-Performance Optimization Software

• Applicant: Professor Dr. Thorsten Koch
• Subject Area: Mathematics
• Term: from 2018 to 2021
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 391087700

The software Ubiquity Generator is a uniquely flexible framework to port a broad class of algorithms from the fields of discrete and global optimization to modern high performance computing architectures with small overhead. It holds the record for the coordinated use of 80,000 CPU cores to solve a single mixed-integer optimization problem. As such it bears high relevance in two aspects: first, for algorithmic research at the interface of applied mathematics and computer science in the context of rapidly evolving parallel hardware architectures; second, for the solution of complex optimization problems for which sequential algorithms currently fall flat.Most immediately, the goal of the proposed project is to improve the usability and visibility of the Ubiquity Generator framework, and to enable its long-term, sustainable development by defining clear workflows for software development that facilitate the interaction of a larger group of distributed developers and users. Its application in future research projects shall be facilitated by the development of a template for a data management plan that adheres to the standards of good scientific practice specified by DFG and other funding agencies.On a broader scope, the project aims at achieving these goals by creating general infrastructures and formulating best practices not only for UG, but for other academic software projects in the field of high performance computing, optimization, and beyond. This includes in particular the development of citation recommendations and metadata schemes implementing the international standards of the FORCE11 initiative, experimentation with an OAIS-compliant digital preservation system for source code, and the extension of the widely-used library platform OPUS for publishing versioned software.

• DFG Programme: Research data and software (Scientific Library Services and Information Systems)
• Participating Institution: FIZ Karlsruhe; Leibniz-Institut für Informationsinfrastruktur GmbH (FIZ KA)
• Co-Investigators: Dr. Ambros Gleixner; Dr. Wolfgang Peters-Kottig; Dr. Yuji Shinano; Dr. Wolfram Sperber