Working Group on Numerical Experimentation

Working Group on Numerical Experimentation
Name	Working Group on Numerical Experimentation
Abbreviation	WGNE
Formation	1970s
Type	Research consortium
Headquarters	Geneva
Regions served	Global
Membership	International researchers

Working Group on Numerical Experimentation is an international consortium of researchers and institutions focused on computational studies, numerical methods, and algorithmic benchmarking. Founded amid shifting computational priorities in the late twentieth century, the group brought together specialists from national laboratories, universities, and intergovernmental agencies to address reproducibility, standards, and large-scale simulation. Its influence spans collaborations with major centers, contributions to benchmarking suites, and advisory roles for scientific bodies.

History

The group traces origins to meetings that involved participants from CERN, National Institute of Standards and Technology, Lawrence Berkeley National Laboratory, Los Alamos National Laboratory, and IBM research labs, and later attracted contributors from Massachusetts Institute of Technology, Stanford University, University of Cambridge, Princeton University, and ETH Zurich. Early milestones included coordination with work at Argonne National Laboratory, exchanges with NASA, and workshops held in cities such as Geneva, Paris, Tokyo, and Washington, D.C.. Over successive decades the consortium engaged with initiatives associated with European Space Agency, National Science Foundation, Deutsche Forschungsgemeinschaft, and projects linked to DOE computing facilities, while members published alongside editors from Nature and Science.

Mission and Scope

The mission emphasizes rigorous numerical experimentation, transparent benchmarking, and methodological reproducibility, aligning work with standards promulgated by entities like ISO, IEEE, ACM, and SIAM. Scope includes algorithm validation for problems addressed at CERN detectors, climate models used by Intergovernmental Panel on Climate Change, and simulations informing World Health Organization advisory analyses. The group sets practices that intersect with computational efforts at Harvard University, Yale University, Columbia University, University of Oxford, and University of Toronto.

Organizational Structure

The consortium is organized with steering committees, technical advisory panels, and working teams drawing members from Los Alamos National Laboratory, Sandia National Laboratories, Oak Ridge National Laboratory, and academic departments at Imperial College London and California Institute of Technology. Governance features elected chairs, rotating secretariats hosted by institutions such as University of California, Berkeley and University of Illinois Urbana-Champaign, and liaisons to agencies including European Commission research directorates and JAXA. Publication oversight involves editorial boards with representatives from journals like Journal of Computational Physics, Communications of the ACM, and Proceedings of the National Academy of Sciences.

Major Projects and Publications

Major projects include the development of standardized benchmarks paralleling efforts at TOP500, creation of datasets analogous to initiatives at Kaggle and collaborative test suites comparable to those used by U.S. Department of Energy centers. High-profile outputs were disseminated through venues such as SIAM Journal on Scientific Computing, special issues coordinated with IEEE Transactions on Computers, and white papers presented at conferences like Supercomputing Conference and International Conference on Machine Learning. The group has produced influential reports cited in policy discussions at European Research Council meetings and technical standards referenced by National Institutes of Health grant review panels.

Collaborations and Impact

Collaborations have linked the consortium with CERN simulation groups, climate consortia advising the Intergovernmental Panel on Climate Change, and biomedical modeling teams at Johns Hopkins University and Imperial College London. Impact appears in cross-institutional training programs with Carnegie Mellon University, capacity-building workshops funded by World Bank science initiatives, and cooperative toolchains integrated by Google, Microsoft Research, and Amazon Web Services for high-performance workloads. The group's guidance influenced procurement and architecture choices at national facilities including Oak Ridge National Laboratory and shaped curricula at universities such as University of Michigan.

Methodologies and Tools

Methodologies emphasize open experimental protocols inspired by practices at Reproducibility Project initiatives and incorporate software engineering conventions advocated by Linux Foundation and Apache Software Foundation. Tools developed or standardized by participants include benchmarking suites interoperable with ecosystems like TensorFlow, PyTorch, and numerical libraries from MPI implementations, BLAS variants, and software originating at Netlib. Workflow automation leverages continuous integration patterns used at GitHub and containerization approaches pioneered by projects at Docker and orchestration methods related to Kubernetes.

Category:Scientific organizations Category:Computational science