Motivated by the needs of large-scale scientific applications, Argonne’s Mathematics and Computer Science division is developing robust, reliable, efficient numerical methods and tools; delivering them to the community as portable, high-performance software and libraries; and engaging with the scientific community to expand capabilities as needed for work toward next-generation predictive simulations and analysis. Products of our research include solvers for algorithmic differentiation, differential algebraic equations, linear and nonlinear equations, numerical optimization, variational inequalities and complementarity problems, and partial differential equations.
Formulating and testing new technologies to meet the needs of large-scale scientific applications, through projects such as the PATH solver for mixed complementarity programming and the TAO toolkit for advanced optimization.
Creating open-source libraries that have been deployed on the world’s leading supercomputers, through projects such as PETSc.
Developing technologies for generating efficient derivative code for models implemented as computer program, through projects such as OpenAD/F.