The theory, algorithms, and software of linear algebra are familiar tools across mathematics, the applied sciences, and engineering. This ubiquity of linear algebra masks a fairly recent growth of nonlinear algebra in mathematics and its applications to other disciplines. The proliferation of nonlinear algebra has been fueled by recent theoretical advances, efficient implementations of core algorithms, and an increased awareness of these tools.

The benefits of this nonlinear theory and its tools are manifold. Pushing computational boundaries has led to the development of new mathematical theories, such as homotopy methods for numerical algebraic geometry, tropical geometry and toric deformations, and sums of squares methods for polynomial optimization. This uncovered many concrete nonlinear mathematical objects and questions, many of which are ripe for computer experimentation. In turn, resulting mathematical breakthroughs often lead to more powerful and efficient algorithms... (more)

This workshop will focus on core algorithms in the three crucial areas in nonlinear algebra: numerical algebraic geometry, symbolic computation, and combinatorial methods. There have been tremendous advances in algorithms in these areas. As applications become more sophisticated, and require more computing resources, the basic algorithms and implementations need to step up to match the demand from applications. This workshop will bring together experts to exchange ideas on new algorithms that are needed and on improvement of existing ones. It will incite collaboration on hybrid algorithms involving computational methods from the three areas. Examples of open problems to be addressed include: certification of numerical methods, and combining numerical, symbolic and combinatorial methods to allow a much larger reach for decomposition algorithms.

This workshop will focus on techniques and structures in real algebraic geometry and optimization, including computational tools for semi-algebraic sets, semidefinite programming techniques for polynomial optimization, and applications of these tools to problems in computer vision. Real algebraic geometry provides powerful tools to analyze the behavior of optimization problems, the geometry of feasible sets, and to develop new relaxations for hard non-convex problems. On the other hand, numerical solvers for semidefinite programs have led to new fast algorithms in real algebraic geometry. Algebraic methods over the real numbers are essential for many real-world applications. This workshop aims to explore the cutting edge of techniques in real algebraic geometry and convex optimization as well as applications of these tools to problems in computer vision and other information sciences.

Applications often pose many algorithmic, computational, and theoretical challenges, and overcoming these challenges has been a driving force behind many recent innovations in nonlinear algebra. This workshop will bring together mathematicians and practitioners with a focus on recently developed methods that have been motivated by solving problems arising in applications. Three key hallmarks of the methods presented are efficient computation of solutions, exploitation of structure, and reformulation of numerically unstable systems. Some of the topics planned for discussion include algebraic cryptanalysis and coding theory, chemical reaction networks, computational biology, computer-aided geometric design, applications of enumerative and tropical geometry, gauge and string theory in physics, and applications to statistics such as probabilistic graphical models and singular learning theory.