***NOTE: Change in time and location***
Code generation for embedded convex optimization

Jacob Mattingley and Stephen Boyd

Optimization is widespread, but nearly always constrained to problems solved in a few minutes or seconds; often, with a human in the loop. However, using automatic code generation and specially modified algorithms, our solvers can reliably find solutions in milliseconds or microseconds. Such solvers have potential application in many fields, including signal processing, automatic control, machine learning and finance.

In this talk, I will describe the special requirements and opportunities in embedded convex optimization. I will briefly demonstrate CVXGEN, a convenient tool for generating library-free, flat C code ready for compilation into a high-speed solver. CVXGEN is just as simple to use as conventional parser-solvers like CVX or YALMIP, but creates solvers that are thousands of times faster, and suitable for embedding. The primary focus of the talk will be on the techniques we use to make the solvers extremely reliable and fast.