Nonsmooth Optimization

Many problems derived from computational chemistry and biology, optimal control, and data analysis among others, lead to minimizing or maximizing nonsmooth functions. These problems are typical applications of numeric optimization algorithms. The main goal of this project is to implement and improve nonsmooth optimization algorithms inspired by the family of Spectral Projected methods, and the benchmark is known problems against other optimization algorithms.

Student Outcomes

Students will become familiar with conducting numerical experimentation, compiling results, and writing reports and papers to be presented at specialized conferences.

Student Qualifications

Programming skills, basic knowledge of MATLAB, pre-requisite: STMATH 324 and CSS 161, and the student has to be self-motivated.

Student Responsibilities

Review background material, Create heuristics to solve undesirable numerical behaviors, Implement optimization numerical methods, Attend weekly meetings, and Write reports, presentations and posters to be presented at conferences and meetings.