Andrea Brilli

Postdoctoral researcher in Operations Research

I am a postdoctoral researcher at the Department of Computer, Control, and Management Engineering (DIAG) at Sapienza University of Rome. My research focuses on derivative-free optimization, nonsmooth optimization, and multi-objective optimization.

I completed my PhD in Operations Research, developing novel algorithms for optimization problems where gradient information is unavailable or unreliable. My work combines theoretical analysis with practical algorithm development and implementation.

I am currently working toward designing a general directional derivative-free algorithmic framework that leverages the advantages of direct search and line-search methods while mitigating their inherent limitations. There might be news soon!

Research Interests

  • Derivative-free optimization methods
  • Nonsmooth constrained optimization
  • Multi-objective optimization
  • Numerical algorithms and convergence analysis

Teaching

I teach courses in mathematical programming and optimization methods at the university level.

My Projects

log-ds

A Matlab implementation of derivative-free direct search methods. The algorithm is able to address general nonconvex constrained optimization problems with a penalty-interior point method, addressing non-relaxable constraints if necessary. The performance is enhance by quadratic models and simplex gradients. The implementation is undergoing a cleaning stage for an official release, but a pre-release is available if anyone is curious. For anyone interested in testing the algorithm, send me an e-mail and we can set things up together!

log-dfl

A Python implementation of a derivative-free line-search method for general nonconvex constrained optimization problems. As for log-ds, send me an e-mail for any support!