ICCOPT2025USC

Hi! I'm Matt Menickelly, and I'm a computational mathematician at Argonne National Laboratory, part of the US Department of Energy's network of national laboratories. I broadly characterize my research as being in "mathematical optimization", and I particularly focus on expensive optimization.

What do I mean by expensive? This can be interpreted several ways, but imagine about any application where evaluating an objective or a constraint function in an optimization model is a serious bottleneck. The function could be expressed as a computer simulation that requires nonnegligible wall clock to perform. The function could be derived from experimental output that entails a material cost to a laboratory.

Often, this means my research ends up intersecting with derivative-free optimization. I'm generally thinking about developing practical (derivative-free) optimization methods that are worried about judicious allocation/selection of function evaluations more than anything else. I love using machine learning techniques as subroutines in optimization methods to do these sorts of thing, which is what I'm talking about at ICCOPT.

In general, if you have expensive optimization models, come chat with me about your problem. I love to get my hands dirty in applications and think about how to exploit structure in specific problems to get tailored performance of algorithms for specific optimization models.