##### Department of Mathematics,

University of California San Diego

****************************

### Math 278 - Center for Computational Mathematics

## Joey Reed

#### UCSD

## Benchmarking Derivative-Free Optimization Algorithms

##### Abstract:

As optimization problems become increasingly complex, the availability and computability of derivatives becomes problematic. As a result, it is important to use so-called direct search methods, which are optimization algorithms that do not use derivative information. We present a small collection of test problems that is designed to facilitate the benchmarking of different direct search algorithms. All implementation has been done in Matlab. This is joint work with Jorge More of Argonne National Lab.

### October 14, 2008

### 11:00 AM

### AP&M 2402

****************************