Optimization in model reduction

Abstract EN

Optimization techniques and optimal approximations in the context of model reduction of high dimensional systems by low order models are reviewed.

Both time domain and frequency domain criteria are considered.

First, classical optimal reduction techniques are briefly summarized, leading mostly to highly computational iterative algorithms with no guaranteed convergence to the global solution.

The more recent optimal Hankel approximation techniques are then surveyed.

The solution to these problems led to explicit closed form algorithms with highly tractable and efficient computations.

The various techniques and algorithms are simulated on simple examples for illustration and comparison.