An evolutionary algorithm for solving parameter identification problems in elliptic systems

In this paper, we construct an evolutionary algorithm. It yields good performance on a collection of elliptic parameter identification problems. The evolutionary algorithm has a good tolerability for the noise in the observed data. Even when the noise level is up to 10%, we can also get such a good result. The result of numerical experiments shows explicitly that the algorithm is very fit for solving this kind of inverse problem but not very sensitive to the noise.