Numerical analysis of an inverse coefficient problem for a chemical transformation model

Inverse coefficient problems for a non-stationary chemical transformation model are considered. The objective of this work is to test an approach consisting in reducing the inverse problem to a quasi-linear matrix equation based on sensitivity operators constructed from an ensemble of independent solutions of adjoint equations. A Newton-Kantorovich-type algorithm is used to solve the thus obtained matrix equations. This approach is tested on a chemical transformation scheme with 22 species and 20 reactions. The reconstruction results are compared with several sets of unknown reaction rates according to the influence characteristics. The analysis seems to be useful for selecting sets of reaction rates that can be reconstructed by the inverse problem solution.