We consider the Gauß-Newton method for the solution of nonlinear least squares problems in the context of parameter identification problems. A scheme is presented in which the discretization error is controlled in such a way that in the initial phase of the algorithm a rather coarse grid level is being used whereas a refinement takes place as one moves closer to the solution of the problem. The theoretical justification is a mesh independence result on the rate of convergence for the iterates produced by the Gauß-Newton method. If the refinement of the mesh is done at a linear or quadratic rate then the method with mesh refinement during the iteration retains the linear or quadratic rate of convergence.

Key words and phrases

Gauß-Newton method parameter identification mesh independence 
1980 Mathematics Subject Classification (1985 Revision35R30 65K10 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [1]
    E. L. Allgower and K. Böhmer, Application of the mesh independence principle to mesh refinement strategies, SIAM J. Numer. Anal. 24 (1987), 1335–1351.CrossRefGoogle Scholar
  2. [2]
    E. L. Allgower, K. Böhmer, F. A. Potra, and W. C. Rheinboldt, A mesh-independence principle for operator equations and their discretizations, SIAM J. Numer. Anal. 23 (1986), 160–169.CrossRefGoogle Scholar
  3. [3]
    H. T. Banks and D. W. Iles, A comparison of stability and convergence properties of techniques for inverse problems, LCDS 86–3, Brown University, Providence, RI, 1986.Google Scholar
  4. [4]
    H.T. Banks and K. Klinisch, Estimation Techniques for Distributed Parameter Systems, Birkhäuser—Verlag, Boston, Basel, Berlin, 1989.CrossRefGoogle Scholar
  5. [5]
    J. E. Dennis and R. B. Schnabel, Numerical Methods for Nonlinear Equations and Unconstrained Optimization, Prentice-Hall, Englewood Cliffs, N.J., 1983.Google Scholar
  6. [6]
    V. Friedrich and B. Hofman, A Two Grid Approach to Identification and Control Problems For Partial Differential Equations, Large Scale Scientific Computing (P. Deuflhard and B. Engquist, eds.), Birkhäuser, Boston, Basel, Stuttgart, 1987, pp. 257–268.CrossRefGoogle Scholar
  7. [7]
    M. Heinkenschloss, Mesh independence for nonlinear least squares problems with norm constraints, Tech. Rep., Universität Trier, FB IV — Mathematik, 1990.Google Scholar
  8. [8]
    M. Heinkenschloss and E. W. Sachs, The role of growth rates for Gauß-Newton methods and parameter identification problems, Fifth Symposium on Control of Distributed Parameter Systems (A. El Jai and M. Amouroux, eds.), 1989, pp. 79–84.Google Scholar
  9. [9]
    C. T. Kelley, Operator prolongation methods for nonlinear equations, Computational Solution of Nonlinear Systems of Equations (E. L. Allgower and K. Georg, eds.), Vol. 26 of AMS Lectures in Applied Mathematics, Providence, RI, 1990, American Mathematical Society, pp. 359–388.Google Scholar
  10. [10]
    C. T. Kelley and E. W. Sachs, Approximate quasi-Newton methods, Mathematical Programming, ser. B 48 (1990), 41–70.CrossRefGoogle Scholar
  11. [11]
    M. Kroller and K. Kunisch, A numerical study of an augmented Lagrangian method for the estimation of parameters in a two point boundary value problem, Tech. Rep. 87 (1987), Technical University of Graz, Austria.Google Scholar
  12. [12]
    S. Omatu and J.H. Seinfeld, Distributed Parameter Systems. Theory and Applications, Oxford University Press, Oxford, New-York, Toronto, 1989.Google Scholar

Copyright information

© Springer Basel AG 1991

Authors and Affiliations

  • M. Heinkenschloss
    • 1
  • M. Laumen
    • 1
  • E. W. Sachs
    • 1
  1. 1.FB IV - MathematikUniversität TrierTrierWest Germany

Personalised recommendations