# Shape Parameter Estimation in RBF Function Approximation

Shape Parameter Estimation in RBF Function Approximation

A. Karageorghis P. Tryfonos

Department of Mathematics and Statistics, University of Cyprus, Cyprus

Page:
246-259
|
DOI:
https://doi.org/10.2495/CMEM-V7-N3-246-259
N/A
|
Revised:
N/A
|
Accepted:
N/A
|
Available online:
N/A
| Citation

OPEN ACCESS

Abstract:

The radial basis function (RBF) collocation method is applied for the approximation of functions in two variables. When the RBFs employed include a shape parameter, the determination of an appropriate value for it is a major issue. In this work, this is addressed by including the value of the shape parameter in the unknowns along with the coefficients of the RBFs in the approximation. The variable shape parameter case when a different shape parameter is associated with each RBF in the approximation is also considered. Both approaches yield nonlinear systems of equations, which are solved by a standard non-linear solver. The results of several numerical experiments are presented.

Keywords:

collocation, function approximation, radial basis functions

References

[1] Afiatdoust, F. & Esmaeilbeigi, M., Optimal variable shape parameters using genetic algorithm for radial basis function approximation. Ain Shams Engineering Journal, 6(2), pp. 639–647, 2015. https://doi.org/10.1016/j.asej.2014.10.019

[2] Buhmann, M.D., Radial Basis Functions, Cambridge University Press, Cambridge, 2003.

[3] Chen, W., Fu, Z.J. & Chen, C.S., Recent Advances in Radial Basis Function Collocation Methods, Springer Briefs in Applied Sciences and Technology, Springer, Heidelberg, 2014.

[4] Cheng, A.H.D., Multiquadric and its shape parameter - a numerical investigation of error estimate, condition number, and round-off error by arbitrary precision computation. Engineering Analysis with Boundary Elements, 36(2), pp. 220–239, 2012. https:// doi.org/10.1016/j.enganabound.2011.07.008

[5] Fasshauer, G.E., Newton iteration with multiquadrics for the solution of nonlinear PDEs. Computers & Mathematics with Applications, 43(3–5), pp. 423–438, 2002. https://doi. org/10.1016/s0898-1221(01)00296-6

[6] Fasshauer, G.E., Meshfree Approximation Methods with MATLAB, Interdisciplinary Mathematical Sciences, Vol. 6, World Scientific Publishing Co. Pte. Ltd., Hackensack, NJ, 2007.

[7] Fasshauer, G.E., Gartland, E.C. & Jerome, J.W., Newton iteration for partial differential equations and the approximation of the identity. Numerical Algorithms, 25(1/4), pp. 181–195, 2000. https://doi.org/10.1023/a:1016609007255

[8] Fasshauer, G.E. & Zhang, J.G., On choosing optimal shape parameters for RBF approximation. Numerical Algorithms, 45(1–4), pp. 345–368, 2007. https://doi.org/10.1007/ s11075-007-9072-8

[9] Fornberg, B. & Flyer, N., Solving PDEs with radial basis functions. Acta Numerica, 24, pp. 215–258, 2015. https://doi.org/10.1017/s0962492914000130

[10] Fornberg, B. & Wright, G., Stable computation of multiquadric interpolants for all values of the shape parameter. Computers & Mathematics with Applications, 48(5–6), pp. 853–867, 2004. https://doi.org/10.1016/j.camwa.2003.08.010

[11] Jankowska, M.A. & Karageorghis, A., Variable shape parameter Kansa RBF method for the solution of nonlinear boundary value problems. Engineering Analysis with Boundary Elements, 103, pp. 32–40, 2019. https://doi.org/10.1016/j.enganabound.2019.02.005

[12] Jankowska, M.A., Karageorghis, A. & Chen, C.S., Kansa RBF method for nonlinear problems. International Journal of Computational Methods and Experimental Measurements, 6(6), pp. 1000–1007, 2018. https://doi.org/10.2495/cmem-v6-n5-1000-1007

[13] Jankowska, M.A., Karageorghis, A. & Chen, C.S., Improved Kansa RBF method for the solution of nonlinear boundary value problems. Engineering Analysis with Boundary Elements, 87, pp. 173–183, 2018. https://doi.org/10.1016/j.enganabound.2017.11.012

[14] Kansa, E.J., Multiquadrics—a scattered data approximation scheme with applications to computational fluid-dynamics. II. Solutions to parabolic, hyperbolic and elliptic partial differential equations. Computers & Mathematics with Applications, 19(8–9), pp. 147–161, 1990. https://doi.org/10.1016/0898-1221(90)90271-k

[15] Karageorghis, A. & Tryfonos, P., Determination of shape parameter in RBF approximation. WIT Transactions on Engineering Sciences, Vol. 122, WIT Press, pp. 153–162, 2019.

[16] The MathWorks, Inc., 3 Apple Hill Dr. Matlab, Natick, MA.

[17] More´, J.J., The Levenberg-Marquardt algorithm: Implementation and theory. In Numerical Analysis, (Proc. 7th Biennial Conf., Univ. Dundee, Dundee, 1977), Lecture Notes in Math., Vol. 630, Springer, Berlin, pp. 105–116, 1978.

[18] Rippa, S., An algorithm for selecting a good value for the parameter c in radial basis function interpolation. Advances in Computational Mathematics, 11(2/3), pp. 193–210, 1999. https://doi.org/10.1023/a:1018975909870

[19] Sarra, S.A. & Sturgill, D., A random variable shape parameter strategy for radial basis function approximation methods. Engineering Analysis with Boundary Elements, 33(11), pp. 1239–1245, 2009. https://doi.org/10.1016/j.enganabound.2009.07.003

[20] Tryfonos, P., Some numerical experiments on radial basis function approximation. MSc thesis, Department of Mathematics and Statistics, University of Cyprus, 2018.

[21] Tsai, C.H., Kolibal, J. & Li, M., The golden section search algorithm for finding a good shape parameter for meshless collocation methods. Engineering Analysis with Boundary Elements, 34(8), pp. 738–746, 2010. https://doi.org/10.1016/j.enganabound.2010.03.003

[22] Xiang, S., Wang, K.M., Ai, Y.T., Sha, Y.D. & Shi, H., Thrigonometric variable shape parameter and exponent strategy for generalized multiquadric radial basis function approximation. Applied Mathematical Modelling, 36(5), pp. 1931–1938, 2012. https:// doi.org/10.1016/j.apm.2011.07.076