Analysis of formulas for numerical differentiation of functions with large gradients on a Bakhvalov mesh

N.A. Zadorin

Sobolev Institute of Mathematics, Siberian Branch, Russian Academy of Sciences, Novosibirsk, 630090 Russia

E-mail: nik-zadorin@yandex.ru

Received February 3, 2021

ORIGINAL ARTICLE

Full text PDF

DOI: 10.26907/2541-7746.2021.3-4.261-275

For citation: Zadorin N.A. Analysis of formulas for numerical differentiation of functions with large gradients on a Bakhvalov mesh. Uchenye Zapiski Kazanskogo Universiteta. Seriya Fiziko-Matematicheskie Nauki, 2021, vol. 163, no. 3–4, pp. 261–275. doi: 10.26907/2541-7746.2021.3-4.261-275. (In Russian)

Abstract

The article gives an estimate of the error of the classical formulas for the numerical differentiation of a function of one variable with large gradients in the exponential boundary layer. It is assumed that the function is decomposed in the form of the sum of the regular and singular components, which is valid for the solution of a boundary value problem for the ordinary second-order differential equation with a small parameter ε affecting the highest derivative. It is known that the application of the classical polynomial formulas of numerical differentiation to such a function in the case of a uniform mesh can lead to unacceptable errors. The article estimates the error of the formulas for numerical differentiation on the Bakhvalov mesh, which is condensed in the boundary layer region. Bakhvalov’s mesh is widely used to construct uniformly converging difference schemes; therefore, the error estimation of the numerical differentiation formulas on this mesh is of interest. The estimates of the error on the Bakhvalov mesh are obtained taking into account the uniformity in the small parameter for the classical difference formulas widely used to calculate the first, second, and third derivatives. The results of numerical experiments are presented, which agree with the obtained error estimates. A numerical comparison of the obtained errors on the Bakhvalov and Shishkin meshes and on a uniform mesh is carried out.

Keywords: function of one variable, boundary layer, large gradients, Bakhvalov mesh, formulas for numerical differentiation, error estimation

Acknowledgments. The study was supported by the Russian Foundation for Basic Research (project no. 19-31-60009).

References

1. Bakhvalov N.S. The optimization of methods of solving boundary value problems with a boundary layer. USSR Comput. Math. Math. Phys., 1969, vol. 9, no. 4, pp. 139–166. doi: 10.1016/0041-5553(69)90038-X.
2. Shishkin G.I. Setochnye approksimatsii singulyarno vozmushchennykh ellipticheskikh i parabolicheskikh uravnenii [Grid Approximations of Singular Perturbation Elliptic and Parabolic Equations]. Yekaterinburg, Ural. Otd. Ross. Akad. Nauk, 1992. 233 p. (In Russian)
3. Zadorin A.I., Zadorin N.A. Spline interpolation on a uniform grid for functions with a boundary-layer component. Comput. Math. Math. Phys., 2010, vol. 50, no. 2, pp. 211– 223. doi: 10.1134/S0965542510020028.
4. Kopteva N.V., Stynes M. Approximation of derivatives in a convection-diffusion two-point boundary value problem. Appl. Numer. Math., 2001, vol. 39, pp. 47–60.
5. Shishkin G.I. Approximations of solutions and derivatives for a singularly perturbed elliptic convection-diffusion equations. Math. Proc. R. Ir. Acad., 2003. vol. 103A, no. 2, pp. 169–201.
6. Zadorin A.I. Analysis of numerical differentiation formulas in a boundary layer on a Shishkin grid. Numer. Anal. Appl., 2018, vol. 11, no. 3, pp. 193–203. doi: 10.1134/S1995423918030011.
7. Zadorin A.I., Zadorin N.A. Interpolation formula for functions with a boundary layer component and its application to derivatives calculation. Sib. Elektron. Mat. Izv., 2012, vol. 9, pp. 445–455.
8. Il’in A.M. Differencing scheme for a differential equation with a small parameter affecting the highest derivative. Math. Notes, 1969, vol. 6, no. 2, pp. 596–602. doi: 10.1007/BF01093706.
9. Zadorin A., Tikhovskaya S. Formulas of numerical differentiation on a uniform mesh for functions with the exponential boundary layer. Int. J. Numer. Anal. Model., 2019, vol. 16, no. 4, pp. 590–608.
10. Il’in V.P., Zadorin A.I. Adaptive formulas of numerical differentiation of functions with large gradients. J. Phys.: Conf. Ser., 2019, vol. 1260, art. 042003. doi: 10.1088/1742-6596/1260/4/042003.
11. Blatov I.A., Zadorin A.I., Kitaeva E.V. An application of the exponential spline for the approximation of a function and its derivatives in the presence of a boundary layer. J. Phys.: Conf. Ser., 2018, vol. 1050, art. 012012, pp. 1–7. doi: 10.1088/1742-6596/1050/1/012012.
12. Blatov I.A., Zadorin A.I., Kitaeva E.V. Approximation of a function and its derivatives on the basis of cubic spline interpolation in the presence of a boundary layer. Comput. Math. Math. Phys., 2019, vol. 59, no. 3, pp. 343–354. doi: 10.1134/S0965542519030047.
13. Linß T. The necessity of Shishkin decompositions. Appl. Math. Lett., 2001, vol. 14, no. 7, pp. 891–896. doi: 10.1016/S0893-9659(01)00061-1.
14. Linß T. Layer–Adapted Meshes for Reaction-Convection-Diffusion Problems. Berlin, Springer, 2010. 233 p.
15. Blatov I.A., Zadorin N.A. Interpolation on the Bakhvalov mesh in the presence of an exponential boundary layer. Uchenye Zapiski Kazanskogo Universiteta. Seriya Fiziko-Matematicheskie Nauki, 2019, vol. 161, no. 4, pp. 497–508. doi: 10.26907/25417746.2019.4.497-508. (In Russian)

The content is available under the license Creative Commons Attribution 4.0 License.