Fall 2021 Stable and convergent difference schemes for weakly singular convolution integrals
Wesley Davis, Richard Noren
J. Integral Equations Applications 33(3): 271-288 (Fall 2021). DOI: 10.1216/jie.2021.33.271

Abstract

We obtain new numerical schemes for weakly singular integrals of convolution type called Caputo fractional order integrals using Taylor and fractional Taylor series expansions and grouping terms in a novel manner. A fractional Taylor series expansion argument is utilized to provide fractional-order approximations for functions with minimal regularity. The resulting schemes allow for the approximation of functions in Cγ[0,T], where 0<γ5. A mild invertibility criterion is provided for the implicit schemes. Consistency and stability are proven separately for the whole-number-order approximations and the fractional-order approximations. The rate of convergence in the time variable is shown to be O(τγ), 0<γ5 for uCγ[0,T], where τ is the size of the partition of the time mesh. Crucially, the assumption of the integral kernel K being decreasing is not required for the scheme to converge in second-order and below approximations. Optimal convergence results are then proven for both sets of approximations, where fractional-order approximations can obtain up to whole-number rate of convergence in certain scenarios. Finally, numerical examples are provided that illustrate our findings.

Citation

Download Citation

Wesley Davis. Richard Noren. "Stable and convergent difference schemes for weakly singular convolution integrals." J. Integral Equations Applications 33 (3) 271 - 288, Fall 2021. https://doi.org/10.1216/jie.2021.33.271

Information

Received: 12 January 2021; Revised: 9 March 2021; Accepted: 30 March 2021; Published: Fall 2021
First available in Project Euclid: 17 February 2022

MathSciNet: MR4383252
zbMATH: 07543104
Digital Object Identifier: 10.1216/jie.2021.33.271

Subjects:
Primary: 26A33 , 45D05 , 65R20

Keywords: composite quadrature , fractional derivative , integral equations , numerical analysis , numerical convergence

Rights: Copyright © 2021 Rocky Mountain Mathematics Consortium

Vol.33 • No. 3 • Fall 2021
Back to Top