Novel Variants of Diffusive Representation of Fractional Integrals: Construction and Numerical Computation

Renu Chaudhary,Kai Diethelm
2023-12-18
Abstract:In this paper, we revisit the diffusive representations of fractional integrals established in \cite{diethelm2023diffusive} to explore novel variants of such representations which provide highly efficient numerical algorithms for the approximate numerical evaluation of fractional integrals.
Numerical Analysis
What problem does this paper attempt to address?
The main objective of this paper is to develop efficient numerical techniques for approximating the Riemann-Liouville fractional integral, reducing computational complexity and memory usage. Specifically, the authors achieve this by exploring new variants of the diffusion representation of fractional integrals. The main contributions of the paper are as follows: 1. **New Variants of Diffusion Representation**: Based on the existing diffusion representation theory, the authors propose a new representation method that allows the integrand to exhibit exponential decay characteristics at the endpoints of the domain. This feature helps improve the efficiency of numerical integration. 2. **Exponential Sum Approximation of Kernel Functions**: By expressing the kernel function of the fractional integral as a sum of exponentials and using the Prony method to reduce the number of terms, the numerical computation process is optimized. This method significantly reduces computational complexity and memory usage. 3. **Gauss-Laguerre Formula**: In addition to the exponential sum approximation method, the authors also developed the Gauss-Laguerre formula as another approximation method for fractional integrals. This provides more options for practical applications. These improved methods not only enhance computational efficiency but also provide strong support for the practical application of fractional calculus. Future research directions include extending the application range of the diffusion representation, exploring more forms of permissible transformations and their impact on the efficiency of numerical algorithms, and further analyzing the error bounds and convergence properties of numerical algorithms.