A corrected \(\mathrm{L}1\) method for a time-fractional subdiffusion equation (Q6101675): Difference between revisions

The paper titled ``A corrected \(\mathrm{L}1\) method for a time-fractional subdiffusion equation'' by Zheng Yang and Fanhai Zeng focuses on enhancing numerical methods for solving fractional differential equations, specifically addressing the computational challenges associated with the Caputo fractional derivative operator. Recognizing the widespread application of piecewise linear interpolation in discretizing the Caputo fractional derivative, leading to the widely used L1 method, Yang and Zeng introduce a corrected \(\mathrm{L}1\) method (C\(\mathrm{L}1\)) that aims to offer an improved accuracy level akin to the \(\mathrm{L}1\) method but with significant reductions in storage and computational demands due to the nonlocal nature of the Caputo operator. The core scientific problem tackled in the manuscript is the efficient and accurate discretization of the Caputo fractional derivative operator, particularly in the context of fractional initial value and time-fractional subdiffusion problems. The traditional \(\mathrm{L}1\) method, while popular, faces limitations in accuracy and computational efficiency, especially for problems characterized by initial singularities or requiring long-time computations. To address these issues, the authors propose the C\(\mathrm{L}1\) method, which introduces correction terms to the standard \(\mathrm{L}1\) discretization scheme, thereby achieving a better approximation of the fractional derivative. They rigorously prove the convergence of the numerical schemes derived from the C\(\mathrm{L}1\) method and develop a fast, memory-saving variant to mitigate the computational cost associated with the nonlocality of the fractional derivative. This variant, referred to as the fast C\(\mathrm{L}1\) method, utilizes a time-stepping approach that significantly reduces both storage requirements and computational time. Empirical validations demonstrate the theoretical advancements, showcasing the C\(\mathrm{L}1\) method's superior accuracy over the traditional \(\mathrm{L}1\) scheme, especially in cases of strong solution singularities or small fractional order parameters. Moreover, the fast C\(\mathrm{L}1\) method is shown to maintain this improved accuracy while drastically reducing computational costs, as confirmed by numerical experiments involving both fractional initial value and subdiffusion problems. The significance of this research lies in its potential to advance the computational handling of fractional differential equations, which are prevalent in modeling anomalous diffusion processes in various scientific and engineering domains. By providing a more accurate and computationally efficient method for discretizing the Caputo fractional derivative, Yang and Zeng's work contributes to the broader applicability and effectiveness of numerical simulations in studying complex, real-world phenomena characterized by fractional dynamics.