Answers are grounded in this paper and cite their sources.
We propose and analyse a fully discrete method for the time-fractional Gray–Scott reaction–diffusion system with a Caputo derivative of order α∈(0,1) and homogeneous Neumann boundary conditions. The spatial discretisation is a Galerkin method in the space of C2 cubic splines, spanned by uniform cubic B-splines, in which the Neumann condition is natural and no boundary modification of the basis is required. The temporal discretisation is the nonuniform Alikhanov (L2-1σ) formula on graded or graded-then-uniform meshes, which resolves the weak singularity of the solution at t=0. All terms, including the autocatalytic nonlinearity uv2, are evaluated implicitly at the offset point tn−α/2, and the resulting nonlinear systems are solved by Newton’s method. We show numerically that this implicit offset-point treatment remains robust in regimes where extrapolation-based linearisations become unstable, namely small α (e.g., α=0.3) and strongly graded meshes, and we give a cancellation-free evaluation of the discrete convolution kernels. Using a globally Lipschitz cut-off of the nonlinearity and the discrete fractional Grönwall inequality, we prove unique solvability of the cut-off scheme under a step-size condition, its stability, and the error bound ON−min{rα,2}+h4 in L2(Ω). Under an additional mild coupling of the time and space mesh sizes the cut-off is inactive, so the original scheme has a solution that is unique among discrete solutions bounded by the cut-off level and satisfies the same error bound. Numerical experiments in one and two space dimensions confirm second-order temporal and fourth-order spatial accuracy. In two applications, Pearson-type spot formation and one-dimensional pulse splitting, the memory of the Caputo derivative slows the dynamics in the tested parameter regimes: pulse splitting is delayed and the pulse speed drops eightfold from α=0.99 to α=0.5, and a perturbation that decays for α=0.99 evolves into four spots that remain stationary up to t=4000 for α=0.9 and 0.7, whereas for α=0.5 it only begins to split near t=4000.
No takes yet. Share an insight, caveat, or question.
Iqbal et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: