Work

Error-Balanced Adaptive B-Spline Galerkin Methods with Fast Memory for Time-Fractional Porous Medium Equations

Year: 2026
Type: article
Abstract: We develop an adaptive B-spline Galerkin method for the Caputo time-fractional porous medium equation with heterogeneous conductivity in one and two space dimensions. The method combines truncated hierarchical B-splines refined at degenerate fronts, an L2-type Caputo approximation of order 3−α evaluated in a cancellation-free form on strongly graded meshes, a sum-of-exponentials compression of the memory with a proven relative error bound that is uniform in time, and a residual-based time-step controller. For linear problems, we prove a computable a posteriori bound for the temporal error; it held at every time level in all tests, with effectivity between 2.1 and 5.1 at the final time. After a one-time calibration, the controller met prescribed tolerances to within 1% on an independent problem. Experiments verified against published benchmark tables, exact and self-similar solutions and independent reference computations show that hierarchical refinement reaches uniform-mesh accuracy with 2.1 to 3.6 times fewer peak degrees of freedom for an expanding front in two dimensions, that the spatial error converges at the optimal rate p+1 for degrees up to five, and that the computed infiltration follows the exact law Stα/2 to a relative error of 3×10−6 at t=1. We also show that a published closed-form Barenblatt profile holds only for m=1.
Source: Mathematics
Language: English
Open Access status: gold
Estimated APC paid: $3,137
OpenAlex ID: W7234586630
FWCI:
Cites: 39
Cited by:
Related to: 10