Decoupled matrix Riccati differential equations approach for robust boundary data completion in time-fractional diffusion problems

Abstract

Abstract This research introduces an innovative algorithmic framework tailored to solve the inverse boundary data completion problem for time-fractional diffusion equations in a bounded domain, especially under partially specified Neumann and Dirichlet conditions. This issue is notoriously ill-posed in the Hadamard sense which demands a sophisticated and nuanced approach. Our method innovatively transforms this problem into a system of first-order differential equations, linked with Matrix Riccati Differential Equations. Moving beyond traditional methods, our framework integrates a state-of-the-art decoupling algorithm, which effectively blends the strategic depth of optimal control theory with the precision of the Golden Section Search algorithm. This integration determines the optimal regularization parameter essential for ensuring the stability and the reliability of the solution. The robustness and effectiveness of our approach have been rigorously verified through extensive numerical experiments, proving its resilience even in conditions marked by significant noise levels. AMS Subject Classifications: 34K20, 35Pxx, 35S16, 60K50, 35R11