On the resolution of the non-smooth inverse Cauchy problem by the primal-dual method

Laghrib, Amine

DOI: https://doi.org/10.1007/s11075-024-01950-5

IF: 2.37

2024-11-09

Numerical Algorithms

Abstract:This paper delves into addressing an inverse Cauchy problem using the primal-dual method. This approach is motivated by the inherent challenges posed by the irregular nature of the unknown parameter, characterized as type, which more accurately reflects real-world phenomena than traditional regularity assumptions. Our approach involves reformulating the inverse problem into a PDE-constrained optimization framework, incorporating fidelity and regularization terms in the norm. Subsequently, we apply a primal-dual method to effectively navigate the non-differentiability of the cost function. In tandem with theoretical advancements, we rigorously establish the stability of the solution, particularly in the presence of noisy data, ensuring convergence. To substantiate the efficacy and stability of our algorithm, we undertake a series of numerical experiments spanning both two and three dimensions. These experiments not only demonstrate the performance of our methodology but also underscore its resilience in handling real-world noisy data scenarios. Finally, we present experimental results demonstrating the efficiency of our proposed approach. These results include instances with both noisy and noise-free data, represented in two and three dimensions.

mathematics, applied