Answers are grounded in this paper and cite their sources.
Explores coefficient reconstruction in parabolic equations, indicating robust methods for ill-posed problems.
The inverse problem under consideration concerns a one-dimensional parabolic equation whose thermal diffusivity takes the quadratic-in-space form as(τ)κ2+bs(τ)κ+cs(τ). The unknowns are three time-dependent coefficients as(τ),bs(τ),cs(τ) together with the temperature field T(κ,τ). The direct problem supplies initial data, Neumann boundary conditions, and three over-determination conditions: two boundary temperatures and the spatial integral of T. We prove two theorems. The first theorem establishes the local-in-time existence of a solution under explicit regularity and sign conditions on the given data ξ,νk,δℓ,θ and compatibility at τ=0. The second theorem guarantees the uniqueness of this solution. Despite uniqueness, the inverse reconstruction remains ill-posed: small perturbations in the over-specified data can cause large deviations in the recovered coefficients. For the forward model, we implement two numerical schemes: (i) a Crank–Nicolson finite difference methodology (CN-FDM) on a uniform grid and (ii) a semi-discretized Crank–Nicolson approach combined with Bell spectral collocation in space (CN–Bell). The inverse step minimizes a Tikhonov-regularized least-squares functional using MATLAB’s (R2026a) lsqnonlin. Two numerical examples (smooth and non-smooth), tested with both exact synthetic data and artificially added noise, demonstrate stable and accurate coefficient reconstructions. The framework applies directly to heat conduction and porous media flow where diffusivity varies quadratically in space.
No takes yet. Share an insight, caveat, or question.
Mousa J. Huntul (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: