Adaptive wavelet methods and sparsity reconstruction for inverse heat conduction problems (Q609553)

The authors apply adaptive wavelet methods for a numerical treatment of inverse heat conduction problems. The problem is modeled as heat conduction equation with Dirichlet data on the inner boundary and Neumann data on the outer boundary. The nonhomogeneous problem is split into two parts; one homogeneous and a constant term. The inverse problem is thus formulated as a linear parabolic one with new boundary data on the outer boundary. It is assumed that the boundary data has a sparse representation in a basis and is well suited for capturing the overall, smooth temperature distribution as well as some very localized structures with very few coefficients. The problem is thus solved by an adaptive wavelet scheme. The wavelets are well suited for the detection of sparse structures and also provide convergent solutions. Some theorems on the mapping of the nonhomogeneous problems to homogeneous problems are stated and proved.