抄録
This paper discusses the improvement of an ill-posed linear system solutions concerning with multidimensional field distribution problems. The discrete Wavelet transforms arrange the system matrix as well as input vector to solve, compressing the data density to mother wavelets. This transformed system reduces the effects of a vector representing multidimensional data. An inverse source problem in magnetic field is taken up to compare the solutions given by generalized vector sampled pattern matching method (GVSPM). As a result, we have found that lower order Wavelet bases perform efficient analysis and higher accuracy of solutions.
