Adaptive placement of the pseudo-boundaries improves the conditioning of the inverse problem.