In many practical inverse problems, however, non-uniqueness of the solutions or other ill-conditioned solutions is not desired.
In practice, one has only data with noise due to errors in the measurements and inaccuracies of the model itself (Engl and Kugler, 2003).