** Next:** Combining model-space and data-space
** Up:** Model versus data normalization
** Previous:** Computational cost

If the estimation problem, ,
is underdetermined, then a standard approach is to find the solution with
the minimum norm. For the *L*2 norm, this is the solution to
| |
(98) |

As a corollary to the the methodology outlined above for creating
model-space weighting functions, Claerbout (1998a) suggests constructing
diagonal approximations to by probing the
operator with a reference data vector, .
This gives data-space weighting functions of the form,

| |
(99) |

which can be used to provide a direct approximation to the
solution in equation (),
| |
(100) |

Alternatively, we could use as a data-space
preconditioning operator to help speed up
the convergence of an iterative solver:

| |
(101) |

** Next:** Combining model-space and data-space
** Up:** Model versus data normalization
** Previous:** Computational cost
Stanford Exploration Project

5/27/2001