** Next:** Decomposition of model space
** Up:** DISCUSSION
** Previous:** DISCUSSION

After about 24 hours of contemplating migration
with weighted median stacks I began to doubt they would work.
The problem is that they do not seem to migrate plane layers!
The Fresnel zone is less than half the hyperbola's length.
Thus the median seems to be unaffected by the tangency
of the hyperbola to the parallel line.
This is a paradox that *L*_{1} is convex and yet it seems
that a search of a full set of basis functions
does not lead us anywhere.
Perhaps a ``trimmed mean'' would do the job.

** Next:** Decomposition of model space
** Up:** DISCUSSION
** Previous:** DISCUSSION
Stanford Exploration Project

11/12/1997