On Thu, 2004-04-22 at 14:51, Vadim V. Zhytnikov wrote:
> Find maximal (Z_max) and minimal (Z_min) values for the plot
> Replace T by Z_max or Z_min choosing one by looking at the slope
> sign at adjacent points (are we going up or down approaching singular
> point). If this method fails (e.g. slope is different from different
> directions) then replace T by average of adjacent values.
I thought of something like this, also. At the very least I think the
singular points should be twice the height of the highest non-singular
points.
I'm worried, however, that this line of thought leads to a routine that
produces potentially misleading results. What does a plot of
sqrt(sin(x^2+y^2)),[x,-4,4],[y,-4,4]
look like using this algorithm?
--Jim