On Wed, Mar 12, 2008 at 6:52 PM, dlakelan <dlakelan at street-artists.org> wrote:
> Mario Rodriguez wrote:
> > dlakelan escribi?:
> >> nticks doesn't seem to affect the workings of the explicit plot type in
> >> draw, however the docs say that it does.
> ...
>
> > Sorry, but the documentation doesn't reflect the real meaning of nticks,
> > which is the same as in the plot-routine case. This is my fault, of course.
>
> I have a function that takes a relatively long time to compute (it
> involves doing a numerical integral inside a root finder, a typical
> evaluation is about 2 seconds). When I plot it via plot2d or draw2d it
> takes FOREVER. It appears that this is due to an interaction between the
> adaptive plotting and the function. When the function changes rapidly,
> it also takes longer to calculate the integral at each step, and the
> plotting routine thinks it needs more points... So the adaptive plotting
> routine basically zeros in on the MOST EXPENSIVE portion of the graph,
> and decides to demand a lot of points there... At least that's what I
> think is going on.
>
> Would it be possible to create a variable that controls the max plot
> time, so that the adaptive algorithm has a time budget that keeps it
> from going off into neverland?
>
> Just a thought. In the mean time I've converted to using parametric
> plots with nticks=15, so that my plot completes in 20 seconds rather
> than 30 minutes. Not as smooth, but acceptable.
Decreasing the value of adapt_depth might help with plot2d and
draw2d(explicit(...)).
--
Andrej