I was a bit confused about the gridding formats in gnuplot. What is the
difference between pm3d and dgrid3d. Are they just alternative approaches
for plotting non-gridded data? Or is there something more or less
scientific about either of them.
In case I do have data already on a grid are either of these settings
(pm3d
and dgrid3d) irrelevant? Or not?
What I did not understand about dgrid3d was in what sense it was different
from just a weighted interpolation to a desired grid? I've used splines,
liner-interpolation, cubics etc. to process non-gridded data in the past
to
fit a grid to enable isoline plotting. Is dgrid3d also an analogous
"interpolation" scheme?
I think I understood from the docs that each point that dgrid3d "produces"
depends not only on the adjacent points but essentially on "all" points
(of
course with the farther points weighed less depending on the scaling
parameter) Did I understand it right or am I mistaken? Can one set a
smaller "point-neighborhood" so that a certain point's interpolation
depends only on a few neighbors so as not to create a combinatorial
explosion for a large number of scattered data-points? I guess setting a
large power in the scaling function gets the same effect but then the
function still has to do all the computations, right?
Finally, I'm still mystified by the sentence in the docs that say:
"The `dgrid3d` option is a simple low pass filter that converts scattered
data to a grid data set."
How is this low-pass-filtering? I apologize if I'm just not seeing
something that's obvious.
--
Rahul
