Re: Dual scale histograms

Jay Lemke (jllbc who-is-at cunyvm.cuny.edu)
Fri, 21 Nov 1997 20:56:27 -0500

I certainly think it would be interesting to some some actual data plotted
as manifolds (that's the official word for 'blankets' in mathematica) where
there are different time scale on orthogonal axes.

I can indeed imagine some cases where a local event might appear to send
ripples spreading in the longer time dimension, though it is not clear to
me that this will always be readily visible. It is also the inverse case of
the one that conceptually worries me more: not events with longterm
consequences (I suspect we actually overplay this paradigm as it is), but
large-scale processes that affect events on time-scales much shorter than
is supposed to be 'normal'. We might need some other sort of visual
representation to make that more vivid.

I'm sure there are a lot of interesting possibilities to be explored here.

JAY.

PS. Fourier analysis would indeed seem to be very relevant to all these
cases, since in its simplest form it is about the superposition
('accumulation') of effects on different time scales, both those differing
'discretely' and those that could be getting continuously shorter or longer.

For the non-cognoscenti, Fourier analyzed how the troughs and peaks of
waves with different spacing between peaks add up or cancel out; the
spacing corresponds to the time-scale of the movement. It turns out that
any possible movement can be represented as the sum of (maybe an infinite
series) of neat little regular waves with different periodicities.

---------------------------
JAY L. LEMKE

CITY UNIVERSITY OF NEW YORK
JLLBC who-is-at CUNYVM.CUNY.EDU
---------------------------