[Date Prev][Date Next][Thread Prev][Thread Next][Author Index][Date Index][Thread Index]

[sw@xxxxxxxx: Re: Sunless-Sea page on EnfiladeTheory]



----- Forwarded message from sw@xxxxxxxx -----

In-Reply-To: <3C3F10D7.6205CBFA@xxxxxxxx>
Date: Fri, 11 Jan 2002 13:07:08 -0800
From: sw@xxxxxxxx
Subject: Re: Sunless-Sea page on EnfiladeTheory

Jeff, if you're there,

Wouldn't it be better if we could carry out (or at least log) this
discussion on the site's message board?

At 8:20 AM -0800 1/11/02, roger gregory wrote:

>     The inclusion of self balancing trees in gold was a mistake or 
>at best a temporary hack,

My page on the Ent implies that they're unbalanced, I forgot about the
splaying.  [Thanks, John, for reading & adding the comment.]

>And most importantly, lets start a discussion on the problems that 
>might benefit form ent&enf theory.
>Solving the theory is 1000 times easier than programming and selling 
>a solution.

I'm lazy & at this point only wanted to make sure the basic intro stuff
was on that site so that smart browsers might get interested (& have some
help understanding more of what's there).

>So steve can you explain your intuition?  I'll look up "Delaunay 
>triangulation" but I've got to run now.

I guess you mean my saying that algorithms can be incrementalized with
enfilades?  Trying to think of an explanation, I realized that this can
bootstrap off the notion of parallelizing: the way you would split a
computation to parallelize it would work to split it into a tree.
I.e. breaking up the problem & assigning work corresponds to going
down the tree; finishing, summarizing and reporting back corresponds
to widding; and the *change* in the computation that would
happen if some of the inputs were changed corresponds to an edit operation
on the enfilade.  Rebalancing corresponds to pretending that the work
was split more fairly--or presciently!

The Delaunay triangulation and the Voronoi diagram--I still don't remember
which is which--are "duals" of each other.  They both relate to finding
the point in a set of points that is closest to a given location.  One
draws line segments between nearest-neighbor points, the other draws
borders between the neighborhoods around the points.  By the way,
I'm told this is one of a million cool pluggins available (for a
charge) from Oracle for their database.  This is "interesting" (in the
curse sense maybe) because the point is to divide a space into *non*-
overlapping areas.

I roughly designed the word-wrap and pagination enfilade in the mid-1980s.
The wids are lookup tables, and the widding operation is
function composition on the tables.

I wrote an in-memory enfilade that wids counts of instances of each ascii
character, & used it in in a compression algorithm, this year.


  --Steve
-- 
Your presence is required.

----- End forwarded message -----

-- 
mailto:xanni@xxxxxxxxxx                         Andrew Pam
http://www.xanadu.com.au/                       Chief Scientist, Xanadu
http://www.glasswings.com.au/                   Technology Manager, Glass Wings
http://www.sericyb.com.au/                      Manager, Serious Cybernetics
http://two-cents-worth.com/?105347&EG		Donate two cents to our work!
P.O. Box 477, Blackburn VIC 3130 Australia	Phone +61 401 258 915