Friday, May 16, 2003

hi paul

kominis? i dont know him, give me his email.

sure, this is going to be quite hard.
chew from cornell has approved that chart for me, see his comment in my web

i am thinking for my whole project, but i guess it should be quite universal.
it's about triggering the points.
so i will have function to identify whenever "any number of" walkers are in "particular distance" to each other. anything in "" should be in some user form, user enter the desired number.

i've seen in your code (agent emergent local perception) that you have used look for nearest neighbour, this should be a variant of this function or even a conditional from that.

now i think we could use nest sorting. if we have some sort of grid behind, that represent "particular distance of walker from each other", we sort the locating which x number of walkers have triggered which point. then perhaps we can use the center of that cell of grid as the point.
so this will trigger a voronoi point. lots of walker grouping on a particular cell creates lots of voronoi points.

these meeting points then be stored in array of positions, i guess. then we should triangulate.

the problem of triangulation could be about traces. you have used to represent traces visually present as lines. what should be added is a function to create "traces" from emergent voronoi point to another point with conditional about the same as "do not bump" other lines which i think will say the same as "do not intersecting" at the same time calculating area of these triangles and locate the biggest area. this should solve the convex hull problem.

thats about delaunay triangulation. if i am right, we can built voronoi on the fly from it.
let me have some more thinking this weekend.

i hope this is what we're looking for; the database.

i am tempted to go to the opening of Matrix Reloaded which will be next wed early in the evening. but my husband suggest i do some work in the morning first. thus i will be early next wed.

hope you are eating well,

>From: "Paul S. Coates"
>To: "choesnah idarti"
>Subject: RE: flowchart + calling Mathematica from VB + meet thurs morning
>Date: Fri, 16 May 2003 10:43:52 +0100
>you know that theo (kominis?) last years msc used mathematica to explore 2d
>cas and such, he had a chum who showed him how to do it, but i still think we
>must do it ourselves. your algorithm sounds convincing. can you think about
>the database, i mean how you represent the points, lines triangles and so on.
>it makes all the diff between a difficult project and an easy one. the idea
>is to come up with a structure (TYPE in basic) that holds the data
>appropriateley, especially being able to cluster individual elements (say
>lines) into different ways of referencing them.
>see you thurs