home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky sci.math:17795 sci.math.symbolic:3394 comp.graphics:13504
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!swrinde!gatech!udel!darwin.sura.net!jvnc.net!netnews.upenn.edu!amy.med.upenn.edu!bill
- From: bill@amy.med.upenn.edu (Bill Tozier)
- Newsgroups: sci.math,sci.math.symbolic,comp.graphics
- Subject: Help find optimum display of directed graphs
- Message-ID: <103943@netnews.upenn.edu>
- Date: 7 Jan 93 16:49:04 GMT
- Sender: news@netnews.upenn.edu
- Followup-To: sci.math
- Organization: Biology Department, University of Pennsylvania
- Lines: 51
- Nntp-Posting-Host: amy.med.upenn.edu
-
- Hello, everyone,
- I need some help with a problem in visualizing mathematical graphs.
- What I'm trying to do is display graphs that include numerous
- loops and cycles in such a way that they're easy to
- understand in a 2-D or 3-D image. For instance:
-
- a -> b
- b -> c
- c -> d
- d -> a
- d -> e
-
- is a simple digraph that would end up looking, in this
- "optimized" form I'm looking for, like this:
-
- a --> b
- ^ |
- | |
- | v
- d <-- c
- /
- /
- L
- e
-
- The criteria for this are (1) that the nodes are as
- far from one another as can be arranged, (2) the edges
- are all approximately the same length, and (3) no edges
- are superimposed.
- One can imagine the image I'd like by linking all connected
- nodes on this graph by, say, loose springs. The minimum
- energy (where "energy" is defined by the above criteria)
- position of all the nodes is what I'm interested in.
- I've actually been trying a simpleminded optimization
- of the sort I just described--with springs, and the
- nagging suspicion that it's been done ages ago in a
- thorough and symbolic way is still with me.
- Does anyone have any suggestions? Note that,
- for trees and other very sparsely connected graphs,
- there really isn't any problem. But for very
- highly connected graphs, finding a layout that
- is simple to "see" is extremely difficult, as
- numerous edges must be discernable in this project.
-
- Thanks for any useful advice. I'd appreciate
- anyone's hints re: Mathematica, or just a theoretical
- optimization in 2-space or 3-space.
-
- Bill Tozier
- bill@amy.med.upenn.edu
-
-