Talk:Force-based algorithms
From Wikipedia, the free encyclopedia
Pseudocode pieces for algorithms usually give (or discuss) initial conditions. I think here the initial layout of the nodes is supposed to be random (but taking care to avoid nodes in exact same locations) and zero velocities.Balrog-kun 22:47, 23 January 2007 (UTC)
We have found that smart initialization is a significant advantage. I don't agree that slow running times are an intrinsic property of force-directed layout - even the article points to fast implementations (c.f. Barnes-Hut and fast MDS techniques.) That naive algorithms are slow just means that they are ... naive. And, speaking of MDS, the article should make this connection obvious (an old gripe). In general the article is thoughtful and well-written. --StephenCNorth 18:06, 6 July 2007 (UTC)
this_node.velocity := (this_node.velocity + timestep * net_force) * damping
Shouldn't we try to keep the units consistent? Multiplying an amount of time times a force does not yield units of velocity. Then again, we'd have to add mass into the algorithm to keep it true, or we can rename the "net force" to something more appropriate. 208.252.106.139 20:06, 23 July 2007 (UTC)
- Yes and no. Fruchterman & Reingold ("Graph Drawing by Force-directed Placement", Software -- Practice and Experience, 21(11):1129-1164, starting at page 1131, near the bottom) carefully note that "We were inspired by natural systems such as springs and macro-cosmic gravity, but must point out that the 'forces' are not correctly named. We use forces to calculate velocity for each time quantum (and thus displacement, since the time of a quantum is unity), whereas true forces induce acceleration. The distinction is extremely important, because the real definition leads to dynamic equilibria (pendulums, orbits), and we seek static equilibria."
- Looking at this article, it is clear it is heavily imbued with the false impression numerical modeling of physical systems are being undertaken by these "force-directed" layout algorithms, as opposed to simply taking what was needed to solve the problem, mangling some established terminology from other disciplines along the way. mdf 15:02, 31 October 2007 (UTC)
Where does the pseudocode algorithm come from? A source link would be helpful…Hypotroph 06:53, 29 October 2007 (UTC)
- I don't know, but the pseudo-code given here is essentially the same given by the Fruchterman & Reingold reference I note above. Also note that the pseudo-code represents the 'naive' implementation. mdf 15:02, 31 October 2007 (UTC)
[edit] Why SPAM?
Yesterday I put a link to http://morcego.arca.ime.usp.br on external links (sorry for being anonymous), and it was removed as spam. So I ask for explanations on criteria for considering this spam, since: 1 - The page is about a software that implements exactly the concept shown here; 2 - the software is Free Software, licensed under LGPL; 3 - Someone with scientific interest in the subject "Force-based algorithms" can see a real example of one and read the source code of its implementation;
So, I ask permission to put the link back, or an explanation for this being a spam.
Tks, Luis —Preceding unsigned comment added by LuisFagundes (talk • contribs) 16:24, 10 November 2007 (UTC)
- You mean as 189.33.147.115 (talk · contribs · deleted contribs · logs · block user · block log), where your sole contributions were to add the same link two three articles. See WP:EL and WP:SPAM, especially Wp:spam#How_not_to_be_a_spammer --Ronz 16:41, 10 November 2007 (UTC)
- Tks for that, I've reviewed the article and all the links. I usually prefer to contribute anonymously and this was my first linking around, so I was not used to spam rules. I see that my motivations were not always the best. Since I make this software with no commercial interest at all, am a strong believer on Wikipedia and Free Software, I wrongly assumed that free software promotion was ok, sorry for that.
-
- As I see: one of my links were not legitimate (in mind maps article), so thanks for removal. One was very proper and was kept, and this specific one is deserver discussion. I think Morcego fits as an illustration of the concept, both visually and its source code. Maybe it would be more proper to link direct to http://morcego.arca.ime.usp.br/Physical+Model, that is the exact page in site that explains the Force-based algorithm used by the software. But that article don't add much to the concept itself, the software is the illustration, so a link to the homepage might also be appropriate.
-
- So, if your judgment based on the content and not on behaviour tells you so, I ask you to please put back the link, either to homepage or to the Physical Model page.
