Depicting networks is a tricky subject, and too often leads to representations that are a tangled and complicated mess. Such diagrams are often referred to with terms like ball of yarn or string, a birds nest, cat hair or simply hairball.

It’s also common for a network diagram to be engaging and attractive for its complexity (usually aided and abetted by color), which tends to hide how poorly it conveys the meaning of the data it represents.

This presentation is an excellent survey of methods, and highly recommended for anyone getting started with graph and network data. It’s a useful aid in the “how should I represent this data?” question.

Random Walk is a series of visualisations by Daniel A. Becker showing the relationship between chaos and order in random systems.

The project simulates randomness in a series of visualizations, covering 14 double-sided A2 posters, aiming at delivering insights into the unfamiliar phenomenon. The project was part of a diploma thesis at the University of Applied Sciences in Mainz, Germany

http://www.random-walk.com/

Overriding the maps created by the galaxy generation program produce interesting results.

As I mentioned previously the game is planned to have multiple methods of travelling from system to system the one I’ve recently been looking into is something like a jump-point network which links all systems to at least one neighbour and provides a route from any system to any other.

There are two main graphs which fulfil this quality one is the Delaunay triangulation and the other is it’s sub-graph a Gabriel graph both graphs can be constructed via straight edge and compass.

The Delaunay triangulation is constructed by selecting three points and drawing a circumscribing circle which touches each point if this circle does not wholly contain any points (the initial three points are not contained and sit on the perimeter) then the triangle fills the Delaunay condition and may be added to the list of links between points.

A Gabriel graph is similar but only considers two points:

As I say a Gabriel graph is a sub-graph of the Delaunay triangulation so any links that appear in a Gabriel graph appear in a Delaunay triangulation but some of the links belonging to triangle which fulfils the Delaunay condition will not appear in a Gabriel graph.

Both of these graphs can be transformed into 3D networks by considering a sphere between two points in the case of a Gabriel graph and the Circumsphere of four points in the case of a Delaunay triangulation.

Unfortunately a 2D representation of such a graph looks pretty indecipherable.

A little more progress.

Recently I’ve volunteered to collaborate on a Space RTS game, nothing fancy just an attempt to relive the joys of Master of Orion while I’m waiting for the guy who’s working on the scripting interface to finsih off I decided to have a little play around with the procedural generation of galaxies.

Anyone can create a bob of stars (more on this later) so I decided to try a twin armed spiral galaxy first.

The spiral is pretty easy it’s random displacement and placement of stars around a 2D spiral defined by:

x = A * exp(B * t) * cos(t)

y = A * exp(B * t) * sin(t)

So far so good, tweak A and B to get the right tightness of curve, add in the random noise, invert the x and y values every so often to create a second spiral and you’re there as far as basic testing goes:

Ok to be fair you may want to tweak the constants a bit to make the arms a little shorter but that’s pretty pretty, but it doens’t look like a galaxy for two reasons one real spiral galaxies have a big blob in the middle from which the arms extend, and two all the stars look the same aside from being very tightly packed in the core.

More on my adventures in galaxycraft later, along with a system for mapping a jumpgate network.