expose full behavior of Barabasi-Albert generator algorithm

[nicolaspayette]

Currently, the nw:generate-preferential-attachment-primitive uses a fixed value of 1 for the numEdgesToAttach parameters and automatically runs the algorithm for a number of iterations equals to the specified desired number of vertices.

Maybe we could have, in addition to that simplified way of doing things, other primitives for sophisticated users:

• nw:barabasi-albert-init with number of initial vertices and number of edges to attach
• nw:barabasi-albert-step to iterate the algorithm

This would require to keep track of which vertices where created by the current instance of the algorithm. It is doable, but not trivial.

[nicolaspayette]

In discussion with Bill Rand, the idea of simply adding a generate-preferential-attachment-with-min-degree came up: the only difference would be an additional min-degree parameter that would correspond to numEdgeToAttach in Jung or k in the original algorithm.

[jbadham]

yes, that's all that is necessary - how many edges to add with each node (min-degree). The only trick is that you have to create a complete (all nodes connect to all other nodes) graph with min-degree nodes before adding the individual nodes. But it's not a hard thing to code for a user (onne they have worked out the joys of the rnd:weighted-n-of syntax), though I guess having it built in would make it quicker.

A fixed degree generator would also be useful.

[qiemem]

Fixed by e8741ba8dbaa2774cc27370941f30bf78579d655!