LouvainCore why consider edgeData (or weight) as Long only?

[5abhim]

EdgeData (or Edge Weight) could be a double as well.
What is the significance of internalWeight and nodeWeight in LouvainCore class? and how will they change when edge weight changes to Double from Long?

[eric-kimbrel]

nodeWeight is the weighted degree of a node (sum of all edge weights).

internalWeight represents the sum of all self edge weights. This is typically 0 when starting the algorithm but after a compression stage (where all nodes of a single community are compressed down to a single node) there will be many self edges. Rather than representing the self edges as true edges in the graph framework (which would increase memory and computation required) we simply keep a sum of all the self edges for calculations when it is needed.

Both nodeWeight and internalEdgeWeight would need to change to Double if you changed edge weight to Double. Also we would want to check the "q" method to ensure the computation and variable types are still correct.