The experiment result output: all the graphs (running time comparison betweenn HBC, BBC; the speed-up ratio), and the data after processing.
All the data of the experiments for Heuristic Betweenness Centrality and Brandes BC.
[Milestone] Analysis - automatically generate all the graph for all *.out1 input. Those graphs are:- Line chart of running time for BC and HBC with the increasing number of nodes for each graph type [CN | PL | ER] on [router | server] > 4 charts- Bar chart for running time of BC, HBC on [router | server] for Real Community Networks Ninux, FFG, FFW > 2 charts...
Plot the running time of BC and HBC for 3 types of graphs: Erdos, Community Network and Power Lab
Analyze the experiment result.Input: *.out1Output: *.out2, *.out3
.out2 the average running time for each graphs, from 10 repetitions..out3 the average running time for each [graph type (Community Network, Erdos), and number of nodes (from 100 - 1000)]...
Analyze the bi-connected components for CN and PL graphs
[HBC - Python] Backup code
[f] WBBC = HBC :)
Heuristic BC for different graphs
Compute Component Tree Weights - Bugs fixed- Component-vertex pair (B, v): minus 1 unit more- Vertex-component pair (v, B): initialize size = 0
Results: Correct even for Sample 4. Still not correct for Ninux Graph.
More examples to debug the Computing Component Tree Weights - I found problems for the simple4.edges
Debugging the Algorithm 1: Compute Component Tree Weights [Puzis 2012]
Done with creating the communication importance matrix (or Traffic Matrix). This is necessary in calculating betweenness centrality later
Heuristic Betweenness for Structurally Equivalence Class
[r] Heuristic Betweenness for bi-connected components
Heuristic betweenness centrality - block cut tree
Add label to x and y-axis in graph
Script to plot comparison graph between networkx and Boost Graph Library
Comparision between networkx and Boost Graph Library
Fiddling around with C++ - pointer and template function
Fiddling around with networkx library