dreambydream / graphfly

GraphFly focus on dependency-flow exploiting.

GraphFly generates the D-trees of initial graph. And the initial graph is divided into many dependency-flows.

$ python main.py

GraphFly can be easily integrated in existing streaming graph systems. Following commands is we use GraphBolt systems (https://github.com/pdclab/graphbolt) to execute graph algorithms based on dependency-flows processing:

$ ./BFS -source 1 -numberOfUpdateBatches 1 -nEdges 100000 -streamPath ../inputs/live-journal/[edge streams file] -outputFile /tmp/output/bfs_log ../inputs/livejournal/ [dependency flows file]

$ ./SSSP -source 1 -numberOfUpdateBatches 1 -nEdges 100000 -streamPath ../inputs/livejournal/[edge streams file] -outputFile /tmp/output/sssp_log ../inputs/livejournal/ [dependency flows file]

$ ./SSWP -source 1 -numberOfUpdateBatches 1 -nEdges 100000 -streamPath ../inputs/livejournal/[edge streams file] -outputFile /tmp/output/sswp_log ../inputs/livejournal/ [dependency flows file]

$ ./CC -source 1 -numberOfUpdateBatches 1 -nEdges 100000 -streamPath ../inputs/livejournal/[edge streams file] -outputFile /tmp/output/cc_log ../inputs/livejournal/ [dependency flows file]

$ ./PageRank -numberOfUpdateBatches 1 -nEdges 100000 -streamPath ../inputs/livejournal/[edge streams file] -outputFile /tmp/output/pr_log ../inputs/livejournal/ [dependency flows file]

$ ./LabelPropagation -numberOfUpdateBatches 1 -nEdges 100000 -streamPath ../inputs/livejournal/[edge streams file] -seedsFile ../inputs/livejournal/lj-seeds-file -outputFile /tmp/output/lp_log ../inputs/livejournal/ [dependency flows file]