Closed msharmavikram closed 5 years ago
I didn't feel comfortable putting special logic into the reader just because the lowest node id is 1. That would prevent files with a node id of 0 from working. So the graphs right now effectively have a node 0 with no outgoing edges. A better solution might be to transform the graph files so they have numbering starting at 0 instead of 1.
@dannyk0104 what is it about cross decomp that you think requires this?
On further reflection, the EdgeListFile should be a true reflection of the contents on disk. If you need to subtract one from all nodes, feel free to call a function on each retrieved edge.
Nodes start from 0 but in our graphs, we have nodeid starting from 1.
in graph challenge code base - we perform a subtract of 1 to ensure we start our node from 0. We see this subtraction missing in the read_tsv file.
Danny depends on this for enabling cross decomposition algorithm.