January 2020
Intermediate to advanced
640 pages
16h 56m
English
Now that we have completed our brief tour of the Calculator API, it's time to shift our focus to the most important part of the implementation: the compute function.
At the end of each super-step, vertices are expected to evenly distribute their PageRank score to their neighbors. Under our graph processing model, this task is facilitated by broadcasting a message. The IncomingScoreMessage type describes the payload for the exchanged messages:
type IncomingScoreMessage struct { Score float64 } func (pr IncomingScoreMessage) Type() string { return "score" }
To bootstrap the calculator, we need to set the initial PageRank score for every vertex in the graph to the value 1/N, where ...