Skip to content

fixed dynamic pagerank vertex update computation#2738

Closed
p1anv wants to merge 2 commits into
apache:masterfrom
p1anv:master
Closed

fixed dynamic pagerank vertex update computation#2738
p1anv wants to merge 2 commits into
apache:masterfrom
p1anv:master

Conversation

@p1anv

@p1anv p1anv commented Oct 9, 2014

Copy link
Copy Markdown

There seems to be a typo here when comparing to static version or to the usual pagerank algorithm. if we take a node in the graph, on each iteration, we add a random jump to any node in the graph with equal probability of resetProb/N. Then summing for all nodes in the graph yields the resetProb on line 155, which has nothing to do with the pagerank value for last iteration.

There seems to be a typo here when comparing to static version or to the usual pagerank algorithm. if we take a node in the graph, on each iteration, we add a random jump to any node in the graph with equal probability of resetProb/N. Then summing for all nodes in the graph yields the resetProb on line 155, which has nothing to do with the pagerank value for last iteration.
@AmplabJenkins

Copy link
Copy Markdown

Can one of the admins verify this patch?

1 similar comment
@AmplabJenkins

Copy link
Copy Markdown

Can one of the admins verify this patch?

@ankurdave

Copy link
Copy Markdown
Contributor

@planvin I think this is the intended behavior, because msgSum is a delta, not a new PageRank value. Therefore it needs to be added to the previous rank. This has confused other people in the past: #2100 (comment)

@ankurdave

Copy link
Copy Markdown
Contributor

ok to test

@SparkQA

SparkQA commented Oct 9, 2014

Copy link
Copy Markdown

QA tests have started for PR 2738 at commit a582d1e.

  • This patch merges cleanly.

@AmplabJenkins

Copy link
Copy Markdown

Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/21552/Test FAILed.

As mentioned in #1000, but we still need to add the random jump
@p1anv

p1anv commented Oct 9, 2014

Copy link
Copy Markdown
Author

commited another attempt, sorry I did not see about the delta graph in #2100 (and not #1000 as I wrote in the commit description)

@SparkQA

SparkQA commented Oct 9, 2014

Copy link
Copy Markdown

QA tests have started for PR 2738 at commit 7305560.

  • This patch merges cleanly.

@SparkQA

SparkQA commented Oct 9, 2014

Copy link
Copy Markdown

QA tests have finished for PR 2738 at commit a582d1e.

  • This patch fails Spark unit tests.
  • This patch merges cleanly.
  • This patch adds no public classes.

@AmplabJenkins

Copy link
Copy Markdown

Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/21553/Test FAILed.

@SparkQA

SparkQA commented Oct 9, 2014

Copy link
Copy Markdown

QA tests have finished for PR 2738 at commit 7305560.

  • This patch fails Spark unit tests.
  • This patch merges cleanly.
  • This patch adds no public classes.

@AmplabJenkins

Copy link
Copy Markdown

Test FAILed.
Refer to this link for build results (access rights to CI server needed):
https://amplab.cs.berkeley.edu/jenkins//job/SparkPullRequestBuilder/21555/Test FAILed.

@p1anv

p1anv commented Oct 10, 2014

Copy link
Copy Markdown
Author

I am clearly missing something here. Sorry for that. I am closing the PR.

@p1anv p1anv closed this Oct 10, 2014
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants