You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@spark.apache.org by "Shay Elbaz (Jira)" <ji...@apache.org> on 2020/08/10 06:26:00 UTC
[jira] [Updated] (SPARK-32578) PageRank not sending the correct
values in Pergel sendMessage
[ https://issues.apache.org/jira/browse/SPARK-32578?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Shay Elbaz updated SPARK-32578:
-------------------------------
Description:
The core sendMessage method is incorrect:
{code:java}
def sendMessage(edge: EdgeTriplet[(Double, Double), Double]) = {
if (edge.srcAttr._2 > tol) {
Iterator((edge.dstId, edge.srcAttr._2 * edge.attr))
// *** THIS ^ ***
} else {
Iterator.empty
}
}{code}
Instead of using the source PR value, it's using the PR delta (2nd tuple arg). This is not the documented behavior, nor a valid PR algorithm AFAIK.
This is a 7 years old code, all versions affected.
was:
The core sendMessage method is incorrect:
{code:java}
def sendMessage(edge: EdgeTriplet[(Double, Double), Double]) = {
if (edge.srcAttr._2 > tol) {
Iterator((edge.dstId, edge.srcAttr._2 * edge.attr))
// THIS ^
} else {
Iterator.empty
}
}{code}
Instead of sending the source PR value, it sends the PR delta. This is not the documented behavior, nor a valid PR algorithm AFAIK.
This is a 7 years old code, all versions affected.
> PageRank not sending the correct values in Pergel sendMessage
> -------------------------------------------------------------
>
> Key: SPARK-32578
> URL: https://issues.apache.org/jira/browse/SPARK-32578
> Project: Spark
> Issue Type: Bug
> Components: GraphX
> Affects Versions: 2.3.0, 2.4.0, 3.0.0
> Reporter: Shay Elbaz
> Priority: Major
>
> The core sendMessage method is incorrect:
> {code:java}
> def sendMessage(edge: EdgeTriplet[(Double, Double), Double]) = {
> if (edge.srcAttr._2 > tol) {
> Iterator((edge.dstId, edge.srcAttr._2 * edge.attr))
> // *** THIS ^ ***
> } else {
> Iterator.empty
> }
> }{code}
>
> Instead of using the source PR value, it's using the PR delta (2nd tuple arg). This is not the documented behavior, nor a valid PR algorithm AFAIK.
> This is a 7 years old code, all versions affected.
>
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscribe@spark.apache.org
For additional commands, e-mail: issues-help@spark.apache.org