[SPARK-8563] Bug that IndexedRowMatrix.computeSVD() yields the U with wrong numCols

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.0.2, 1.1.1, 1.2.2, 1.3.1, 1.4.1
• Fix Version/s: 1.0.3, 1.1.2, 1.2.3, 1.3.2, 1.4.1, 1.5.0
• Component/s: MLlib
• Target Version/s: 1.0.3, 1.1.2, 1.2.3, 1.3.2, 1.4.1, 1.5.0
• Assignee: 19 Lee
• Reporter: 19 Lee
• Created: 23/Jun/15 11:08
• Updated: 02/Jul/15 18:41
• Resolved: 30/Jun/15 21:14

Description

IndexedRowMatrix.computeSVD() yields a wrong U which U.numCols() = self.nCols.

It should have been U.numCols() = k = svd.U.numCols()

self = U * sigma * V.transpose
(m x n) = (m x n) * (k x k) * (k x n)
-->
(m x n) = (m x k) * (k x k) * (k x n)

Proposed fix: https://github.com/apache/spark/pull/6953

Issue Links

• links to: [Github] Pull Request #6949 (lee19)
• links to: [Github] Pull Request #6953 (lee19)