Github user mengxr commented on a diff in the pull request:
https://github.com/apache/spark/pull/964#discussion_r13743001
--- Diff:
mllib/src/main/scala/org/apache/spark/mllib/linalg/distributed/RowMatrix.scala
---
@@ -201,6 +202,31 @@ class RowMatrix(
}
/**
+ * Multiply the Gramian matrix `A^T A` by a DenseVector on the right.
+ *
+ * @param v a local DenseVector whose length must match the number of
columns of this matrix.
+ * @return a local DenseVector representing the product.
+ */
+ private[mllib] def multiplyGramianMatrix(v: DenseVector): DenseVector = {
+ val n = numCols().toInt
+
+ val bv = rows.aggregate(BDV.zeros[Double](n))(
+ seqOp = (U, r) => {
+ val rBrz = r.toBreeze
+ val a = rBrz.dot(v.toBreeze)
+ rBrz match {
+ case _: BDV[_] => brzAxpy(a, rBrz.asInstanceOf[BDV[Double]], U)
--- End diff --
`axpy` uses multimethod. It should be unnecessary to check the types here.
Did you see a performance improvement by making the types explicit?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
