Github user brkyvz commented on a diff in the pull request:
https://github.com/apache/spark/pull/15730#discussion_r96038711
--- Diff:
mllib/src/main/scala/org/apache/spark/mllib/linalg/distributed/BlockMatrix.scala
---
@@ -459,14 +464,155 @@ class BlockMatrix @Since("1.3.0") (
*/
@Since("1.3.0")
def multiply(other: BlockMatrix): BlockMatrix = {
+ multiply(other, 1)
+ }
+
+ /**
+ * Left multiplies this [[BlockMatrix]] to `other`, another
[[BlockMatrix]]. This method add
--- End diff --
Ok, I feel this is very verbose in terms of documentation. Can we summarize
it somehow?
```scala
/**
* Left multiplies this [[BlockMatrix]] to `other`, another
[[BlockMatrix]]. The `colsPerBlock`
* of this matrix must equal the `rowsPerBlock` of `other`. If `other`
contains
* `SparseMatrix`, they will have to be converted to a `DenseMatrix`. The
output
* [[BlockMatrix]] will only consist of blocks of `DenseMatrix`. This may
cause
* some performance issues until support for multiplying two sparse
matrices is added.
* Blocks with duplicate indices will be added with each other.
*
* @param other Matrix `B` in `A * B = C`
* @param numMidDimSplits Number of splits to cut on the middle dimension
when doing multiplication.
For example, when multiplying a Matrix `A` of size `m x n` with Matrix `B`
of size `n x k`, this parameter
configures the parallelism to use when grouping the matrices. The
parallelism will increase from `m x k` to
`m x k x numMidDimSplits`, which in some cases also reduces total shuffled
data.
```
Wrote out a sketch, please put in proper format, i.e. omitted `*` on the
last lines
---
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.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
