[
https://issues.apache.org/jira/browse/MAHOUT-1813?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15196353#comment-15196353
]
ASF GitHub Bot commented on MAHOUT-1813:
----------------------------------------
GitHub user andrewpalumbo reopened a pull request:
https://github.com/apache/mahout/pull/194
MAHOUT-1813: Functional "apply" DSL for distributed and in-memory matrices
We have functional "Assign" for in-memory matrices, e.g.:
```
mxA := { x => x + 1 }
mxA ::= { x=> x * 2 }
```
However, we lack similar unary elementwise function capability with
distributed matrices, because distributed matrices are logically immutable.
The suggestion here is to use apply(func) to augment that capability for
DRMs:
```
drmA(x => x + 1)
drmA(x => 2 * x)
```
Handling sparse/dense functional assignments: Second optional parameter,
evalZeros, can be one of TRUE, FALSE, or AUTO (default AUTO).
It regulates treatement of 0 elements in matrices: whether we can skip
assignments to 0 elements. AUTO (which is default) tests the
supplied function whether it changes 0 elements:
evalZeroTest = f(0) != 0
Obviously, this will not work in all situations (e.g. non-deterministic
functions), so we want to retain ability to explicitly override this
with `true` or `false`.
For completeness mostly, similar functional no-side-effect applies are
implemented for in-memory types, e.g.:
```
mxA(x => x + 1)
mxA(x => 2 * x)
mxA( (row, col, x) => x + 1)
```
etc.
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/andrewpalumbo/mahout MAHOUT-1813
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/mahout/pull/194.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #194
----
commit 3a42189f254da5e1c2d3781f0865f4640f33fe94
Author: Andrew Palumbo <[email protected]>
Date: 2016-03-15T20:23:29Z
Functional apply for in-core matrices and DRMs
----
> Functional "apply" DSL for distributed and in-memory matrices
> -------------------------------------------------------------
>
> Key: MAHOUT-1813
> URL: https://issues.apache.org/jira/browse/MAHOUT-1813
> Project: Mahout
> Issue Type: New Feature
> Affects Versions: 0.11.2
> Reporter: Andrew Palumbo
> Assignee: Andrew Palumbo
> Fix For: 0.12.0
>
>
> We have functional "Assign" for in-memory matrices, e.g.:
> {code}
> mxA := { x => x + 1 }
> mxA ::= { x=> x * 2 }
> {code}
> However, we lack similar unary elementwise function capability with
> distributed matrices, because distributed matrices are logically immutable.
> The suggestion here is to use apply(func) to augment that capability for DRMs:
> {code}
> drmA(x => x + 1)
> drmA(x => 2 * x)
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
