You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@mahout.apache.org by QIFAN PU <qi...@berkeley.edu> on 2018/04/19 02:08:00 UTC

distributed cholesky on mahout

Hi,

I'm wondering if distributed cholesky decomposition on mahout is supported
now.
From this doc:
https://mahout.apache.org/users/sparkbindings/ScalaSparkBindings.pdf
It seems that the implementation is single-node?

Thanks,
Qifan

Re: distributed cholesky on mahout

Posted by Qifan Pu <qi...@gmail.com>.
Got it. I need to do cholesky on a large square matrix. Thanks for the
info, Ted, Trevor & Dmitriy. Good to know.

On Thu, Apr 19, 2018 at 2:43 PM, Dmitriy Lyubimov <dl...@gmail.com> wrote:

> no distributed Cholesky as far as i know.
> Thin QR or ssvd.
>
> On Wed, Apr 18, 2018 at 7:08 PM, QIFAN PU <qi...@berkeley.edu> wrote:
>
> > Hi,
> >
> > I'm wondering if distributed cholesky decomposition on mahout is
> supported
> > now.
> > From this doc:
> > https://mahout.apache.org/users/sparkbindings/ScalaSparkBindings.pdf
> > It seems that the implementation is single-node?
> >
> > Thanks,
> > Qifan
> >
>

Re: distributed cholesky on mahout

Posted by Dmitriy Lyubimov <dl...@gmail.com>.
no distributed Cholesky as far as i know.
Thin QR or ssvd.

On Wed, Apr 18, 2018 at 7:08 PM, QIFAN PU <qi...@berkeley.edu> wrote:

> Hi,
>
> I'm wondering if distributed cholesky decomposition on mahout is supported
> now.
> From this doc:
> https://mahout.apache.org/users/sparkbindings/ScalaSparkBindings.pdf
> It seems that the implementation is single-node?
>
> Thanks,
> Qifan
>

Re: distributed cholesky on mahout

Posted by Trevor Grant <tr...@gmail.com>.
Hey Qifan,

I think you can do a distributed QR if the matrix is thin.

http://mahout.apache.org/docs/latest/algorithms/linear-algebra/d-qr.html

dqrThin(drmA) I think you want.



On Thu, Apr 19, 2018 at 10:09 AM, Ted Dunning <te...@gmail.com> wrote:

> There was a variant of cholesky decomposition in Mahout at one time not so
> long ago. I would guess that it is still there.
>
> It is difficult to make a truly distributed version of QR decomposition,
> but for the purposes of the randomized SVD in Mahout, it wasn't actually
> necessary to have a true QR.
>
> I don't have a pointer handy and I am not even sure that this code is still
> in Mahout.
>
> Sorry about that.
>
>
>
> On Wed, Apr 18, 2018 at 7:08 PM, QIFAN PU <qi...@berkeley.edu> wrote:
>
> > Hi,
> >
> > I'm wondering if distributed cholesky decomposition on mahout is
> supported
> > now.
> > From this doc:
> > https://mahout.apache.org/users/sparkbindings/ScalaSparkBindings.pdf
> > It seems that the implementation is single-node?
> >
> > Thanks,
> > Qifan
> >
>

Re: distributed cholesky on mahout

Posted by Ted Dunning <te...@gmail.com>.
There was a variant of cholesky decomposition in Mahout at one time not so
long ago. I would guess that it is still there.

It is difficult to make a truly distributed version of QR decomposition,
but for the purposes of the randomized SVD in Mahout, it wasn't actually
necessary to have a true QR.

I don't have a pointer handy and I am not even sure that this code is still
in Mahout.

Sorry about that.



On Wed, Apr 18, 2018 at 7:08 PM, QIFAN PU <qi...@berkeley.edu> wrote:

> Hi,
>
> I'm wondering if distributed cholesky decomposition on mahout is supported
> now.
> From this doc:
> https://mahout.apache.org/users/sparkbindings/ScalaSparkBindings.pdf
> It seems that the implementation is single-node?
>
> Thanks,
> Qifan
>