Mailing-List: contact user-help@mahout.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@mahout.apache.org
Received-SPF: pass (nike.apache.org: local policy)
DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws;
  s=s1024; d=yahoo.com;
  h=X-YMail-OSG:Received:X-Mailer:References:Message-ID:Date:From:Reply-To:Subject:To:Cc:In-Reply-To:MIME-Version:Content-Type;
  b=0tvimsxkM9swZCL78EJwvYoS9aWDbF2vVbPoReHzZH+afj4ZRUJcL03qE3KKVxj4UYgyO69VbomxAt6qJnaE+lAh+kwHwzfA7N7TSpF3vZLIXk8IbS/OAf4Fe6SJEydU8z1RZApu/CRKzFuOpiJpeJFe4v9AWXULuJhQo+8WzZA=;
References: <1327472120.72432.YahooMailNeo@web114201.mail.gq1.yahoo.com>
 <5A5FC957-863B-405C-9C33-CEFEC98585AE@gmail.com>
 <CAJwFCa1k-spMEKgdnPoW+wcxPKjVpFWgksX=fLaDKVAaWGsxtA@mail.gmail.com>
 <1327525851.98941.YahooMailNeo@web114217.mail.gq1.yahoo.com>
 <CAJwFCa19oTGxBK37G+e0qQnY9Y_ud2o790xiW1xbA0jiGuzLzA@mail.gmail.com>
Message-ID: <1327534288.16319.YahooMailNeo@web114215.mail.gq1.yahoo.com>
Date: Wed, 25 Jan 2012 15:31:28 -0800 (PST)
From: B & E <ejawce@yahoo.com>
Reply-To: B & E <ejawce@yahoo.com>
Subject: Re: mahout matrix package
To: Ted Dunning <ted.dunning@gmail.com>
Cc: "user@mahout.apache.org" <user@mahout.apache.org>
In-Reply-To: 
 <CAJwFCa19oTGxBK37G+e0qQnY9Y_ud2o790xiW1xbA0jiGuzLzA@mail.gmail.com>
MIME-Version: 1.0
Content-Type: multipart/alternative;
 boundary="32398890-1719506081-1327534288=:16319"

--32398890-1719506081-1327534288=:16319
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: quoted-printable

thank you very much!=0A=0A=0A________________________________=0A From: Ted =
Dunning <ted.dunning@gmail.com>=0ATo: B & E <ejawce@yahoo.com> =0ACc: "user=
@mahout.apache.org" <user@mahout.apache.org> =0ASent: Wednesday, January 25=
, 2012 1:39 PM=0ASubject: Re: mahout matrix package=0A =0A=0A=0A=0A=0AOn We=
d, Jan 25, 2012 at 1:10 PM, B & E <ejawce@yahoo.com> wrote:=0A=0AIt's for p=
art of a few straightforward matrix operations but over a very trivial size=
 of matrix (~ 10 x 5) within a mapper/reducer job.=A0=0A=0ASuppose you have=
=A0=0A=0A=A0 =A0 =A0Matrix a;=0A=A0 =A0 =A0Vector b;=0A=0Aand you want to s=
olve for x such that a.times(x) =3D b, then this should work:=0A=0A=A0 =A0 =
=A0Matrix bMatrix =3D=A0new Matrix(b.size(), 1).assignColumn(0, b);=0A=A0 =
=A0 =A0Vector x =3D new QRDecomposition(a).solve(bMatrix).viewColumn(0);=0A=
=0A=0AThere is a bit of mess here because there isn't a solve(Vector) metho=
d so we have to construct a matrix and then get the required vector back ou=
t.=0A=0AIf you absolutely want the inverse over my suggestion that you don'=
t want it, try this=0A=0A=A0 =A0 =A0Matrix inverse =3D new QRDecomposition(=
a).solve(new DiagonalMatrix(n))=0A=0AYou will have to pick the right n.=0A=
=0A=0A>=0A>For the inverse method, for now more like just looking for a han=
dy brainless one liner (since the matrix operations is trivial now).=A0=0A=
=0AThere you go.=0A=A0=0ABut do have a tradeoff concern in longer term when=
 the matrix calculations becomes more involved (in terms of complexity but =
not so much of size), would we have better off by using another matrix pack=
age in the first place VS the out-of-box convenience to get some quick proo=
f-of-concepts now.=A0=0A=0AThat is up to you. =A0I still don't know what yo=
u are doing and I don't know what your limits are. =A0Mahout's math package=
 is very good with sparse matrices and is moderate with dense matrices. =A0=
It uses an API style that is really nice after you get to understand it, bu=
t it has a learning curve.=0A=0A=A0=0A=0A>=0A>thanks.=0A>=0A>=0A>=0A>______=
__________________________=0A> From: Ted Dunning <ted.dunning@gmail.com>=0A=
>To: user@mahout.apache.org =0A>Cc: B & E <ejawce@yahoo.com> =0A>Sent: Wedn=
esday, January 25, 2012 3:18 AM=0A>Subject: Re: mahout matrix package=0A> =
=0A>=0A>Matrix inverse is almost never a good idea. =A0The same effect can =
usually be had using a decomposition at far less cost. =A0For instance, for=
 solving a linear system, QR decomposition provides two sub-matrices that c=
an easily have an inverse multiply operation applied to them avoiding the n=
eed for actually computing the inverse.=0A>=0A>=0A>Can you say more about w=
hat you want to do?=0A>=0A>=0A>On Tue, Jan 24, 2012 at 11:47 PM, Raphael Ce=
ndrillon <cendrillon1978@gmail.com> wrote:=0A>=0A>I believe matrix inversio=
n in general may not be so easy to distributed within a map reduce framewor=
k, however you could consider using the SSVD job to calculate an approximat=
e matrix inverse.=0A>>=0A>>=0A>>On 24 Jan, 2012, at 10:15 PM, B & E wrote:=
=0A>>=0A>>> Hi,=0A>>>=0A>>> I have a mahout newbie question -- I wanted to =
do some matrix operations in mahout and just realized Mahout's matrix packa=
ge doesn't seem to have any API on providing an inverse of a matrix.=0A>>>=
=0A>>> Just curious was there a particular reason for such, or just so happ=
ens no code was contributed to add that method to the matrix classes?=0A>>>=
=0A>>> thanks.=0A>>>=0A>>> Ejawce=0A>>=0A>>=0A>=0A>=0A>
--32398890-1719506081-1327534288=:16319--