| Re: [eigen] Matrix - Scalar CwiseUnaryOps |
[ Thread Index |
Date Index
| More lists.tuxfamily.org/eigen Archives
]
- To: eigen@xxxxxxxxxxxxxxxxxxx
- Subject: Re: [eigen] Matrix - Scalar CwiseUnaryOps
- From: Benoit Jacob <jacob.benoit.1@xxxxxxxxx>
- Date: Wed, 4 May 2011 08:55:03 -0400
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:mime-version:in-reply-to:references:date :message-id:subject:from:to:content-type; bh=XlJqEKm7Pb72cZ+6K5se0UZ9A2slSdIzhItPTlaeHn0=; b=WWYKO0U2TTeFrv3amkwgUNSxTPmLrnSGiBQ6vLzcLZavHU9GLJ+P/nqr+eLBERJ5up 4L1yPCDnupdVEfXg6uB2TA1jnSP9dLhwWDATE8CpAnTVN9+oI3PBdddogvtmjKliJlqP 5JEB0XOkHT6MVT7KWPZ0XpGLR632YvMfZPugo=
- Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :content-type; b=nCHUGvTqwPYva+EEABvbSiwcjTUGy6Q+Z8RF8UjcHv7Y+oooY3A6xJ3yCRcOusnbBi CPhEL0q3gvuMpQ23cvjDj8+EMITJLIpEYPhG6TFt0FnvzDtNo0ffE0sh4Jh77Xac9nh1 x7Cb8WqxjMmfR3CE4Dxmqiw36lCKwPIJX7BCU=
Yes, the ambiguity is between interpreting a+M as
1) aE+M where E is the matrix all of whose coefficients are 1; this
is what we do for arrays; or
2) aI + M where I is the identity matrix; this is definitely what
mathematicians mean by a+M whenever they use that notation. This is
required if you want basic matrix arithmetic rules to be self
consistent, e.g. distributivity: (a+M)*N = aN + M*N
This ambiguity is the primary reason why we leave this undefined in Eigen.
Benoit
2011/5/4 Thomas Capricelli <orzel@xxxxxxxxxxxxxxx>:
>
> Hi,
>
> From a mathematical point of view a.M is defined but a+M is not...
> (a=scalar, M=Matrix).
>
> my 2 cents..
>
> Thomas
>
> --
>
> Thomas Capricelli <orzel@xxxxxxxxxxxxxxx>
>
> http://www.freehackers.org/thomas
>
> On Wednesday 04 May 2011 11:39:13 Hauke Heibel wrote:
>
>> Hi,
>
>>
>
>> I know that we have the array functionality for those operations but I
>
>> wonder whether we should enable +/- operations between matrices and
>
>> scalars.
>
>>
>
>> We already have * and / and thus I think it might make sense. It's
>
>> just for convenience and I have met quite a few occasions where this
>
>> feature would lead to shorter code. What do you think?
>
>>
>
>> - Hauke
>
>>
>
>>
>
>>
>