| Re: [eigen] Initial implementation of tensor support |
[ Thread Index |
Date Index
| More lists.tuxfamily.org/eigen Archives
]
- To: Eigen development <eigen@xxxxxxxxxxxxxxxxxxx>
- Subject: Re: [eigen] Initial implementation of tensor support
- From: Rhys Ulerich <rhys.ulerich@xxxxxxxxx>
- Date: Sat, 2 Nov 2013 20:58:04 -0500
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :content-type; bh=p6K3mlxrlTLN1mW2Wa7M5rIs3dJ7Uxc0vsqfFBaTgSM=; b=xyc9z9BYCSBT4eWrcPSz4zcVJrMr7k90bn9Zo3MPsHgQyAI7ct+FhQXrKej3jt2XZ6 HD0B9XCX2kYus592vm7AF8zAy2YYRFXuteQQi//6c3p07yaaX1+ZSRJ+wsKcYVeEtl1k DNvI1OP7fDnMUw3bvsrGVG8pewjPl5/2+cpVkLkxOvmD/0ALDbw+ZvxGnMp+zEtVTMxO DZDj4QUsPTllA4mxSkRXFCrXvTvxQ+uBQ8inKMX7hD1lsflkLMIVHP0Uo8S2CmuGGRcz CO3Ks062+seu/JG5ocE5F4FobakZmpfiTfpVTCTzAAyxeyh+bT8fJ3mKrYlc6T0hN4zK 5Zrg==
> I've also a question regarding the ColMajor/RowMajor option. I guess they
> rather mean something like forward/backward storage order, but is that
> option really needed??
One feature I've seen in ndarray (http://ndarray.googlecode.com) for
multidimensional arrays is a template parameter indicating the number
of dimensions which are contiguously stored in row-major order
(http://ndarray.googlecode.com/svn/doc/index.html). This permits
matching algorithm striding using compile-time dispatch where
possible. I always thought using a signed template parameter would
permit tracking the analogous column-major guarantee by reversing the
sign. If the template parameter was 0, then no promises are made
about its contiguousness.
- Rhys