Re: [eigen] Iterators with dense matrices: (was: Areas of eigen3 variation and using eigen2/3 in generic code)

[Benoit Jacob <jacob.benoit.1@xxxxxxxxx>]

Hi,

I have done the renaming, so the names begin() and end() are now free
for you to use for iterators!

Side note: sed's syntax is *really* strange. I spent a ridiculous
amount of time getting it just 90% automated.

Benoit

2010/1/4 Benoit Jacob <jacob.benoit.1@xxxxxxxxx>:
> Hi,
>
> Thanks for sharing all these thoughts. I put a few comments below.
>
> Later today I will do the renaming, making it possible to add proper
> iterators support with begin()/end(). But, as Hauke mentions, we are
> currently in a rush to prepare Eigen 3.0 and we can't afford to
> prioritize iterators at the moment. Also, as I mentioned above, once
> this renaming is done, nothing prevents you from adding iterators
> yourself, even without patching eigen at all:
>
> http://eigen.tuxfamily.org/dox/CustomizingEigen.html#ExtendingMatrixBase
>
> 2010/1/4 Jesse Perla <jesseperla@xxxxxxxxx>:
>> 3) Cursors for Sparse Matrices
>> This is getting above my pay grade... I hope Paul or someone who does a lot
>> of sparse algorithms can correct it.
>> Where cursors/iterators are essential is with sparse algorithms.  The reason
>> is that you need to go through only the non-zero elements.
>
> We already have this, Gael who wrote the Sparse module based it
> completely on iterators.
>
>> 4) Iterators for Matrices and Tensors
>> Assuming that Eigen moves towards having multidimensional structures,
>
> In fact, Eigen is completely specialized to 1D (Vectors) and 2D
> (matrices). Given how hard the 2D case already is to optimize, and
> given how convenient it is to be able to completely specialize the API
> on these cases, we don't plan to support higher-dimensional arrays.
>
> Benoit
>