| Re: [eigen] Matrices based on compact-triangular storage |
[ Thread Index |
Date Index
| More lists.tuxfamily.org/eigen Archives
]
- To: eigen@xxxxxxxxxxxxxxxxxxx
- Subject: Re: [eigen] Matrices based on compact-triangular storage
- From: Benoit Jacob <jacob.benoit.1@xxxxxxxxx>
- Date: Fri, 30 Jul 2010 11:08:35 -0400
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:mime-version:received:received:in-reply-to :references:date:message-id:subject:from:to:content-type :content-transfer-encoding; bh=jLjOv7jhXpZ40hXIo9rxeseqv1vVKHDLWAaHw7G3F6A=; b=Oe9YJCxIiBkNoToF3RK3F0jV/3+fcflb6n3JxHHrNqNDUVh6wY5xog0GQ2R6VEjqtm +N55an+14bgF/MfPCkGWbUQwOji1kGGVMQYYkJGN14R88OqX/SBdlBvkAm6OlRgLpKdG ELEiSEjZPdNJTEWzmOlpIhxq0Ra8UsNKubXTQ=
- 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:content-transfer-encoding; b=eZovgW9/Hgi1eWVDXqoT72dVVKWMZGQ0d6yo0kI8k/2rcZaoeNBJDOwtrxGovKJTXm vuhAz01SEvP/IbIGYP2NBVrGQjtU6MDNS/lG+vTFODyjUtxRz+ZQLzXDwLtPFpnYqLyb Mvk+U4rnv2NaSCdhuqOytFDJcTdPYmauaUCX8=
2010/7/30 Manoj Rajagopalan <rmanoj@xxxxxxxxx>:
> Hi Eigen developers,
>
> I have the following working:
>
> 1. CompactTriangularMatrix
> 2. CompactSelfAdjointMatrix
> 3. LLT for #2
> 4. Architecture (see attached crude diagram)
> (a) matrices indirectly inherit CompactTriangularStorageBase (which
> inherits EigenBase) which allows assignments, increments etc.
> (b) Wrappers to expressions (like addition, products etc.) indirectly
> inherit CompactTriangularExpressionBase (no assignment etc.)
>
> Much work remains like products and solve() but the basic evalTo(),
> addTo(), subTo(), CTOR-initialization and assignment from triangularView<> or
> selfadjointView<> of dense expressions is all working.
>
> Is this a good time for me to submit a patch?
It's always a good time to push that to a public fork on bitbucket.org.
Then when we start doing Eigen 3.1 we can discuss and pull.
Sorry I don't have time now to enter a big design discussion for Eigen
3.1 material.
> Any suggestions at this stage
> by Eigen developers will be good. As I add more, undoing and redoing stuff
> will become difficult for me.
>
> Since there is a fair amount of code involved, I was thinking that
> introducing it in small installments would make sense. Initially, just the
> triangular matrix and tests. Later, self-adjoint, and still later LLT etc..
>
> Introducing this now would let others test the code, file bugs and help me
> polish this module soon.
A fork on bitbucket allows all of this ;-)
Benoit
>
> Thanks,
> Manoj
>
>