Boost logo

Boost :

Subject: [boost] Fwd: Re: [gsoc18][ublas] Proposal to add advanced matrix operations
From: Mike Gresens (mike.gresens_at_[hidden])
Date: 2018-01-20 11:21:37

Hi Mike,

Apart from the operations I was proposing, the idea of solving
polynomials via QR reduction seems a great idea to me. I can add this
module in the list of operations to be added to Boost.ublas in my proposal.
I am looking for someone to mentor me on this project.

Shikhar Srivastava

On Sat, Jan 20, 2018 at 10:34 AM, SHIKHAR SRIVASTAVA
<shikharsri1996_at_[hidden] <mailto:shikharsri1996_at_[hidden]>> wrote:

    Hi Mike,

    It looks like some of the operations that I suggested were already
    implemented as a part of GSOC15, though never got merged into the
    main branch.

    My Idea was to add generalised modules for the operations (Now) -

     1. LU and Cholesky
     2. QR and QL
     3. SVD

    On Fri, Jan 19, 2018 at 9:14 PM, Mike Gresens
    <mike.gresens_at_[hidden] <mailto:mike.gresens_at_[hidden]>>

        Hi Shikhar,

        does this include something for solving polynomials (via
        balanced-QR reduction of the companion matrix)?


        Best regards,

        Am 19.01.2018 um 07:37 schrieb SHIKHAR SRIVASTAVA via Boost:

            Hi everyone,

            I am a 4th year undergraduate student pursuing a degree in
            Computer Science
            and Engineering. I have strong programming experience in C++
            internships, self projects and programming events. I wish to
            be a part of
            gsoc18 under boost and am particularly interested in the
            linear algebra
            library Boost.ublas.

            The ublas library can be made more useful for Machine
            Learning applications
            like recommendation systems, clustering and classification,
            recognition by adding some operations required in those.
            I propose to add advanced matrix operations to ublas including -

             Â    1. Triangular Factorisation (LU and Cholesky)
             Â    2. Orthogonal Factorisation (QR and QL)
             Â    3. Operations to find Singular Value lists
             Â    4. Eigenvalue algorithms
             Â    5. Singular Value Decomposition (SVD)
             Â    6. Jordan Decomposition
             Â    7. Schur Decomposition
             Â    8. Hessenberg Decomposition

            This is a very brief description of what I would like to
            propose. Any
            suggestions that help in refining the requirements before
            making a draft
            proposal  would be great.

            Shikhar Srivastava

            Unsubscribe & other changes:

Boost list run by bdawes at, gregod at, cpdaniel at, john at