Re: [eigen] New Levenberg Marquardt stuff

[ Thread Index | Date Index | More Archives ]

So, I have two main use cases:

1) I just want basic linear algebra stuff like GEMM and LU
factorization. Eigen is overkill for that and typically in these cases
the code is pure C (we actually still have an old AIX box which has such
an old C++ compiler, most template stuff won't compile). In this case
I'll just call BLAS/Lapack directly or roll my own short routines.

2) I need super advanced features like MRRR tridiagonal eigensolver, or
Lapack's (much faster) SVD or GEP solvers, which Eigen doesn't
implement. So I have to call BLAS/Lapack directly.

I'm still sitting on this mailing list in hopes that some charitable
soul will spend the man hours to port those advanced features :).

On 12/07/2012 02:59 PM, Thomas Capricelli wrote:
> Il 07/12/2012 23:44, Victor Liu ha scritto:
>> I have often thought about using Eigen for my own scientific computing
>> projects, but each time I have been deterred by what I consider its
>> enormous size (don't even get me started about Boost)
> Eigen big ? I disagree totally. if you remove doc/tests, the code is
> worth 3.5Mb (Eigen/src). That's NOTHING compared to what it does. And
> it's only includes...
> I'm not the author, i'm allowed to say how great it is :-)
> Thomas

Mail converted by MHonArc 2.6.19+