|
From: | Marco Caliari |
Subject: | Re: Status for GSoC project : Improve logm,sqrtm and funm |
Date: | Tue, 29 Mar 2016 18:33:31 +0200 (CEST) |
User-agent: | Alpine 2.10 (DEB 1266 2009-07-14) |
On Tue, 29 Mar 2016, Svante Signell wrote:
On Tue, 2015-10-20 at 18:40 +0100, Carnë Draug wrote:On 20 October 2015 at 16:07, Mudit Sharma <address@hidden> wrote:All, I would like to know the status of the project:Improve logm,sqrt,funmn as it was being done for GSoC'15. I am concerned for it as I have interests in doing it even without any funding.Hi Mudit The student stopped replying 1-2 weeks after the start and we never heard from him again. He failed on the midterm evaluation, having made pretty much no work. If you wish to work on this, you should speak with Marco Caliari (cc'ed on this email) as he has details on what algorithms to try to implement, including source code to start the work.Hi, sorry for chiming in here, but anyway. I see that the expm implementation in octave only use Pade approximation to calculate the matrix exponential. I have an implementation I made in Fortran 77 in the late 90's. It uses scaling, pade approximation and squaring.
Hi, the current Octave's expm uses Pade' with scaling and squaring, as well. The state of the art for the matrix exponential by Pade approximation isA new scaling and squaring algorithm for the matrix exponential, Al-Mohy, A. H. and Higham, N. J., 2009
which is freely available as a m-file http://eprints.ma.man.ac.uk/1442/
1) How much work is it to call that Fortran code from Octave? 2) Do you have some real difficult test examples, including execution times and memory requirements? 3) I a .c/.c++ or .m implementation preferred to Fortran?
A m-file is of course easier to maintain. A standard Pade` approximation with scaling and squaring involves only one or two short, unnested for loops. I can't imagine a Fortran/C++ implementation much faster than a m-file. Did you compare your implementation with Octave's expm? expm.m is a very short file to read and compare with.
Best regards, Marco
[Prev in Thread] | Current Thread | [Next in Thread] |