the current implementation of logm & funm for GSoC

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

the current implementation of logm & funm for GSoC

Abd El Rahman Nour

Hi,

 

I’m currently planning to write a proposal to work on the improvement of logm, sqrtm, and funm for google summer of code.

I have one problem tho, there’s a lack of details about the currently existing implementations of these algorithms.

 

For sqrtm(), there’s a reference for a paper (written in 1999) and I’m assuming that the current implementation follows the algorithm in that paper.

There was a more recent paper published (in 2013) by the same researcher(s) that contains some improvement on the previous paper, and it’s the currently implemented algorithm in MATLAB’s sqrtm() function, as stated in the documentation on their website.

 

For logm and funm tho, octave doesn’t mention anything in the documentation, but the papers history is the same here (existing old paper, with a new and updated paper proposing improvements, written by the same person (also the same one that wrote the sqrtm papers))

 

I’m really excited to work on the project, and I’ve skimmed through both the old paper and the new paper for sqrtm

And I skimmed through the old logm and the 3 new papers (yes, there are 3, but there’s only 1 main paper) and I wanna read them thoroughly, but I just don’t know what the current implementation is, and thus idk if those new papers are implemented or not (because there was an attempt on this same project idea back in 2015, after those papers were published)

 

Please can anyone help clarify.

 

In short, my question is, what paper was followed for the implementation of logm & funm ?

(Note: just giving the name of an algorithm mentioned isn’t enough, old & new papers follow the same algorithm with the same name, just with some modification for optimization reasons, so saying “The implementation utilizes a Padé approximant and the identity” is of no use to me.

 

Thank you

- AbdulRahman



 

Sent from Mail for Windows 10

 

Reply | Threaded
Open this post in threaded view
|

Re: the current implementation of logm & funm for GSoC

PhilipNienhuis
Abd El Rahman Nour wrote

> Hi,
>
> I’m currently planning to write a proposal to work on the improvement of
> logm, sqrtm, and funm for google summer of code.
> I have one problem tho, there’s a lack of details about the currently
> existing implementations of these algorithms.
>
> For sqrtm(), there’s a reference for a paper (written in 1999) and I’m
> assuming that the current implementation follows the algorithm in that
> paper.
> There was a more recent paper published (in 2013) by the same
> researcher(s) that contains some improvement on the previous paper, and
> it’s the currently implemented algorithm in MATLAB’s sqrtm() function, as
> stated in the documentation on their website.
>
> For logm and funm tho, octave doesn’t mention anything in the
> documentation, but the papers history is the same here (existing old
> paper, with a new and updated paper proposing improvements, written by the
> same person (also the same one that wrote the sqrtm papers))
>
> I’m really excited to work on the project, and I’ve skimmed through both
> the old paper and the new paper for sqrtm
> And I skimmed through the old logm and the 3 new papers (yes, there are 3,
> but there’s only 1 main paper) and I wanna read them thoroughly, but I
> just don’t know what the current implementation is, and thus idk if those
> new papers are implemented or not (because there was an attempt on this
> same project idea back in 2015, after those papers were published)
>
> Please can anyone help clarify.
>
> In short, my question is, what paper was followed for the implementation
> of logm & funm ?
> (Note: just giving the name of an algorithm mentioned isn’t enough, old &
> new papers follow the same algorithm with the same name, just with some
> modification for optimization reasons, so saying “The implementation
> utilizes a Padé approximant and the identity” is of no use to me.
>  

See this thread for some more info:
http://octave.1599824.n4.nabble.com/GSOC-2017-Idea-Improve-logm-sqrtm-funm-td4682267.html

Philip




--
Sent from: http://octave.1599824.n4.nabble.com/Octave-Maintainers-f1638794.html

Reply | Threaded
Open this post in threaded view
|

Re: the current implementation of logm & funm for GSoC

Nir Krakauer-3
Also, you can simply view the source code and revision history for any Octave function, including logm and sqrtm, at
(use the search bar)

funm is in the linear-algebra package, whose repository is at
(the repository locations for all packages are given in https://octave.sourceforge.io/packages.php )
Reply | Threaded
Open this post in threaded view
|

Re: the current implementation of logm & funm for GSoC

PhilipNienhuis
Nir Krakauer-3 wrote
> Also, you can simply view the source code and revision history for any
> Octave function, including logm and sqrtm, at
> https://hg.savannah.gnu.org/hgweb/octave
> (use the search bar)
>
> funm is in the linear-algebra package, whose repository is at
> https://sourceforge.net/p/octave/linear-algebra/ci/default/tree/
> (the repository locations for all packages are given in
> https://octave.sourceforge.io/packages.php )

A newer funm.m (don't know if it was ever completely finished, apart from
porting to C++), is here:
https://github.com/RickOne16/matrix

Philip



--
Sent from: http://octave.1599824.n4.nabble.com/Octave-Maintainers-f1638794.html