Quantcast

[GSoC 2017 Application] Improve Iterative Methods for Sparse Linear Systems

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

[GSoC 2017 Application] Improve Iterative Methods for Sparse Linear Systems

Rui Xie
Hi Marco,

I'm a Chinese undergraduate student and I major in both mathematics and electrical engineering. I have used MATLAB a lot in the past two years for course projects. I'd like to apply for GSoC this year with Octave and I'm interested in the project "Improve iterative methods for sparse linear systems".

I have read Cristiano Dorigo's report for his work, and implemented an additional method minres in a patch at https://savannah.gnu.org/patch/?9282. It is now only a draft. Could you give me some suggestions? Thanks.


Best regards,
Rui Xie


Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: [GSoC 2017 Application] Improve Iterative Methods for Sparse Linear Systems

Marco Caliari-4
On Wed, 8 Mar 2017, Rui Xie wrote:

> Hi Marco,
> I'm a Chinese undergraduate student and I major in both mathematics and
> electrical engineering. I have used MATLAB a lot in the past two years for
> course projects. I'd like to apply for GSoC this year with Octave and I'm
> interested in the project "Improve iterative methods for sparse linear
> systems".
>
> I have read Cristiano Dorigo's report for his work, and implemented
> an additional method minres in a patch
> at https://savannah.gnu.org/patch/?9282. It is now only a draft. Could you
> give me some suggestions? Thanks.
>
>
> Best regards,
> Rui Xie
Dear Rui Xie,

I suggest you the following. First, apply Cristiano's patch you can find
here

https://savannah.gnu.org/patch/?9108 #39638

Second, add the tests you find in pcg.m to your minres.m. As you know,
minres does not require the matrix to be positive definite. Therefore, add
tests for symmetric and Hermitian matrices NOT positive definite.

Regards,

Marco
Loading...