Quantcast

GSoC 2017 - ode15s : Matlab Compatible DAE solver

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 - ode15s : Matlab Compatible DAE solver

Ivan Gonzalez Polanco
Hello, I'm Ivan Gonzalez, and I'm looking forward to work on the implementation of a Matlab compatible adaptative BDF solver.
To start getting familiar with the project, I have a few questions:
1)What does it mean that the solver must be adaptative? (as in, being able to change it's stepsize on a whim, or to adapt to different kinds of DE?)
2)As the SUNDIALS library will be used, and has both C and a MEX interface, would the backend of the solver have to be C-only or MEX-only? as I understand now, MEX files are only C files that are read as scripts, without having to be compiled.

Greetings,
Ivan.
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: GSoC 2017 - ode15s : Matlab Compatible DAE solver

Francesco Faccio
2017-03-08 8:11 GMT+01:00 Ivan Gonzalez Polanco <[hidden email]>:
Hello, I'm Ivan Gonzalez, and I'm looking forward to work on the implementation of a Matlab compatible adaptative BDF solver.
To start getting familiar with the project, I have a few questions:
1)What does it mean that the solver must be adaptative? (as in, being able to change it's stepsize on a whim, or to adapt to different kinds of DE?)
2)As the SUNDIALS library will be used, and has both C and a MEX interface, would the backend of the solver have to be C-only or MEX-only? as I understand now, MEX files are only C files that are read as scripts, without having to be compiled.

Greetings,
Ivan.
 
Hi Ivan,

the description of the project is obsolete and need to be updated. I worked on this project during GSoC 16 and the implementation is already included in Octave core [1].

@Carlo, could you please update the project description?

Regarding your questions:
1) being adaptive means that the solver is variable-step (actually it's also variable order)
2) we developed  __ode15__ , a C++ interface for ode15i and ode15s. MEX files are specific for Matlab architecture and very slow in Octave in comparison to oct-files, so we decided to use the C / C++ interface

If you are still interested in this project I suggest you to follow this discussion [2]. A new project description will be soon updated.

Sorry for the wrong description,

Francesco Faccio


[1]http://hg.savannah.gnu.org/hgweb/octave/file/4890b1c4a6bd/libinterp/dldfcn/__ode15__.cc
http://hg.savannah.gnu.org/hgweb/octave/file/4890b1c4a6bd/scripts/ode/ode15i.m
http://hg.savannah.gnu.org/hgweb/octave/file/4890b1c4a6bd/scripts/ode/ode15s.m

[2]http://lists.gnu.org/archive/html/octave-maintainers/2017-01/msg00196.html
Loading...