Octave

Gouzouni

Joined 3 March 2014
User page Discussion

User:Gouzouni (edit)

Revision as of 23:05, 17 March 2014

169 bytes added ,  17 March 2014
→‎Y: Your task
Line 67: Line 67:


== Y: Your task ==
== Y: Your task ==
* I have selected two tasks from the proposals list. The first is the development of the symbolic package [http://wiki.octave.org/Summer_of_Code_Project_Ideas#Rewrite_symbolic_package link ] and the second one is the improvement of the MPI package [http://wiki.octave.org/Summer_of_Code_Project_Ideas#Improve_MPI_package link ]. I would really like to work on any of these tasks but the truth is that if I had to choose I would go with the symbolic package.
* The task I have chosen is the implementation of [http://wiki.octave.org/Summer_of_Code_Project_Ideas#High_Precision_Arithmetic_Computation "High Precision Arithmetic Computation"]. The part that I would like to focus on would be the completion and test of the multi-precision class along with the implementation of several linear algebra algorithms which will support the use of multi-precision inputs.


I think that the symbolic package is one of the most usually used from the users of programs like Octave, and this is why I want to work on it.
* A rough estimated timeline for my work on the task would be :
I have build Octave's Development Sources and I am currently reading the code from the symbolic and the multi - precision package. A first step on the symbolic package will be the completion of the multi - precision class by including array and complex number support. Next step will be the development of a symbols class.


* As a rough timeline I would expect the array and complex number support of the multi - precision class to be completed by the mid - term. After that the symbols class development can begin.
First month ( beggining until the mid-term evaluation ) : adding array and complex number support on the existent multi-precision class implementation and testing them.
 
Second month : Getting basic linear algebra algorithms working natively on multi-precision matrix. Some of those algorithms would be :
Singular Value Decomposition ( SVD ),
QR factorization,
Cholesky factorization ( chol ),
Inverse matrix computation,
Eigenvalues and eigenvectors calculation,
Solving linear systems,
Root of non-linear function finder ( fzero ),
Minimum value finder ( fmins ).
 
Those algorithms will also be tested and debugged during the second month..
 
Third month : Finish what is left from previous month and add some more algorithms tested and debugged. Also check the documentation and anything else that comes up.
Gouzouni
67

edits

Retrieved from "https://wiki.octave.org/Special:MobileDiff/4565"