Hi,

I have been working on the dense matrix for the last few days. The PR is [1]

I have written some Profile results in [2]. Here I draw some conclusions.

There is an absolute need of a more efficient data structure to store and 
use dense matrices. The current implementation with all of the 
 _handle_creation_matrix, copy and lambda calls seems to be bottlenecks to 
me.  I think that for internal use, matrices can be represented just as 
list of lists like if we call Matrix.tolist(). What is really the issue now 
is how the matrices should be represented in top levels, we can't just 
return list of lists. Should this issue be deferred for above levels, where 
we would have a unified representation of all the matrices and all the user 
interface stuff? Or should I tackle this problem right now.

Also, it seems that there is no need to enter the individual elements like 
ZZ(3) instead of 3, the results are no different. The functions make the 
calls themselves in their course of execution. 

I now believe that the addition of domains along with a new data structure 
will help matrices improve it's speed a lot.

-Saurabh

[1] https://github.com/sympy/sympy/pull/2248
[2] http://sympymatrix.blogspot.in/2013/07/profile-results-of-dense-matrix.html

-- 
You received this message because you are subscribed to the Google Groups 
"sympy" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To post to this group, send email to [email protected].
Visit this group at http://groups.google.com/group/sympy.
For more options, visit https://groups.google.com/groups/opt_out.


Reply via email to