Clear Filters
Clear Filters

matlabFunction very slow

4 views (last 30 days)
Martin
Martin on 29 Jul 2011
Hi,
I am using the symbolic math toolbox to generate analytical derivatives of single valued function in approximately 190 unknowns. This works all fine (within 3 min). However, if I use matlabFunction to generate an m-file consisting of the first-order conditions, this actually takes hours... I am even more concerned by the fact that the cpu-usage is only around 0.2 percent (I have eight cores, so it would be still below 2 percent for one core).
Why is matlabFunction so slow compared to fprintf? Is it just because it tries to optimize the output? Why does matlabFunction use only a negligible fraction of the cpu?
Thanks for your answers, Best, Martin
  1 Comment
Walter Roberson
Walter Roberson on 29 Jul 2011
Related: http://www.mathworks.com/matlabcentral/answers/12491-why-is-matlabfunction-so-slow

Sign in to comment.

Answers (2)

the cyclist
the cyclist on 29 Jul 2011
What is the memory usage like? If you need to use virtual memory, the whole thing will slow down dramatically.

Martin
Martin on 29 Jul 2011
Actually, it uses virtual memory in a modest way. That was my first thought as well. This should not be the problem: matlabFunction is always quite slow even for small functions. Is there a way to write the function unoptimized?

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!