Order of Complexity of Code Generation from Symbolic Expressions using matlabFunction

Hello,

I am currently generating m-Functions from symbolic expressions and am wondering about the time complexity of the generation and optimization algorithm used in the function "matlabFunction(symbolic_expression)".

I created m-Code from a 7x1-Vector with symbolic expressions (Coriolis-Torques of a Robot). This took 3 days calculating time without further optimizations on the symbolic expression. Now I want to create the m-Code for a 7x7-Matrix (Coriolis-Matrix) with expressions similar to those in the aforementioned 7x1-Vector.

Is the time complexity of "matlabFunction" linear, i.e. O(n)? Then the calculations would take approximately 7 times as long in my understanding (that means 21 days).

Or is the time complexity higher due to cross-element-optimizations of the code (i.e. O(n^2) ). Then it would take about forever.

Any suggestions would be helpful to determine if I can expect a result or am just increasing my institutes electricity bill.

P.S.: I use Matlab 2014a.

Answer 1

[Moritz Schappler]

I just wanted to close this question. After 24 days I aborted the calculation. It seems, that the optimization algorithm has more than linear complexity. With Maple, the whole calculation could be done much faster.