MATLAB: Costly anonymous function declaration for solving a equation system with parameters

Question

Hi i have to solve a equation system (which is a matlab functiion)for many parameters values, so i run a loop over the parameter values, then use an anonymous function to pass the parameters to the equation system, then use fsolve to solve this anonymous function. now i understand that solving takes some time, but setting up the anonymous function takes almost the same time. why? can this be done faster?

The script looks like this

for i=1:ni
   parameters= par_fun(i);
   my_anon_fun = @my_fun(x)(x,parameters);
   [sol fval exitflag]= fsolve(my_anon_fun,starval);
end

where my_fun is the eqauation system and par_fun give the parameter vector for the equation system. The profiler shows a lot of time spent in the line my_an_fun = @myfun(x)(x,parameters);

Thanks Dominik

Answer 1

Perhaps it would be faster for you to use nested functions rather than anonymous functions. I think (thought I haven't tested this idea) that this would avoid the overhead of creating anonymous functions.

Alan Weiss

MATLAB mathematical toolbox documentation