Aimo Törn: Probabilistic Algorithms


Prev   11   Next


Optimization

As a probabilistic local optimization algorithm you implemented Timo Järvi's Unirandi, which is a random direction, line search algorithm.

As a global optimization algorithm you implemented multistart and in your assignment your own version of Törn's Topographical Global Optimization Method.

You tested your algorithm on some standard test problems one of which is the so called Griewank function

  f(x) = 1/d(x12 +..+xn2) -
cos(1-1/2xi+1)·..· cos(n-1/2xi+1)
where n=10, d=4000, box: [-600,600]10. The function has several thousand local minima in the box with the global minimum f(0)=0.