This example demonstrates the AIMMS multistart algorithm. The multistart algorithm can be used to find better solutions for a nonlinear program by solving the program several times using different starting points.

On the first page in this project the heath map of a non-convex objective function is plotted. On the second page the impact of starting points on the outcome is demonstrated. On the third and last page the multistart algorithm is demonstrated step by step.

This example uses the basic multistart algorithm as described in Chapter Advanced Methods for Nonlinear Programs of the AIMMS Language Reference. The dynamic multistart algorithm usually gives better results but it is more complex.


Multistart, Starting point, GMP, 3D chart, Network object.

Problem Type

NLP (small)


A zip file with this example can be downloaded here.


This example is used in a webinar about the multistart algorithm.