simulannealbnd function output varies every time it is run

simulannealbnd function gives differnt optimal values and different optimal points when it is run multiple times.

Why is it so happening? Why it will not converge to a global optimal point?

Answer 1

[Walter Roberson]

Simulated Annealing is probabilistic by definition. It would not be uncommon for it to get stuck in a local minimum and not find a narrow global minimum.