Note: If the Randomize
optimization sequence checkbox is cleared, each optimization will use the
same initial complex and hence will converge to the same solution.
For a well-behaved problem with a single optimum solution, GoldSim can be counted on to locate the optimum. For more complex problems with multiple local optima, however, the solution found by GoldSim may not necessarily be the true optimum (it could be a local optimum).
For complex models with local optima, the choice of bounds and initial values for the optimization variables can be important, and can determine whether GoldSim converges to a local optimum, rather than to the global optimum.
The best way to convince yourself that you have found a global, rather than a local, optimum, is to run the optimization multiple times with different initial complexes.
In order to ensure that GoldSim creates a different initial complex every time it starts an optimization, the Randomize optimization sequence checkbox should be checked (which is the default). If you then run multiple optimizations (by pressing the Optimize! button repeatedly), GoldSim will rerun the optimization with a different initial complex each time.
Note: If the Randomize
optimization sequence checkbox is cleared, each optimization will use the
same initial complex and hence will converge to the same solution.
Although it is impossible to be sure that you have found a global optimum, if you repeat the optimization multiple times (randomizing the initial complex) and converge to the same solution, your confidence that you have found a global optimum will certainly be increased.