Solve non-linear problems with a guarantee of global optimality

 

We solve optimization problems that are unsolvable using conventional optimization technology.

Octeract's optimization software guarantees that the best (global) solution to large-scale MINLP problems will always be located.

What is wrong with current optimization methods?

Stochastic methods (here PSO) never guarantee convergence to the global solution

 Multistart methods don't provide this guarantee either, and the number of starting points to achieve a good resolution increases exponentially with the number of dimensions

Multistart methods don't provide this guarantee either, and the number of starting points to achieve a good resolution increases exponentially with the number of dimensions

Modern non-linear methods can never provide a guarantee of global optimality. Considering that optimization is used to find solutions to multi-billion dollar problems, it is important to find the best possible solution every time.

How is Octeract technology different?
venn_white.png

Octeract software is powered by deterministic global optimization methods, which can uniquely provide this guarantee.

Although powerful, these methods have always been too slow to solve large-scale problems because it was not known how to exploit parallel hardware.

Octeract's proprietary technology breaks through this barrier by enabling the user to invest additional computational resources to solve an optimization problem much more quickly, and still retain the guarantee of finding the, extremely valuable, best solution.