Numerical optimization using high-performance compute on AWS

As discussed in the previous sections, most of the numerical optimization problems are NP-hard and highly compute-intensive for finding a reasonable solution. The software tool employing these algorithms has to carry out a large-scale search over a very complicated multidimensional objective function in search of the global optimum. Because of the complexity, the number of dimensions, non-convexity, and sometimes discontinuities present in these objective functions, it is almost impossible to find the global optimum in a finite amount of time, even with today’s compute resources.

However, for most of these problems, several commercially available and open source software tools find a very good solution (local optimum) in a reasonable amount of time. These tools can be run on the infrastructure and compute resources provided by AWS. Let’s discuss some of the common commercial and open source tools that can...