SigOpt is constantly working to extend the capabilities of our platform by supporting different types of optimization problems. In addition to the optimization of continuous parameters, we already support integer and categorical parameters, multisolution optimization, multimetric optimization, and other advanced features.
Today, we are excited to announce the general availability of Constraints, a feature that gives customers more fine-grained control of an experiment’s parameter space. Now, a subset of the continuous parameters in an experiment can be subject to a linear constraint to restrict the parameter space that SigOpt searches. It is useful for scenarios where known interdependencies between parameters mean only a region of the parameter space is valid. Beyond just defining that parameters alpha, beta, and gamma range from zero to one, customers can also inform SigOpt that alpha + beta + gamma <= 1 by providing a constraint at experiment creation. To learn more about how to use Constraints, read the SigOpt docs or check out our GitHub example.
Constraints have already helped VSA Partners, a data-driven full-service advertising and design agency, optimize a marketing mix strategy for one of their clients. They developed a marketing allocation simulator powered by a Bayesian network to approximate revenue generated based on the allocation of spend across different advertising channels and product offerings. Constraints allowed VSA Partners to define a constrained optimization problem with SigOpt and ensure that their simulator only considers feasible marketing mix configurations that stay under a total marketing budget.
Constraints are available for customers on an Enterprise plan.
If you are an organization interested in using Constraints or have feedback on this feature, please contact your account manager.