globalMOO® makes GMOO, multi-objective optimization software. GMOO finds the inputs to a model that produce the outcomes you want, across many conflicting objectives at once.
globalMOO is a company that makes GMOO, multi-objective optimization software. GMOO finds the inputs to a model that produce the outcomes you want, across many conflicting objectives at once. It works with any model that computes outputs from inputs, including spreadsheets, physics-based simulators, analytical equations and AI/ML models.
GMOO solves the inverse problem. Conventional tools run a model forward again and again, trying inputs until the outputs look right. With GMOO you state the outcomes you want and it back-calculates the inputs that achieve them.
It first learns how your model behaves from a set of training runs, typically 50 to 2,000. After that, each optimization takes seconds, and you can change your objectives without retraining. For each objective you can ask GMOO to match a target value, minimize it or maximize it.
The workflow has five steps:
GMOO is model-agnostic. Models can be black-box and nonlinear, and they don't need to be continuous or differentiable. Inputs and outputs can be continuous, integer, logical or categorical, and objectives can be in different units and scales.
There are three requirements:
GMOO scales to hundreds of input variables and thousands of objectives, using basic multi-core parallelism.
Most optimization methods were designed for a single objective. They handle many objectives in one of two ways, and both have drawbacks:
GMOO optimizes all objectives at once without weighting them. It returns a set of viable solutions rather than a single answer, so decision-makers can choose among real trade-offs.
In globalMOO's published benchmark on 30-variable problems, MOEA/D and DNSGA2 needed up to 100,000 iterations. GMOO needed fewer than 100 iterations, using up to 1,000 training cases.
Compared with Excel Solver, GMOO handles many objectives instead of a single target cell, and it doesn't fail on the non-convex problems that are common in real spreadsheets.
GMOO also performs impact factor calculation. It quantifies how much each input variable influences each outcome. Organizations use this to understand cause and effect in their models and to measure hidden biases in AI and machine-learning models.
Published use cases include:
| Plan | Price | Input variables | Outputs |
|---|---|---|---|
| Trial | Free for one month | 5 | 40 |
| Lite | $30 per month or $300 per year | 20 | 160 |
| Enterprise | Custom | 200+ | 640+ |
All plans include the Excel add-in. Academic, student and product-integration licenses are also available. See pricing for current details.
Use the contact form or email contact@globalmoo.com. A free trial is available at app.globalmoo.com.