Software

MÆSTRO

Published: July 20, 2022

Mæstro (code and documentation) stands for Multi-fidelity Adaptive Ensemble Stochastic Trust Region Optimization and it is an open source plug n play derivate fee stochastic optimization solver. The problem being considered in MÆSTRO involves fitting Monte Carlo simulations that describe complex phenomena to experiments by finding parameters of the resource intensive and noisy simulation that yield the least squares objective function value to a noisy experimental data. This problem is solved using an active machine learning algorithm where in each iteration, a local approximation of the simulation signal and of the simulation noise is constructed over data, which is obtained by running the simulation at strategically placed design points within a trust-region around the current iterate. Then the simulation components of the objective are replaced by their approximations and this analytical and closed-form optimization problem is solved to find the next iterate within the trust-region. Then the trust region is moved and the iterations continue until a satisfactory convergence criteria is met.

Apprentice

Published: July 01, 2022

Apprentice (code and documentation) is an open source package for construction of multivariate analytic surrogate model for computationally expensive Monte-Carlo predictions. The surrogate model is used for numerical optimization of a prediction function since it can be prohibitively expensive to perform optimization over functions with the Monte-Carlo predictions. To summarize, Apprentice can be used for performing three tasks:

• Construct surrogate models to computationally expensive Monte-Carlo predictions
• Formulate a prediction function with surrogate models
• Perform numerical optimization over the prediction function

Outer optimization

Published: April 15, 2021

Outer optimization (CODE) is an open source package to assign weights and solve the tuning problem of finding optimal parameters that minimizes the a least-squares function between approximations of noisy simulations and experimental data or data observed in nature. Instead of setting weights manually based on experience and intuition, the weights are automatically adjusted using a bilevel optimization or a single level robust optimization formulation, thus yielding results efficiently that are less subjective.

Stochastic optimization based on deterministic approximations (SODA)

Published: February 28, 2021

Stochastic optimization based on deterministic approximations (SODA)(CODE) is an open source package containing the algorithm to perform stochastic optimization based on deterministic approximations to efficiently solve the problem of finding control settings for stochastic processes in a large manufacturing service network subject to the satisfaction of stochastic feasibility constraints.

Factory Optima

Published: February 06, 2019

Factory Optima (PAPER) is a web-based prototype system that allows manufacturing process engineers to compose, optimize and perform trade-off analysis of manufacturing service networks based on a reusable repository of performance models.