Published:
Open-source solver that tunes slow, noisy computer simulations until they match real experiment data. It builds a simple stand-in for the simulation as it goes and checks each step against it, so results can be verified rather than trusted blindly. It repeats until the fit is good enough.
Published:
Open-source tool that builds fast, simple formulas to stand in for slow, costly simulations. Because the formulas are quick to run, they can be optimized quickly.
Published:
Open-source tool that decides how much weight each piece of data deserves when tuning a model. It sets the weights automatically instead of by experience and intuition, so results are reached efficiently and are less subjective.
Published:
Open-source method that finds the best control settings for large manufacturing networks where outcomes are uncertain. It swaps the uncertain parts for simple estimates so answers arrive quickly, while staying within required limits.
Published:
Web prototype where manufacturing engineers build, optimize and compare factory designs. It draws on a reusable library of performance models and shows the trade-offs between options.