Engineering & simulationBayesian optimization modelQuality factor over one billion; vibration persisted for about an hour
A machine learning model guided by Bayesian optimization generated a spiderweb-inspired geometry for a microchip that confined internal vibrations, which was then confirmed by simulation.
“So they have a model, the design machine learning model that's guided by optimization, and they fed this Besian optimization model. Hey, this is the parameters that we care about. We want to do the spider web thing. And after a couple iterations, It gave them a design idea.”
Engineering & simulationQuality factor over one billion; oscillation lasted about an hour
Researchers used a machine learning model to optimize the geometry of a vibrating microchip, resulting in a design that isolated vibrations and achieved a quality factor of over one billion, allowing oscillation for an hour at room temperature.
“So the model gave them this design that was kind of like a 6 string design. And what was interesting about it is that it seemed to confine the internal vibrations into like a loop within the string itself. So you have this, uh, stress strain and torsional, um, force distribution through the web that kept the entire vibration local.”