2022 Fiscal Year Annual Research Report
Metallo-Dielectric Janus Particles as Building Blocks for Designer Active Materials
| Project/Area Number |
20K03786
|
|---|
| Research Institution | Kyoto University |
|---|
Principal Investigator |
MOLINA JOHN 京都大学, 工学研究科, 助教 (20727581)
|
|---|
| Project Period (FY) |
2020-04-01 – 2023-03-31
|
| Keywords | Active Particles / Fluid Dynamics / Machine Learning / Flow Inference / Multi-Scale Simulations |
|---|
| Outline of Annual Research Achievements |
As part of an ongoing collaboration, we have focused on active particles in complex environments, i.e., at fluid interfaces and in viscoelastic fluids, typical of biological systems. For the former, we investigated how the swimming type, interface curvature, viscosity ratio, and incoming angle determine the swimmer dynamics. For the latter, we studied the origin of the velocity enhancement/decrease for microswimmers with swirl in viscoelastic fluids. Finally, the work on the collective motion of Quincke rollers, explaining the role of near/far-field hydrodynamics and electrostatics in the different collective modes observed experimentally, has been published.
Furthermore, we have continued the development of Machine Learning (ML) methods to solve Soft-Matter flow problems. First, we used Bayesian ML to develop a Stokesian inference framework to solve forward/inverse problems using partial and/or noisy data. For example, to reconstruct the flow field and infer the location of boundaries using sparse velocity measurements. Second, we expanded our method for learning constitutive relations of complex fluids. It was originally tested on non-interacting polymer models (i.e., linear relations), but has now been applied to entangled polymers (i.e., non-linear relations). Simulating 2D flow problems shows an order of magnitude speed increase compared to Multi-Scale Simulations, with no loss of accuracy. Finally, using reinforcement-learning we have also studied how active/swimming particles can learn to navigate complex flows using local hydrodynamic signals.
|
