Machine learning for phase behavior in active matter systems
- Creators
-
Dulaney, Austin R.
-
Brady, John F.
Abstract
We demonstrate that deep learning techniques can be used to predict motility-induced phase separation (MIPS) in suspensions of active Brownian particles (ABPs) by creating a notion of phase at the particle level. Using a fully connected network in conjunction with a graph neural network we use individual particle features to predict to which phase a particle belongs. From this, we are able to compute the fraction of dilute particles to determine if the system is in the homogeneous dilute, dense, or coexistence region. Our predictions are compared against the MIPS binodal computed from simulation. The strong agreement between the two suggests that machine learning provides an effective way to determine the phase behavior of ABPs and could prove useful for determining more complex phase diagrams.
Additional Information
© The Royal Society of Chemistry 2021. Submitted 19 Feb 2021; Accepted 23 Jun 2021; First published 28 Jun 2021. A. R. D. would like to thank Yisong Yue for thoughtful discussions pertaining to graph neural networks. We gratefully thank S. A. Mallory for providing a majority of the simulation data used in this work. J. F. B. acknowledges support by the National Science Foundation under Grant No. CBET-1803662. We gratefully acknowledge the support of the NVIDIA Corporation for the donation of the Titan V GPU used to carry out this work.Attached Files
Submitted - 2011.09458.pdf
Files
Name | Size | Download all |
---|---|---|
md5:28ad0f87ff9b866a840b93c9cd862354
|
7.3 MB | Preview Download |
Additional details
- Eprint ID
- 107347
- Resolver ID
- CaltechAUTHORS:20210106-130209799
- NSF
- CBET-1803662
- NVIDIA Corporation
- Created
-
2021-01-06Created from EPrint's datestamp field
- Updated
-
2021-07-26Created from EPrint's last_modified field