Machine learning forecasting of active nematics

Abstract

Our model is unrolled to map an input orientation sequence (from time t-8 to t-1) to an output one (t,t + 1…) with trajectray tracing. Cyan labels are −1/2 defect while purple ones are +1/2.

Submersed Micropatterned Structures Control Active Nematic Flow, Topology and Concentration

“…Since previous studies have successfully modelled the dynamics of microtubule/kinesin-based active nematic films with weak effective friction [56], we treat the friction as negligible in the regions of the film superjacent to deep structures. In the regions above shallows, we include lubrication momentum dissipation via a non-zero effective friction coefficient.…”

Kinesin and Myosin Motors Compete to Drive Rich Multi-Phase Dynamics in Programmable Cytoskeletal Composites

“…This complex composite continuously restructures and reconfigures itself in response to the demands of the cell, to enable diverse processes from cytokinesis to mechano-sensing [3][4][5]7,8,[13][14][15][16][17][18][19][20][21] . In vitro systems of reconstituted cytoskeletal proteins, which display rich and tunable dynamics, are also intensely studied as model active matter platforms to shed light on the non-equilibrium physics underlying force-generating, reconfigurable systems 7,12,19,[22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38][39][40] .…”

Learning active nematics one step at a time

“…The use of machine learning for the analysis and prediction of high dimensional spatiotemporal dynamics is an exciting prospect (3). This is nicely demonstrated by the application of deep neural networks to the study of active-nematic systems (2,4). The machine learning-based approach is particularly beneficial in situations where a direct measurement of all the relevant degrees of freedom, or of the underlying parameters, is not possible.…”

“…Unlike liquid crystals, however, active nematics have internal driving, which produces active stresses, locally injecting energy into the system and exciting flows. A widely used realization of active nematics is networks of cytoskeletal filaments with their associated molecular motors (2,4,6,7). The latter generate active forces: Powered by ATP, they "walk" along filaments, constantly attaching and detaching, and promote filament sliding.…”

“…regions between defects (while retaining the defects), reflecting the inherent tendency of nematic constituents to locally align. Remarkably, the machine learning approach is able to predict key defect events, including generation (also called defect unbinding), motility, and annihilation (2,4). While the ability to forecast the behavior of a chaotic system is inherently limited, the machine learning approach appears to predict the short-time behavior of active-nematic systems well, outperforming the predicted dynamics calculated using a simulated system with fitted parameters (2).…”