Characterizing different motility-induced regimes in active matter with machine learning and noise.

Area of Science:

• Physics
• Soft Matter Physics
• Statistical Mechanics

Background:

• Active matter systems exhibit emergent collective behaviors not seen in equilibrium systems.
• Motility-induced phase separation (MIPS) is a key phenomenon in active matter, leading to self-organization.
• Understanding the different phases and transitions within MIPS is crucial for predicting active matter behavior.

Purpose of the Study:

• To investigate the distinct regimes within motility-induced phase separation (MIPS) in 2D run-and-tumble disk systems.
• To differentiate between active fluid, active crystal, and critical regimes using novel analytical methods.
• To compare the efficacy of machine learning and noise fluctuation analysis against traditional measures.

Main Methods:

• Utilized machine learning algorithms and noise fluctuation analysis to study MIPS.
• Developed an order parameter from principal component analysis combined with cluster stability measurements.
• Analyzed noise power spectra of average speed fluctuations.

Main Results:

• Identified three distinct regimes within MIPS: active fluid, active crystal, and critical.
• The principal component-derived order parameter effectively distinguishes these regimes.
• Machine learning demonstrated superior capture of dynamical properties compared to structural measures like maximum cluster size.
• Noise power spectra exhibited a characteristic 1/f^{1.6} signature in the critical regime.

Conclusions:

• The study successfully differentiates and characterizes distinct regimes within MIPS using advanced analytical techniques.
• Machine learning and noise analysis provide powerful tools for understanding complex dynamics in active matter.
• The findings offer a more nuanced view of MIPS, revealing critical behaviors analogous to those in condensed matter physics.