Title Information
Title
Topological Data Analysis of Collective Motion
Name: Personal
Name Part
Bhaskar, Dhananjay
Role
Role Term: Text
creator
Name: Personal
Name Part
Wong, Ian
Role
Role Term: Text
Advisor
Name: Personal
Name Part
Crawford, Lorin
Role
Role Term: Text
Reader
Name: Personal
Name Part
Darling, Eric
Role
Role Term: Text
Reader
Name: Personal
Name Part
Sandstede, Bjorn
Role
Role Term: Text
Reader
Name: Personal
Name Part
Zenit, Roberto
Role
Role Term: Text
Reader
Name: Personal
Name Part
Ziegelmeier, Lori
Role
Role Term: Text
Reader
Name: Corporate
Name Part
Brown University. Biology and Medicine: Biomedical Engineering
Role
Role Term: Text
sponsor
Origin Information
Copyright Date
2021
Physical Description
Extent
xiv, 169 p.
digitalOrigin
born digital
Note: thesis
Thesis (Ph. D.)--Brown University, 2021
Genre (aat)
theses
Abstract
Coordinated migration patterns emerge from the interactions of discrete individuals, ranging from self-propelled particles and mammalian cells at the microscopic scale to robots and animals at the macroscale, representing a fundamental problem that bridges quantitative biology, statistical physics, and data science. Historically, these rich behaviors of self-organization, phase transitions, and pattern formation have been analyzed in a problem-specific manner, which may not be widely applicable. Here, I develop an approach based on topological data analysis to visualize the spatial organization of discrete particles that migrate and proliferate in a biologically-inspired manner. Using persistent homology, the number of connected components, loops, voids, and higher dimensional holes are tracked at multiple spatial scales, thus quantifying the 'shape' of point-cloud data consisting of particle positions at each time point. This information may be represented interchangeably as a topological barcode, persistence diagram, a matrix of Betti numbers, or persistence image. Furthermore, distinct particle configurations are compared by defining a metric to compute the distance between their corresponding topological representations. Using this methodology, phase transitions experimentally observed in proliferating mammary epithelial cells (subject to various culture media and drug treatments), can be automatically classified, revealing distinct regimes of individual motility, formation of branched networks, and compact clusters, are automatically identified. In another case study, rules of interaction are inferred from simulated trajectories of collective motion in animals, including swarming, flocking and milling patterns, using a combination of topological data analysis and machine learning. Overall, the findings indicate that unsupervised topological classification can outperform unsupervised classification based on order parameters, and offers a unique insight into the multiscale dynamics that drive phase transitions in these systems. Broadly, I envision this approach can be applied to elucidate emergent behaviors arising from interacting individuals in a wide variety of biological and physical phenomena.
Subject
Topic
Machine Learning
Subject
Topic
agent-based models
Subject
Topic
topological data analysis
Subject (fast) (authorityURI="http://id.worldcat.org/fast", valueURI="http://id.worldcat.org/fast/00867354")
Topic
Collective behavior
Language
Language Term (ISO639-2B)
English
Record Information
Record Content Source (marcorg)
RPB
Record Creation Date (encoding="iso8601")
20210607
Type of Resource (primo)
dissertations