Inference in particle tracking experiments by passing messages between images.

Area of Science:

• * Physics
• * Computer Science
• * Information Theory

Background:

• * Particle tracking is crucial in various scientific fields, but traditional methods struggle with high particle density and motility.
• * Existing techniques based on proximity in time-consecutive images face trajectory uncertainties.
• * Challenges increase in complex environments like turbulent flows due to strong particle transport.

Purpose of the Study:

• * To develop an efficient algorithm for learning particle dynamics parameters from image data.
• * To address limitations of current particle tracking methods in high-density and high-motility scenarios.
• * To provide a fast and reliable tool for particle tracking applications.

Main Methods:

• * Implemented a belief propagation (BP) algorithm, a type of message-passing algorithm.
• * Designed the algorithm for distributed computation, enabling parallel processing.
• * Tested the BP algorithm on particle tracking in turbulent flow simulations.

Main Results:

• * The BP algorithm demonstrated high accuracy, comparable to exact Markov Chain Monte Carlo methods.
• * BP significantly outperformed other methods in terms of speed.
• * A random distance model was proposed and analyzed to support BP's accuracy.

Conclusions:

• * The developed BP algorithm offers an efficient and reliable solution for particle tracking.
• * The distributed nature of the algorithm facilitates scalable computation.
• * This method provides a systematic approach to particle tracking with broad applicability.