Adaptive Image-based Classification of Circulation Transfer in Double-target Bose–Einstein Condensates
Faculty Mentor
Dr. Mark Edwards
Location
Russell Union Ballroom
Type of Research
On-going
Session Format
Oral Presentation
College
College of Science & Mathematics
Department
Department of Biochemistry, Chemistry, and Physics
Abstract
This project studies image-based classification of quantized circulation states in a double-target Bose–Einstein condensate rotation sensor. The system consists of two overlapping ring–disk targets where circulation transfer between rings depends on the applied overlap barrier strength and the frame rotation rate. I use a region-based convolutional neural network to classify the winding number from single-shot release–interference density images while sweeping the barrier potential, both with and without a central disk phase reference. The analysis emphasizes direct inference of transfer outcomes rather than precision bounds or exponential scaling. I then outline a closed-loop experimental workflow where each density image feeds a supervisory control block. This block updates barrier parameters and imaging settings to maintain performance under controlled signal-to-noise and shot-noise conditions. The result is a repeatable, autonomous measurement cycle that links circulation classification to adaptive recalibration. Such a workflow supports compact atomtronic rotation sensors suited for inertial navigation tasks in aircraft, submarines, and fluid-filled industrial platforms.
Program Description
.
Start Date
4-23-2026 3:00 PM
End Date
4-23-2026 3:15 PM
Recommended Citation
Goyal, Anish, "Adaptive Image-based Classification of Circulation Transfer in Double-target Bose–Einstein Condensates" (2026). GS4 Student Scholars Symposium. 245.
https://digitalcommons.georgiasouthern.edu/research_symposium/2026/2026/245
Adaptive Image-based Classification of Circulation Transfer in Double-target Bose–Einstein Condensates
Russell Union Ballroom
This project studies image-based classification of quantized circulation states in a double-target Bose–Einstein condensate rotation sensor. The system consists of two overlapping ring–disk targets where circulation transfer between rings depends on the applied overlap barrier strength and the frame rotation rate. I use a region-based convolutional neural network to classify the winding number from single-shot release–interference density images while sweeping the barrier potential, both with and without a central disk phase reference. The analysis emphasizes direct inference of transfer outcomes rather than precision bounds or exponential scaling. I then outline a closed-loop experimental workflow where each density image feeds a supervisory control block. This block updates barrier parameters and imaging settings to maintain performance under controlled signal-to-noise and shot-noise conditions. The result is a repeatable, autonomous measurement cycle that links circulation classification to adaptive recalibration. Such a workflow supports compact atomtronic rotation sensors suited for inertial navigation tasks in aircraft, submarines, and fluid-filled industrial platforms.
