Renzhi Sun
Advisor: Dr. Chunhui Du, School of Physics, Georgia Institute of Technology
Defect Spin Quantum Sensing of Magnetic Dynamics in Low-Dimensional Materials
Monday, October 12, 2026
11:00 a.m.
Howey W401
Committee Members:
Dr. Michael Chapman, School of Physics, Georgia Institute of Technology
Dr. Zhigang Jiang, School of Physics, Georgia Institute of Technology
Dr. Hailong Wang, School of Physics, Georgia Institute of Technology
]]>
Sana Aminnaji
School of Biological Sciences
Prospective coding of task state in cerebellar Purkinje cell complex spikes
Wednesday, October 7, 2026, at 9:00 am
In Person Location: Cherry Emerson 204
Open to the Community
Advisor:
Dr. Farzaneh Najafi (School of Biological Sciences)
Committee Members:
Dr. Timothy C. Cope (School of Biological Sciences)
Dr. Audrey Sederberg (School of Psychology, School of Physics)
Dr. Hannah Choi (School of Mathematics)
Dr. Dobromir Rahnev (School of Psychology)
Abstract:
The cerebellum is the brain's canonical learning machine. In the classical account, climbing fibers from the inferior olive deliver an error signal to Purkinje cells, evoking complex spikes that trigger plasticity and correct the next movement. This account is retrospective: the complex spike reports what has already gone wrong. My thesis asks whether these same signals also look forward — whether, before an expected event, Purkinje cell activity reflects what the animal expects and how confidently it expects it. I test this using two-photon calcium imaging of Purkinje cell dendrites in behaving mice, across two paradigms that share animals and imaging field while manipulating certainty by design.
First, I characterize how mice solve a timing task and how complex spikes relate to their internal estimate of its rule. Head-fixed mice produce a two-push joystick sequence whose required interval alternates unpredictably between short and long blocks, so the animal must infer which rule is in force. Mice adapt at trial, block and session timescales, and state-space modeling shows they switch among a small number of timing strategies rather than drifting continuously. In the same animals, complex spike responses separate into groups with distinct sensory, movement and reward profiles that reproduce the known microzonal organization of climbing fiber input. In the quiet interval between trials — containing no stimulus, no required movement and no reward — activity in the movement- and reward-related groups is lower before trials that will be rewarded than before trials that will not, in every animal and every task variant, and above a directly measured slow-drift baseline.
Second, I ask whether this signature survives without movement or reward. The same mice passively viewed or heard a four-element stimulus sequence under fixed inter-stimulus intervals, where the timing of the next stimulus is certain, and jittered intervals, where it is not. Preliminary results indicate that activity preceding a stimulus is suppressed under certainty, in the same class of response group, with nothing to move toward and nothing to earn. I will extend this to the full dataset and test whether the depth of suppression scales with how predictable the upcoming stimulus is.
Third, I will test whether this activity is used rather than merely present. Using optogenetic inactivation of cerebellar output restricted to the interval before a trial, with interleaved within-session controls, I will ask whether the behavioral adjustment that normally follows depends on it, and whether that dependence is specific to the anticipatory window.
Together, these aims ask whether the climbing fiber sits not at the end of a learning loop reporting error, but at its beginning, expressing expectation.
]]>
Doctor of Philosophy in Physics
School of Physics Thesis Dissertation Defense
Sara Sloman
Dr. Chandra Raman, School of Physics, Georgia Institute of Technology (Advisor)
Understanding and Controlling Nonlinear Dynamics in Magneto-Optical Traps
Date: Wednesday, October 7, 2026
Time: 9:00 a.m.
Location: Howey W401
Virtual: https://teams.microsoft.com/meet/233914907052583?p=aGoTcFfvh6PPGVoUl1
Meeting ID: 233 914 907 052 583 / Passcode: wW26HL3u
Thesis Committee:
Dr. Colin Parker, School of Physics, Georgia Institute of Technology
Dr. Zeb Rocklin, School of Physics, Georgia Institute of Technology
Dr. Carlos Sa De Melo, School of Physics, Georgia Institute of Technology
Dr. Bryan Gard, Georgia Tech Research Institute
Abstract:
Magneto-optical traps (MOTs) are overdamped systems, yet they can be prone to nonlinear dynamics and instabilities under realistic lab conditions. With the increasing importance of cold trapped atoms in atomic physics, it has become vital that we develop an understanding of these behaviors, which limit size and overall quality of the MOT. This dissertation investigates the nonlinear dynamics in a sodium-23 MOT through experimental methods. By varying the location of the trap center, stable and unstable states are reached, revealing a controllable method for tuning between these states. High-speed imaging is used classify these states, and bistable states are resolved among these dynamics at the threshold of instability. To further explore the nonlinear nature of MOTs, an analogy is drawn between the trapped atoms and low Reynolds number fluids, where inertia is negligible. A key feature of low Reynolds number systems is kinematic reversibility, the total reversal of the trajectory of a particle with the reciprocal application of an external force. This dissertation presents the first demonstration of kinematic reversibility in a cold atom fluid, which is achieved using controlled magnetic field ramps. The conditions under which reversibility breaks down are investigated and hypothesized to be due to a fluid-jamming behavior. By controllably tuning the trapping geometry due to beam misalignments, we determine a way to induce the breakdown of reversibility, providing further insight into the mechanisms behind this.
]]>
will propose a doctoral thesis entitled,
Low temperature synthesis of MAX-phase Ti2AlC coatings and their performance as hydrogen isotope permeation barriers for fusion technologies
On
Monday, October 12 at 12 p.m.
Pettit Microelectronics Building Room 102A
and/or virtually via MS Teams
Abstract
Realizing practical fusion energy requires materials that can withstand simultaneous neutron irradiation, thermal cycling, and corrosive molten salts without degrading. Conventional barriers such as Al2O3 achieve high reduction factors in laboratory conditions but degrade irreversibly under neutron irradiation and thermal cycling. This work exploits the unique defect tolerance and self-healing behavior of the nanolaminated MAX-phases, specifically Ti2AlC, to address the hydrogen isotope permeation control problem in fusion blanket design. Aim 1 establishes a simplified bilayer synthesis approach combining an aluminum precursor layer with a TiAlC layer, enabling Ti2AlC formation at 650 °C through controlled Al diffusion, reducing the temperature requirements of conventional synthesis while eliminating the per-layer calibration challenges of prior methods. Aim 2 adapts this route to Grade 91 steel, a fusion-relevant structural substrate, and provides the first quantified permeation reduction factors for Ti2AlC-containing coatings through gas-driven deuterium permeation experiments, demonstrating one order of magnitude reduction in hydrogen isotope flux. Aim 3 evaluates barrier durability by measuring coating performance and structure before and after heavy-ion irradiation and molten salt exposure, establishing whether the predicted radiation tolerance and corrosion stability of MAX-phases hold in thin-film form under fusion-relevant conditions. Together, these efforts validate a pathway for engineering MAX-phase permeation barriers on structural materials within realistic fusion reactor thermal budgets.
Committee
]]>
Date: October 7, 2026
Time: 1-3 PM ET
Location: Price Gilbert 4222
Zoom link:
https://gatech.zoom.us/j/98261111313?pwd=xtWSnKAbZJ0ya4ui86blMrJ76ZOqfM.1
Zoom passcode: 599523
Abdelrahman Sharafeldin
Machine Learning PhD Student
The Wallace H. Coulter Department of Biomedical Engineering
Georgia Institute of Technology
Committee
1 Dr. Hannah Choi (Advisor), School of Mathematics, Georgia Institute of Technology
2 Dr. Nabil Imam, School of Computational Science and Engineering, Georgia Institute of Technology
3 Dr. Anqi Wu, School of Computational Science and Engineering, Georgia Institute of Technology
4 Dr. Simon Sponberg, School of Physics, Georgia Institute of Technology
5 Dr. Chethan Pandarinath, Department of Biomedical Engineering, Georgia Institute of Technology
Abstract
Intelligent behavior requires organisms to learn useful representations of the world, seek informative observations, and act effectively despite uncertainty and limited resources. This thesis investigates how predictive generative perception, information-seeking action, and reward-guided learning support these abilities under biological constraints such as embodiment, energy efficiency, and neural architecture. In the first part of this thesis, we develop an active-sensing framework that uses expected reductions in the uncertainty of a generative model to guide exploration, learning environmental structure and visual representations that improve data efficiency in subsequent tasks. In the second part, we extend this framework to multiple sensory modalities and show that the learned visual and mechanosensory information landscapes account for features of hawkmoth flower probing and tracking. We also develop a mode-switching control algorithm that uses perceptual uncertainty to coordinate exploration and reward seeking and improves learning and generalization in simulated nectar search. In the third part, we ask how these computational objectives for perception and action can be instantiated in cortical circuits and develop a cell-type-specific model that combines predictive coding, reinforcement learning, and activity costs under biological connectivity constraints. Trained on a naturalistic change-detection task, this model reproduces several experimentally observed responses to novelty across different cell types without fitting neural response data. These complementary studies connect how information is acquired, represented, and used, offering insights into biological computation and principles for designing artificial agents that learn through interaction.
]]>
will propose a doctoral thesis entitled,
Low temperature synthesis of MAX-phase Ti2AlC coatings and their performance as hydrogen isotope permeation barriers for fusion technologies
On
Monday, October 12 at 12 p.m.
Pettit Microelectronics Building Room 102A
and/or virtually via MS Teams
Abstract
Realizing practical fusion energy requires materials that can withstand simultaneous neutron irradiation, thermal cycling, and corrosive molten salts without degrading. Conventional barriers such as Al2O3 achieve high reduction factors in laboratory conditions but degrade irreversibly under neutron irradiation and thermal cycling. This work exploits the unique defect tolerance and self-healing behavior of the nanolaminated MAX-phases, specifically Ti2AlC, to address the hydrogen isotope permeation control problem in fusion blanket design. Aim 1 establishes a simplified bilayer synthesis approach combining an aluminum precursor layer with a TiAlC layer, enabling Ti2AlC formation at 650 °C through controlled Al diffusion, reducing the temperature requirements of conventional synthesis while eliminating the per-layer calibration challenges of prior methods. Aim 2 adapts this route to Grade 91 steel, a fusion-relevant structural substrate, and provides the first quantified permeation reduction factors for Ti2AlC-containing coatings through gas-driven deuterium permeation experiments, demonstrating one order of magnitude reduction in hydrogen isotope flux. Aim 3 evaluates barrier durability by measuring coating performance and structure before and after heavy-ion irradiation and molten salt exposure, establishing whether the predicted radiation tolerance and corrosion stability of MAX-phases hold in thin-film form under fusion-relevant conditions. Together, these efforts validate a pathway for engineering MAX-phase permeation barriers on structural materials within realistic fusion reactor thermal budgets.
Committee
School of Physics Thesis Proposal
Aniruddha Datta
Advisor: Dr. Shiladitya Banerjee, School of Physics, Georgia Institute of Technology
Adaptive Navigation in Noisy Environments
Friday, October 2, 2026
10:00 a.m.
Howey W401
Committee Members:
Dr. Howard Kim, School of Physics, Georgia Institute of Technology
Dr. Itamar Kolvin, School of Physics, Georgia Institute of Technology
Dr. Peter Yunker, School of Physics, Georgia Institute of Technology
]]>
Advisor: Dr. Koki Ho
Milestone: PhD Thesis Proposal
Degree Program: Aerospace Engineering
Title: Autonomous Landing and Proximity Operation Technology for Poorly-Characterized Small Bodies
Abstract: Small bodies such as asteroids and comets are increasingly central to scientific discovery and commercial development, offering opportunities for exploring planetary defense strategies, insights into the formation of the early solar system, and access to extraterrestrial resources via asteroid mining. Yet their weak, irregular gravity fields, uncertain surface properties, and poorly characterized dynamical environments make autonomous landing and proximity operations uniquely challenging. This thesis advances autonomous small‑body mission capability through three main contributions. First, a time‑of‑flight–free model predictive control (MPC) framework is developed that enables safe, precision landing on irregular asteroid surfaces, thereby improving robustness to environmental uncertainty. This work additionally explores several varying cost function formulations within the MPC to demonstrate applicability to varying mission landing scenarios. Second, this thesis introduces a remotely guided three‑spacecraft mothership–daughtership formation MPC architecture designed for tomographic characterization of asteroid (99942) Apophis with mothership-based measurements to inform daughtership thrust commands in order to maintain antipodal orbital states amidst gravitational and thruster uncertainties. Lastly, this thesis formulates a multi‑lander, multi‑landing‑site MPC sample‑collection strategy that coordinates distributed landers to acquire surface samples from spatially diverse regions of an asteroid, increasing scientific return and supporting future resource‑utilization missions amidst similar dynamic and thrust uncertainties. Together, the contributions in this thesis proposal expand the capabilities of autonomous guidance, navigation, and control systems with its novel MPC algorithm and cost function formulations to improve the robustness, resilience, and scientific capability of spacecraft operating in the vicinity of poorly-characterized small bodies.
Date and time: 2026-10-23, 9am-11am ET
Location: CODA C0915 Atlantic
Committee:
Dr. Koki Ho (advisor), School of Aerospace Engineering
Dr. Brian Gunter, School of Aerospace Engineering
Dr. John Christian, School of Aerospace Engineering
,
,
,
Title: Machine Learning Methods for Quality Monitoring in Manufacturing Applications
Date: Monday, October 19th, 2026
Time: 2:00 pm – 3:30 pm ET
Location: George Tower 1740
Microsoft Teams meeting
Join: https://teams.microsoft.com/meet/235126064810853?p=X0W0PNo3pdabWh1VnI
Meeting ID: 235 126 064 810 853
Passcode: H3rd2n9D
Committee:
Dr. Kamran Paynabar (co-Advisor), H. Milton Stewart School of Industrial and Systems
Engineering
Dr. Jianjun Shi (co-Advisor), H. Milton Stewart School of Industrial and Systems Engineering
Dr. Jing Li, H. Milton Stewart School of Industrial and Systems Engineering
Dr. Xiaochen Xian, H. Milton Stewart School of Industrial and Systems Engineering
Dr. Mostafa Gahrooei, Industrial & Systems Engineering at the University of Florida
Abstract:
Modern manufacturing systems are increasingly driven by complex, high-dimensional, and heterogeneous data collected from distributed production lines and advanced sensing technologies. These data-rich environments present new challenges for process modeling and monitoring, particularly in short-run and customized production settings where limited samples and variability across sites make traditional approaches ineffective. To address these challenges, this thesis develops a series of statistical and machine learning frameworks for high-dimensional process modeling and monitoring under practical constraints such as data heterogeneity, nonlinearity, and privacy preservation. Together these methodologies aim to advance the state of quality monitoring in modern manufacturing environments.
In chapter 2, a federated linear mixed-effects model is proposed to learn process parameters for short-run production settings. Short-run productions are common in manufacturing systems which require customization and flexibility under limited demand and resources. However, traditional statistical process control charts cannot handle such scenarios, since they either ask for data transformation or have strict assumptions which misalign with real-world applications. To address these limitations, we propose a federated framework to estimate process parameters across multiple manufacturing sites and create process monitoring baseline models for each line. Specifically, the restricted maximum likelihood method is used to estimate multivariate linear mixed model through a federated Expectation-Maximation (EM) algorithm. The fixed effect of the model captures the commonality amongst different manufacturing sites, while the random effect captures any heterogeneity between the sites. By using a federated setting to estimate the parameters, local data is kept at each of the manufacturing sites, allowing for increased data privacy and reducing the amount of bandwidth resources needed to transmit information between the local and global servers. Finally, we validate the model by using simulations and a case study using semi-generated paper mill manufacturing data.
Chapter 3 proposes two methods to detect changes in high-dimensional model relationships, also called profiles. Advancements in sensor and data collecting technology have resulted in data sets that are both high dimensional and heterogeneous, with forms such as scalars, waveform signals, images, videos, and 3D point clouds. These advancements have created the need to construct and monitor statistical models that are able to model the relationship between heterogeneous and high dimensional inputs and a high dimensional output. This chapter (1) discusses how to model the relationship between such inputs and output using Multiple Tensor-on-Tensor Regression (MTOT) and (2) presents different approaches for how to monitor this model relationship to detect potential changes over time. Monitoring the model relationship can be done either by monitoring the core tensor of the MTOT model and the residuals of new data samples. Both approaches retain the tensor structure of the profile – something that is lost if dimension reduction is done directly on the inputs and output data. This methodology is validated through several numerical studies including monitoring overlay error from semiconductor wafer shape data.
Finally, Chapter 4 proposes a robust nonlinear tensor on tensor regression model. To do this, a neural network is constructed that can capture the nonlinear relation between tensor inputs and a tensor output where the data is assumed to contain noise. The neural network architecture is an autoencoder with an encoder and decoder that uses tensor train layers (TT-layers) with rectified linear unit (ReLU) activations. The TT-layers enable large-scale tensor mapping with a reduced number of parameters while still preserving the tensor structure, while the ReLU activations capture the nonlinearity of the relationship between the network input and output tensors. Three different loss functions, Huber Loss, Log-Cosh Loss, and MSE Trimmed Loss, are evaluated through a series of numerical studies to explore performance across different noise data scenarios.
]]>Date: Tuesday, October 6th, 2026
Time: 3:00 PM - 4:30 PM EDT
Location: MRDC 3515
Zoom: https://gatech.zoom.us/j/96846452330?pwd=oGfmTGbPoLo2zyp5ulwGl9PjECLl0Y.1
Zhigen Zhao
Ph.D. Student
Institute for Robotics & Intelligent Machines
Georgia Institute of Technology
Committee members
Dr. Ye Zhao (advisor): Woodruff School of Mechanical Engineering, Georgia Institute of Technology
Dr. Jiachen Li: School of Industrial and Systems Engineering and Woodruff School of Mechanical Engineering, Georgia Institute of Technology
Dr. Shreyas Kousik: Woodruff School of Mechanical Engineering, Georgia Institute of Technology
Dr. Sonia Chernova: School of Interactive Computing, Georgia Institute of Technology
Dr. Shiqi Zhang: School of Computing, Binghamton University, State University of New York
Abstract
Robots deployed in unstructured, human-centric environments must carry out long-horizon tasks that couple discrete decisions with contact-rich continuous motion. Task and Motion Planning (TAMP) addresses this by decomposing the problem into a discrete task plan and a continuous motion plan, but classical TAMP requires re-engineering for each new task or environment, scales poorly with problem size and horizon, and accumulates no experience across deployments.
This proposal builds toward generalizable, scalable, and experience-driven robot autonomy by progressively replacing the hand-engineered layers of classical TAMP with learning-based counterparts. First, we formulate TAMP as a single bilevel optimization that couples symbolic search with dynamics-consistent motion and remains scalable by exploiting task structure. Second, we move this optimization offline and learn fast, robust motion policies by imitation, supported by cross-platform teleoperation and egocentric data infrastructure and by a discrete action representation that makes multi-task policies steerable at inference time. Finally, we propose an embodied agent system and an agent memory evaluation benchmark in which an LLM/VLM orchestrates VLA skills and a self-evolving multimodal memory accumulates verified experience across deployments, enabling the robot to continually self-improve from its own experience without retraining.
]]>
Committee:
Dr. Morris Cohen, ECE, Chair, Advisor
Dr. Gregory Durgin, ECE
Dr. Aaron Lanterman, ECE
Dr. Andrew Peterson, ECE`
Dr. Joel Johnson, OSU
]]>]]>
Doctor of Philosophy in Physics
School of Physics Thesis Dissertation Defense
Senlei Li
Dr. Chunhui Rita Du, School of Physics, Georgia Institute of Technology (Advisor)
Multimodal Scanning-Probe Quantum Sensing of Quantum Materials
Date: Friday, October 2, 2026
Time: 10:00 a.m.
Location: Howey N210
Thesis Committee:
Dr. Zhigang Jiang, School of Physics, Georgia Institute of Technology
Dr. Itamar Kimchi, School of Physics, Georgia Institute of Technology
Dr. Hailong Wang, School of Physics, Georgia Institute of Technology
Dr. Bolei Deng, Daniel Guggenheim School of Aerospace Engineering, Georgia Institute of Technology
Abstract:
This dissertation develops scanning nitrogen-vacancy (NV) quantum sensing as a local probe of magnetic and superconducting quantum materials. Scanning NV microscopy is used to image magnetic domains and switching in Mn₃Sn, resolve stacking-dependent magnetic phases in twisted CrI₃, and probe coherent microwave fields, magnetic noise, and vortices in a superconducting Nb resonator. The technique is further applied to twisted Bi₂Sr₂CaCu₂O₈+x, where local magnetic and noise measurements reveal boundary- and field-dependent responses associated with unconventional superconductivity. Together, these studies demonstrate scanning NV microscopy as a versatile multimodal platform for connecting nanoscale electromagnetic signals with microscopic phenomena in quantum materials.
]]>
Committee Members: Dr. Sudheer Chava (Chair), Dr. Manasa Gopal, Dr. Wendi Du (University of South Carolina)
Title: Essays on FinTech, AI and Innovation in Finance
Dissertation Overview:
Essay 1: How Do Banks Attract Deposits From Households?
Households allocate their savings between bank deposits and capital markets, yet deposit competition is typically studied only across banks. Using household portfolio data, we show that banks compete not only with other banks for deposits but also with households’ outside investment opportunities. The same bank pays higher deposit rates in markets where households hold more equity and experiences larger deposit outflows following monetary tightening. We instrument household equity holdings using local exposure to zero-commission trading. Our results show that greater household sensitivity to relative returns increases deposit rates, while a stronger response of outside returns to monetary policy generates larger deposit outflows after tightening. A 37-basis-point increase in the outside return reduces deposits by 1.75 percent and bank franchise value by 3.7 percent, weakening banks’ incentives to limit risk.
Essay 2: Paying for Prompting: How Generative AI Subscriptions Shape Household Finance and Labor Decisions
Using transaction-level consumer data, we study who subscribes to generative AI and how household financial and labor-market outcomes evolve around adoption. We identify paid subscriptions to major generative AI services directly from household transactions and document substantial heterogeneity in adoption across consumers. We find that AI subscription is followed by increases in salary, income, and spending. These gains are accompanied by greater employer mobility: subscribers become less likely to receive pay from their previous employers and more likely to receive pay from new employers, with the largest income gains concentrated among workers who switch employers. We further examine whether generative AI expands workers’ opportunity sets by facilitating transitions across more distant employers. Overall, our findings provide new evidence on how the diffusion of generative AI is associated with household financial outcomes and labor-market reallocation.
Essay 3: Fintech Securitization: Lending Incentives and Consumer Outcomes
We study how access to securitization funding affects fintech lending and consumer outcomes. We exploit the 2020 Term Asset-Backed Securities Loan Facility (TALF), which lowered securitization funding costs for eligible asset classes, and compare fintech lenders with differential exposure to the program. Following the shock, lending recovers more strongly among fintech lenders with prior access to eligible securitization markets. We then use consumer credit bureau data to examine the financial outcomes of borrowers receiving credit from exposed lenders. Among fintech lenders, borrowers subsequently experience greater delinquency and charge-offs, with similar patterns within credit cards. We further examine whether these outcomes reflect expanded credit access to predictably riskier borrowers or weaker screening incentives associated with transferring credit risk. Overall, our analysis sheds light on how securitization affects fintech credit supply, lending incentives, and borrower outcomes.
]]>Committee Members: Dr. Sudheer Chava (Chair), Dr. Suzanne Lee, Dr. Manasa Gopal, Dr. Nikhil Paradkar (University of Georgia)
Title: Essays on Financial Intermediaries and Household Finance: Auto Lending and Credit Unions
Dissertation Overview:
Essay 1: A Different Kind of Drive? Indirect Lending and the Transformation of Consumer Credit Markets
Auto repossessions in the United States have surged to levels last seen in the Great Recession. While standard explanations emphasize borrower credit risk, this paper highlights the role of the origination channel. We exploit the adoption of indirect (dealer-originated) lending programs across credit unions as an empirical setting to identify how distancing lenders from borrowers alters ex-post default resolution. Combining monthly credit bureau data with NCUA Call Reports from 2005 to 2024, we use the staggered adoption of indirect lending to trace how it changes lender behavior. At the lender level, indirect lending adoption significantly increases institution size and auto loan volume, while reducing liquidity and net worth ratios, reflecting a shift toward a more aggressive, growth-oriented financial model. From loan-level data, we show that this growth does not stem from an observably riskier borrower pool. Instead, the indirect channel fundamentally alters institutional forbearance. Conditional on 60 days past due, loans originated post-adoption are 2.6 percentage points (20%) more likely to result in repossession compared to pre-adoption cohorts within the same institution. This 'transactional penalty' is concentrated among lower-income and subprime borrowers yet weakens with the length and breadth of the borrower’s prior relationship with the credit union. The results suggest that rising repossessions reflect not only who borrows but how far lenders stand from the borrowers they serve.
Essay 2: Funding Constraints and the Deposits Channel of Monetary Policy: Evidence from Credit Unions
The standard deposits channel of monetary policy predicts that banks widen deposit spreads when policy rates rise and permit deposit outflows, as wholesale funding partly replaces them. While recent work attributes variation in this channel to clientele sensitivity and financial technology (money market funds, online banks, and digital platforms), we show that an intermediary's own funding constraints also shape it. Credit unions combine a branch-based deposit franchise with no access to external equity and limited access to wholesale funding, making them a natural laboratory for funding constraints that bind, to varying degrees, across all depository institutions. Despite their sleepy, branch-based retail clientele, credit unions pass through substantially more than banks on time deposits (share certificates). Members partially rotated into certificates rather than leaving, and total deposits continued to grow through the 2022–2023 tightening cycle while commercial banks lost $1 trillion. The higher marginal cost of deposits passes through to the asset side: credit union auto loan rates rise 17–23 basis points more than bank rates per 100 basis points of tightening. Within credit unions, deposit passthrough is stronger where alternative funding access is tightest, consistent with funding constraints as the mechanism.
Essay 3: Who Holds the Key? Lender Heterogeneity in Auto Loan Repossession
Auto repossession is the most common collateral seizure in U.S. household finance, exceeding 3 million a year, more than eviction filings, consumer bankruptcies, or mortgage foreclosures, yet the lender's decision to seize remains essentially undocumented. I study how that decision varies with who holds the loan, using a monthly, lender-identified credit bureau panel from 2010 to 2025 that tracks auto loan payment history, the decision to repossess, and how each repossession is resolved. First, in reduced form, I show that lenders differ sharply in their willingness to seize conditional on the same distress: credit unions repossess at a much lower hazard than other lenders, while nonbank lenders seize fastest. Using state-level repossession statutes and staggered state COVID moratoria, I test whether lender types respond differently to the same restriction, both when a ban binds and when it lifts. Second, to recover the primitives governing lender type heterogeneity, I estimate a dynamic model of the repossession decision, separately by lender type, decomposing these differences into lender objectives, securitization and origination channels.
]]>
Seonkyu Shin
(Advisor: Prof. Nazanin Bassiri-Gharb)
will propose a doctoral thesis entitled,
Antiferroelectricity of the PbHfO3–PbZrO3 Thin Films
On
Tuesday, October 13 at 2:00 p.m.
J. Erskine Love Room 311
or Virtually via MS TEAMS
Abstract
Antiferroelectrics are materials in which individual unit cells possess spontaneous polarization, but antiparallel alignment of dipoles in adjacent unit cells results in zero net polarization macroscopically Under a sufficiently high external electric field, the material undergoes a field-induced phase transition from the antipolar state to a polar state, where dipoles can be reoriented along crystallographically allowed direction(s) better aligned with the applied electric field. The antipolar state, however, can be restored upon electric field removal. The field-induced transitions (antipolar-to-polar at Ef and polar-to-antipolar at Ea) are accompanied by large nonlinear and hysteretic changes in dielectric, electro-optic, electrocaloric, and electromechanical responses. These characteristics make them attractive for various functional applications such as high-power (pulsed) capacitors, optical attenuators, phase shifters, solid-state refrigerants, and high force-high displacement actuators.
In 1951, PbZrO3 was the first material to be reported to exhibit antiferroelectricity, and has since been used extensively as the end member of Pb(Zr1-xTix)O3 (PZT) ceramics. PbZrO3 has thus been the archetypal antiferroelectric and a platform for evaluating antiferroelectric phenomena. However, the antiferroelectric nature of PbZrO3 has been the subject of recent literature debate. Anisotropic weak ferroelectricity and modulated (rather than simple antiparallel) Pb displacement observed in multiple experimental studies have raised the possibility that PbZrO3 may be ferrielectric rather than truly antiferroelectric.
PbHfO3 is another antiferroelectric perovskite oxide. Because PbHfO3 is isostructural with PbZrO3, the possibility that PbHfO3 may also be ferrielectric remains an open question. However, despite the first report of antiferroelectricity in PbHfO3 in 1953, systematic studies of polar response in this material remained relatively limited until the early 2000s. The relatively high Ef of bulk ceramic and single-crystal PbHfO3 compared to their PbZrO3 counterparts complicates experimental access to its field-induced phase transition, as dielectric breakdown could occur before complete switching is achieved. Moreover, Pb(HfxTi1-x)O3 did not exhibit the exceptional ferroelectric properties observed in the PZT system. However, since the mid-2000s, advances in processing techniques have enabled the fabrication of high-quality thin films, leading to renewed interest in PbHfO3 in thin film form. This study aims to characterize the antiferroelectricity of PbHfO3 thin films and investigate phase stability and field-induced phase transitions in PbHfO3-PbZrO3 thin films.
The first phase of this study will center on the chemical solution processing and functional characterization of PbHfO3 thin films, pursuing processing-structure-property correlations. PbHfO3 thin films with different preferred orientations are obtained; 84% 021O-oriented and 86% 001O-oriented PbHfO3 thin films are used to investigate the antiferroelectricity of PbHfO3 thin films. (The ”O” subscript underlines the use of the orthorhombic unit cell reference for the crystallographic directions) Functional properties of PbHfO3 are analyzed to compare anisotropic phase stability and field-induced phase transition behavior. To further understand the phase stability and field-induced transition behavior of PbHfO3 thin films, we examine the kinetic stability through frequency dependent measurements. The relative stability of the antipolar and polar phases and the stability of the reversible field-induced phase transition will be evaluated through temperature dependent measurements and fatigue testing, respectively.
Building upon the understanding of antiferroelectricity in PbHfO3 thin films, we will extend this study to PbHfO3/PbZrO3 multi-layered thin films to elucidate the influence of B-site cation distribution on antiferroelectric behavior. The phase stability of the two materials is expected to be modified starting from the interfaces. Additionally, electric field redistribution, lattice mismatch-induced strain, and defect densities may vary depending on the distribution of B-site cations, thereby influencing phase stability and field-induced phase transitions. This study will progress to PbHfO3–PbZrO3 solid-solution thin films to determine the relative stability of antipolar and polar phases and the nature of the field-induced phase transitions within a homogeneous system. Comparison with heterogeneous multi-layered thin films will clarify whether the trends in phase stability and field-induced phase transitions associated with B-site cation distribution persist in the homogeneous solid- solution limit. Ultimately, we aim to establish a composition- and electric field-dependent phase diagram of the PbHfO3-PbZrO3 solid-solution system in thin film form.
Committee
• Prof. Nazanin Bassiri-Gharb – George W. Woodruff School of Mechanical Engineering (Advisor)
• Prof. Juan-Pablo Correa-Baena – School of Materials Science and Engineering
• Prof. Rosario Gerhardt – School of Materials Science and Engineering
• Prof. Eric Vogel – School of Materials Science and Engineering
• Prof. Asif Khan – School of Electrical and Computer Engineering
]]>
Student Name: Anomitra De
Thesis Title: Processing–Structure–Function Relationships for High-Performance Porous Carbon Materials from Kraft Lignin
Thesis Advisor: Sankar Nair
Thesis Co-Advisor: Meisha Shofner
Committee Members: Nian Liu (CHBE), AJ Medford (CHBE), Matthew T McDowell (Department of Mechanical Engineering)
Date: 10/1/2026
Time: 10 AM
Location: Ford ES&T L1120
]]>
School of Psychological and Brain Sciences – Ph.D. Dissertation Defense Meeting
Date: Tue, Nov 10th, 2026
Time: 10:00 AM - 11:00 AM EST
Location: https://gatech.zoom.us/my/kaixue
Dissertation Committee Chair/Advisor:
Dobromir Rahnev, Ph.D. (Georgia Tech)
Dissertation Committee Members:
Tansu Celikel, Ph.D. (Georgia Tech)
Sashank Varma, Ph.D. (Georgia Tech)
Mengyao Li, Ph.D. (Georgia Tech)
Megan Peters, Ph.D. (University College London)
Title: The Computation of Confidence in Perceptual Decision-Making
Abstract: This dissertation investigates the computational mechanisms underlying confidence in perceptual decision-making through three related studies. Study 1 uses transcranial magnetic stimulation (TMS) to establish when confidence is computed, revealing that the dorsolateral prefrontal cortex (DLPFC) contributes to confidence during a broad temporal window overlapping with decision-making. Study 2 tests competing computational theories — Bayesian Confidence Hypothesis (BCH) versus Confidence in Raw Evidence Space (CRES) — in two-choice tasks, finding consistent support for evidence-based rather than probability-based confidence computation. Study 3 extends this investigation to multi-alternative decisions using a novel dot numerosity paradigm, demonstrating that confidence tracks the difference between the two highest evidence values rather than choice probability. Together, these studies advance a mechanistic framework in which confidence emerges in parallel with decisions and operates on sensory evidence rather than probability estimates.
]]>