150813 research outputs found
Sort by
Burning S(e)oul: A Body for Cremation
No full textEvery year, there are over 70,000 fatalities around Seoul, with only two operating crematoria in the city, that is over 100 bodies a day each institution needs to process efciently. By May 26, it would have been six years since my grandfather was gone in those fames. Threading the remnants of mourning, Burning S(e)oul, in forms of a short flm, is a dialogue between “absences” of bodies and architecture. It is presented as a triptych along three parallel timelines divided into fve tableaux. Narrating the aftermaths of death, it refects the bereaved, the deceased, and the workers’ perspective along three mandatory days of grieving. Absence in this paradigm is not solely physical or emotional but rather phenomenological— what appears a quotidian existence of oneself is stripped of its corpse, reafrming that the inherent genius loci of the crematorium instead refect a broader infuence that institutions have experienced since post-war Korea. It argues that the systematized practice of death processing is an apparatus used to sever the genealogy of individual bodies from their role in afrming personal and communal kinships. Embedded within its architectural design, this alienation dismantles time by shifting the condition of death processes as an engineered state, rather than historical or material one. This detachment is emblematic of the country’s postwar trajectory, where rapid modernization prioritized efciency over continuity, severing longstanding rituals that once bond personal grief to communal memory. The friction between an engineered present and an inherited past manifest as a form of cultural desynchronization— one where the ostensibly modern remains haunted by the traditional. This shift extends beyond mere technical or practical concerns; it represents a deliberate method of assimilating a nonlinear societal modernization—one that in its pursuit of progress, distances itself from historical trauma. Yet this tension does not merely mark a transition; it accumulates as a generational melancholy, where the urgency of progress leaves grief suspended in an unresolved state, neither fully severed nor meaningfully preserved.S.M.M.C.P
Report to the President for year ended June 30, 2025, Dean, School of Science
No full textThis report contains the following sections: Highlights, Initiatives and Programs, Education, Research, Fundraising and Philanthropy, Awards and Honors, and Personnel
Geometry and analysis of Ricci curvature and mean curvature flows
No full textIn this thesis, we study the geometry and analysis of spaces with Ricci curvature bounded below from the following three perspectives and the asymptotically conical singularities of mean curvature flows in the following two perspectives. For the spaces with Ricci curvature bounded below, firstly we study the unique continuation problem on RCD spaces, which is a long-standing open problem, with little known even in the setting of Alexandrov spaces. Together with Qin Deng, we proved that on RCD(K,2) spaces both harmonic functions and caloric functions satisfy weak unique continuation properties. Furthermore we constructed counter-examples showing that strong unique continuation in general fails for harmonic and caloric functions on RCD(K,N) spaces where N is greater or equal to 4. Secondly, we consider constructing a canonical diffeomorphism between the n-sphere and a n-dimensional space with Ricci curvature bounded from below by n-1 which is close to the n-sphere in the Gromov-Hausdorff sense. Together with Bing Wang we proved that the first (n+1)-eigenfunctions of Laplacian provides a bi-Holder diffeomorphism and we further give a counter-example showing that the bi-Holder estimate is sharp and cannot be improved to a bi-Lipschitz estimate. Thirdly, we study the Margulis Lemma on RCD spaces. Together with Qin Deng, Jaime Santos-Rodríguez and Sergio Zamora, we extend the Margulis Lemma for manifolds with lower Ricci curvature bounds to the RCD setting. As one of our main tools, we obtain improved regularity estimates for Regular Lagrangian flows on these spaces. For the asymptotically conical singularities of mean curvature flows, firstly together with Tang-Kai Lee, we proved asymptotically conical self-shrinkers as tangent flows of MCFs are unique, generalizing the result in the case of hypersurface proven by Chodosh-Schulze. Secondly, together with Tang-Kai Lee we prove that given any asymptotically conical shrinker, there exists an embedded closed hypersurface such that the mean curvature flow starting from it develops a type I singularity at time 1 at the origin modeled on the given shrinker.Ph.D
Report to the President for year ended June 30, 2025, Department of Earth, Atmospheric and Planetary Sciences
No full textThis report contains the following sections: Educational Activities, Faculty, Communications, Resource Development, and Faculty & Research Highlights
EEA Presidential Address: The Past, Present and Future of Health Care Reform
No full textThe starting point for my speech is the explosive growth in the field of health economics. In
1990, the American Economic Review published just two articles in health economics; now it publishes
about five per year. In the American Economic Journal: Economic Policy and American Economic Journal:
Applied Economics, major new general-interest journals in health economics, about one in eight articles
published in 2017 was in health economics. And what has made health economics so fascinating is that
its impact was felt not just in the scholarly world but also in the policy world as well, most notably
through the Affordable Care Act (ACA) in 2010.
One of the most frustrating aspects of being a health economist is that expectations for health
care suffer from extreme black and white thinking. Is the ACA a failure or a success? Are health care
costs under control or not under control? Is health care reform over or still going? The answer to all of
these is yes! When you have a sector that is 18% of the US economy, there are never simple yes and no
answers.
And in particular, one of the most frustrating aspects of working on health care reform is the
idea that we have ever “done” health care reform. Health care reform is not a single battle; it is an
ongoing war that will never be fully resolved. So when thinking about health care reform, it is important
to understand where we have been, where we are, and where we need to go next – and that’s what I’ll
try to cover in this lecture
Machine Learning through the Lens of Data
No full textMany critical challenges in machine learning—e.g., debugging model behavior or selecting good training data—require us to relate outputs of models back to the training data. The goal of predictive data attribution, the focus of this thesis, is to precisely characterize the resulting model behavior as a function of the training data in order to tackle these challenges. In the first part of this thesis, we introduce a framework, datamodeling, for formalizing and constructing effective methods for predictive data attribution. Despite the complexity of modern machine learning systems (e.g., end-to-end training of deep neural networks using stochastic gradient algorithms), we show that we can accurately predict model outputs from simple linear functions of the training data. We then demonstrate that these predictors—which we call datamodels—provide a versatile primitive for various tasks, ranging from predicting the effect of dataset counterfactuals to identifying brittle predictions. Next, to further improve the scalability of data attribution in this framework, we design a new method trak (Tracing with the Randomly-projected After Kernel) that is both effective and computationally tractable for large-scale, differentiable models. By leveraging a kernel approximation and other classic ideas from statistics and algorithm design, we are able to reduce the challenging problem of attributing the original DNN to that of attributing a simpler surrogate. We demonstrate the effectiveness of trak across various modalities and scales: image classifiers trained on ImageNet, vision-language models (CLIP), language models (BERT and mT5), and diffusion models. In the second part of this thesis, we explore applications of this framework developed in the first part: First, we leverage datamodels for the problem of learning algorithm comparison, where the goal is to detect differences between models trained with two different learning algorithms. Our algorithm, ModelDiff, enables us to automatically surface biases that distinguish different learning algorithms by differentiating how they use the same training data. Lastly, we tackle the challenging problem of machine unlearning, wherein the goal is to “unlearn” a small fraction of training data from a trained model. By leveraging the fact that datamodels can accurately approximate the “oracle” predictions, we design a simple finetuning algorithm that allows us to unlearn at a significantly smaller cost than prior methods.Ph.D
Catalytic implications of confined solvent ensembles within Lewis acid zeolites
No full textLewis acidic zeolites are microporous crystalline materials that offer promise as catalysts for the activation and conversion of biomass-derived precursors in the liquid-phase due to their unique water-tolerance and synthetic versatility. The active site environment in zeolite catalysts is multifaceted in nature and is composed of a primary catalytic binding site, the secondary pore structure that confines such binding sites, and occluded solvent and reactant molecules that interact with adsorbed species. Moreover, Lewis acidic heteroatoms can adopt structurally diverse coordination that selectively catalyze different classes of chemical transformations and can be difficult to control synthetically or characterize spectroscopically. In this thesis, precise mechanistic interpretation of liquid-phase zeolite catalysis was realized through the development of synthetic, spectroscopic, and kinetic methods that decouple complex active site structures and probe the interactions that occur between confined active sites, solvent and reactant molecules, and adsorbed intermediates and transition states.
First, we show how hydrophobic Beta zeolites containing framework Sn atoms catalyze transfer hydrogenation reactions of cyclohexanone in a 2-butanol solvent 10x faster than their hydrophilic analogues. This rate enhancement stems from the ability of hydrophobic Sn-Beta to inhibit the formation of extended liquid-like 2-butanol oligomers and promote dimeric H-bonded 2-butanol networks. The ordered H-bonding solvent network present in hydrophobic Sn-Beta stabilizes the transfer hydrogenation transition state to a greater extent than the liquid-like 2-butanol solvent present in hydrophilic Sn-Beta, giving rise to higher turnover rates on hydrophobic Sn-Beta. Additionally, reactant adsorption within hydrophobic Sn-Beta is entropically-driven by the breakup of intraporous solvent-solvent interactions, resulting in positive enthalpies of adsorption that are partially compensated by an increase in the solvent reorganization entropy. These results emphasize the ability of the zeolite pore to regulate the structure of confined non-aqueous H-bonding solvent networks, which offers an additional dimension to modulate adsorption and reactivity.
Next, we extend our studies to understand how different intraporous alcohol networks reorganize in response to adsorbate sterics and the presence of non-H-bonding co-solvents. Here, we find that first-order rates for methyl-cyclohexanone transfer hydrogenation are ~2-5x higher than for tert-butyl-cyclohexanone, but converge in the zero-order regime across all temperatures (333-393 K) in a bulk 2-butanol solvent. These results show that, while intrinsic bond-activation steps at the active site are largely independent of molecular functionalization of the ketone reactant, adsorption within hydrophobic Sn-Beta is still driven by the breakup of intraporous solvent-solvent interactions. Furthermore, comparisons between bulk toluene or acetonitrile solvents, with 1 M 2-butanol as a reactant, show the significance of intraporous solvent for stabilizing kinetically-relevant species and the complex interdependencies between solvent and catalyst hydrophilicity. Apparent zero-order activation enthalpies and entropies increase with decreasing solvent polarity over hydrophobic zeolites indicating that the transition state is more tightly bound to the open Sn site when first-shell solvent molecules become more polarizing. Conversely, adsorption and activation entropies and enthalpies measured on hydrophilic zeolite in toluene and acetonitrile solvents are nearly identical to those measured in a bulk 2-butanol solvent, suggesting that the intraporous solvating environment in bulk, non-H-bonding co-solvents is similar to that observed when bulk 2-butanol is the solvent.
Finally, we exploit the ability of carbonyl groups to measure electric field differences arising from the different intraporous solvent structures through the vibrational Stark effect. By measuring infrared absorption spectra of Ti-bound acetone in Beta zeolites of varying framework hydrophobicity across a wide range of non-coordinating solvents, we find unique electric field differences arising from distinct solvation under nanoconfinement. Moreover, in the absence of intraporous solvent, we observe a ~7 cm-1 shift in the Ti-bound carbonyl stretching frequency. These results suggest that local differences in the Lewis acid site environment, which influence observed kinetics across reaction classes, arise from the synthetic protocol used to produce each material.
Taken together, the results of this thesis reveal how different solvent-mediated, non-covalent interactions control liquid-phase reactivity within porous, Lewis acid zeolite catalysts. It is our hope that the kinetic and spectroscopic approaches advanced here will provide a useful roadmap for further experimental investigations into the catalytic implications of confined solvent.Ph.D
Report to the President for year ended June 30, 2025, Department of Civil and Environmental Engineering
No full textThis report contains the following sections: Goals, Objectives, Priorities; Development and Fundraising, Research Activities and Accomplishments
Essays on Information Economics
No full textThis thesis contains 5 chapters. Every chapter deals with the question of how information affects equilibrium behavior in strategic problems.
Chapter 1 is my job market paper "Limits of Global Games.'' It considers the impact of information on equilibrium multiplicity in two-player games of strategic complementarities. Games with strategic complementarities often exhibit multiple equilibria. In a global game, players privately observe a noisy signal of the underlying payoff matrix. As the noise diminishes, a unique equilibrium is selected in almost all binary-action games with strategic complementarities - a property known as "limit uniqueness.'' This chapter describes the limits of that approach in two-player games, as we move beyond two actions. Unlike binary-action games, limit uniqueness is not an intrinsic feature of all games with strategic complementarities. When the noise is symmetric, we demonstrate that limit uniqueness holds if and only if the payoffs exhibit a generalized ordinal potential property. Moreover, we provide an example illustrating how this condition can be easily violated.
Chapter 2 is co-authored with Olivier Gossner and is titled "Strategic Type Spaces.'' We provide a strategic foundation for information: in any given game with incomplete information we define strategic quotients as information representations that are sufficient for players to compute best-responses to other players. We prove 1/ existence and essential uniqueness of a minimal strategic quotient called the Strategic Type Space (STS) in which a type is given by an interim correlated rationalizability hierarchy together with the set of beliefs over other players' types and nature that rationalize this hierarchy 2/ that this minimal STS is a quotient of the universal type space and 3/ that the minimal STS has a recursive structure that is captured by a finite automaton.
Chapter 3 is also co-authored with Olivier Gossner and is titled "Information Design for Rationalizability.'' We study (interim correlated) rationalizability in games with incomplete information. For each given game, we show that a simple and finitely parameterized class of information structures is sufficient to generate every outcome distribution induced by general common prior information structures. In this parameterized family, players observe signals of two kinds: A finite signal and a common state with additive, idiosyncratic noise. We characterize the set of rationalizable outcomes of a given game as a convex polyhedron.
Chapter 4 is co-authored with Stephen Morris and Dirk Bergemann and is titled "A Strategic Topology on Information Structures.'' Two information structures are said to be close if, with high probability, there is approximate common knowledge that interim beliefs are close under the two information structures. We define an "almost common knowledge topology'' reflecting this notion of closeness. We show that it is the coarsest topology generating continuity of equilibrium outcomes. We show that finite information structures are dense in the almost common knowledge topology and thus it is without loss to restrict attention to finite information structures in information
design problems.
Finally, chapter 5 is a short note describing an information aggregation mechanism that can be used by players before playing a game of strategic complementarities under incomplete information. In such a game, players may have an incentive to share overly optimistic information with other players, thus inducing them to play higher actions. In this mechanism, players trade a token before playing the game. Players who want to communicate good news must purchase this worthless token and burn resources. The note shows that players only need to observe the market clearing price that arises from the token trades to aggregate their private information. Each element in a player's private information set is encoded as a prime in the prime factorization of the market clearing price. The element that is contained in every player's information set is identified as the prime with the highest multiplicity.
JEL Classification Codes: C72, D82Ph.D
Piloting batch reverse osmosis with a flexible bladder for water recovery from scaling-prone brine
No full textA pilot-scale batch reverse osmosis (RO) system with a flexible bladder was designed to recover additional water from RO concentrate. The sulfate-rich, ~6400-ppm concentrate was sourced from the Yuma Desalting Plant (Arizona, USA), which desalinates agricultural drainage water. The pilot produced 4.4 m3/day of permeate with 150 ppm total dissolved solids from the facility’s concentrate stream with a recovery ratio of 82.6%. Despite producing supersaturated brine, there was no performance deterioration due to scaling. Using a bladder for retentate pressurization limited average power to 633 W and the specific energy consumption to 3.3 kWh/m3. The pilot’s energy data informed a model of large-scale batch RO, which has the potential to desalinate the same water for less than 1 kWh/m3. Additionally, a model was developed to predict scaling likelihood in batch RO. This investigation demonstrates that batch RO is a viable technology for low-energy brine concentration beyond saturation limits.US Bureau of Reclamatio