Publications

An information maximization model of eye movements.

We propose a sequential information maximization model as a general strategy for programming eye movements. The model reconstructs high-resolution visual information from a sequence of fixations, taking into account the fall-off in resolution from the fovea to the periphery. From this framework we get a simple rule for predicting fixation sequences: after each fixation, fixate next at the location that minimizes uncertainty (maximizes information) about the stimulus. By comparing our model performance to human eye movement data and to predictions from a saliency and random model, we demonstrate that our model is best at predicting fixation locations. Modeling additional biological constraints will improve the prediction of fixation sequences. Our results suggest that information maximization is a useful principle for programming eye movements.

The effect of surface oxygen on hydrocarbon reactions catalyzed by platinum crystal surfaces with variable kink concentrations

Understanding Brønsted‐Acid Catalyzed Monomolecular Reactions of Alkanes in Zeolite Pores by Combining Insights from Experiment and Theory

Abstract Acidic zeolites are effective catalysts for the cracking of large hydrocarbon molecules into lower molecular weight products required for transportation fuels. However, the ways in which the zeolite structure affects the catalytic activity at Brønsted protons are not fully understood. One way to characterize the influence of the zeolite structure on the catalysis is to study alkane cracking and dehydrogenation at very low conversion, conditions for which the kinetics are well defined. To understand the effects of zeolite structure on the measured rate coefficient (k app ), it is necessary to identify the equilibrium constant for adsorption into the reactant state (K ads‐H+ ) and the intrinsic rate coefficient of the reaction (k int ) at reaction temperatures, since k app is proportional to the product of K ads‐H+ and k int . We show that K ads‐H+ cannot be calculated from experimental adsorption data collected near ambient temperature, but can, however, be estimated accurately from configurational‐bias Monte Carlo (CBMC) simulations. Using monomolecular cracking and dehydrogenation of C 3 –C 6 alkanes as an example, we review recent efforts aimed at elucidating the influence of the acid site location and the zeolite framework structure on the observed values of k app and its components, K ads‐H+ and k int .

Supernovae 2001iq and 2001ir

RLDRM: Closed Loop Dynamic Cache Allocation with Deep Reinforcement Learning for Network Function Virtualization

Network function virtualization (NFV) technology attracts tremendous interests from telecommunication industry and data center operators, as it allows service providers to assign resource for Virtual Network Functions (VNFs) on demand, achieving better flexibility, programmability, and scalability. To improve server utilization, one popular practice is to deploy best effort (BE) workloads along with high priority (HP) VNFs when high priority VNF's resource usage is detected to be low. The key challenge of this deployment scheme is to dynamically balance the Service level objective (SLO) and the total cost of ownership (TCO) to optimize the data center efficiency under inherently fluctuating workloads. With the recent advancement in deep reinforcement learning, we conjecture that it has the potential to solve this challenge by adaptively adjusting resource allocation to reach the improved performance and higher server utilization. In this paper, we present a closed-loop automation system RLDRM <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sup> RLDRM: Reinforcement Learning Dynamic Resource Management to dynamically adjust Last Level Cache allocation between HP VNFs and BE workloads using deep reinforcement learning. The results demonstrate improved server utilization while maintaining required SLO for the HP VNFs.

An Intervention to Enhance Social, Emotional, and Identity Learning for Very Young Adolescents and Support Gender Equity: Protocol for a Pragmatic Randomized Controlled Trial (Preprint)

<sec> <title>BACKGROUND</title> The onset of puberty is a pivotal period of human development that is associated with significant changes in cognitive, social, emotional, psychological, and behavioral processes that shape identity formation. Very early adolescence provides a critical opportunity to shape identity formation around gender norms, attitudes, and beliefs before inequitable gender norms are amplified during and after puberty. </sec> <sec> <title>OBJECTIVE</title> The aim of the Discover Learning Project is to integrate strategic insights from developmental science to promote positive transformation in social, emotional, and gender identity learning among 10- to 11-year-olds in Tanzania. Through a pragmatic randomized controlled trial, the intervention scaffolds the development of critical social and emotional mindsets and skills (curiosity, generosity, persistence, purpose, growth mindset, and teamwork) delivered by conducting 18 after-school, technology-driven, experiential learning sessions in small, mixed-gender groups. </sec> <sec> <title>METHODS</title> The Discover Learning Intervention is a 3-arm randomized controlled trial that will be delivered to 579 participants selected from four public primary schools in Temeke District, Dar es Salaam, Tanzania. Randomization will be done at the individual level into 3 treatment groups receiving incremental intervention components. The treatment components include Discover Learning content curated into child-friendly videos, facilitated discussions, and a parent-child workbook, to be implemented over two phases, each 6 weeks long. A baseline survey will be administered to participants and their parents prior to the intervention. The process will be observed systematically, and data will be collected using surveys, in-depth interviews, observations, and focus group discussions with adolescents, parents, teachers, and facilitators conducted prior, during, and after each implementation phase. </sec> <sec> <title>RESULTS</title> This study builds on formative and pilot studies conducted with the target population to inform the design of the intervention. The results will generate new evidence that will inform strategies for achieving scale in Tanzania and provide insights for replication of similar programs that are invested in gender-transformative interventions in peri-urban, low-resource settings. </sec> <sec> <title>CONCLUSIONS</title> The Discover Learning Intervention makes an important contribution to the field of adolescent developmental science as an intervention designed for very young adolescents in a low-resource setting. </sec> <sec> <title>CLINICALTRIAL</title> ClinicalTrials.gov NCT04458077; https://clinicaltrials.gov/ct2/show/NCT04458077 </sec> <sec> <title>INTERNATIONAL REGISTERED REPORT</title> DERR1-10.2196/23071 </sec>

Scaling Water Vapor and Carbon Dioxide Exchange from Leaves to a Canopy: Rules and Tools

A differential method for computing local shape-from-texture for planar and curved surfaces

We model the texture distortion at a point in any particular direction on the image plane as an affine transformation and derive the relationship between the parameters of the affine transformation and the surface shape and orientation. We use a technique for estimating affine transforms between nearby image patches which is based on solving a system of linear constraints derived from a differential analysis. It is not necessary to explicitly identify texels or make restrictive assumptions about the nature of the image texture like isotropy. We have developed two different algorithms for recovering surface orientation and shape based on the estimated affine transforms in a number of different directions. The first is a sample linear algorithm based on singular value decomposition. The second is based on nonlinear minimization of a least squares error criterion. Experimental results are presented on images of planar and curved surfaces under perspective projection.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>

LIGO/Virgo S190706ai: No counterpart candidates in KAIT observations

Copper Tantalate by a Sodium‐Driven Flux‐Mediated Synthesis for Photoelectrochemical CO ₂ Reduction (Small Methods 8/2025)

Transitions Into Underage and Problem Drinking: Developmental Processes and Mechanisms Between 10 and 15 Years of Age

Adolescents ages 10–15 experience dramatic changes in their biological, cognitive, emotional, and social development as well as in their physical and social environments. These include the physiological and psychological changes associated with puberty; further development of the brain; changes in family, peer, and romantic relationships; and exposure to new societal and cultural influences. During this period, many adolescents also begin to use alcohol. Alcohol use during adolescence has adverse effects on the body and increases the risk of alcohol dependence later in life. To better understand why some children drink whereas others do not, researchers are examining nonspecific and alcohol-specific factors that put adolescents at risk for, or which protect them from, early alcohol use and its associated problems. Nonspecific risk factors include certain temperamental and personality traits, family factors, and nonnormative development. Examples of nonspecific protective factors include certain temperamental characteristics, religiosity, and parenting factors (e.g., parental nurturance and monitoring). Among the most influential alcohol-specific risk and protective factors are a family history of alcoholism and the influences of siblings and peers, all of which shape an adolescent’s expectancies about the effects of alcohol, which in turn help determine alcohol use behaviors.

JAMA

High‐Porosity Metal‐Organic Framework Glasses

Abstract We report a synthetic strategy to link titanium‐oxo (Ti‐oxo) clusters into metal‐organic framework (MOF) glasses with high porosity though the carboxylate linkage. A new series of MOF glasses was synthesized by evaporation of solution containing Ti‐oxo clusters Ti 16 O 16 (OEt) 32 , linkers, and m ‐cresol. The formation of carboxylate linkages between the Ti‐oxo clusters and the carboxylate linkers was confirmed by Fourier‐transform infrared (FT‐IR) spectroscopy. The structural integrity of the Ti‐oxo clusters within the glasses was evidenced by both X‐ray absorption near edge structure (XANES) and 17 O magic‐angle spinning (MAS) NMR. After ligand exchange and activation, the fumarate‐linked MOF glass, termed Ti‐Fum, showed a N 2 Brunauer–Emmett–Teller (BET) surface areas of 923 m 2 g −1 , nearly three times as high as the phenolate‐linked MOF glass with the highest BET surface area prior to this report.

CCDC 2034516: Experimental Crystal Structure Determination

Development of feedstocks for cellulosic biofuels

The inclusion of cellulosic ethanol in the Energy Independence and Security Act (EISA) of 2007 and the revised Renewable Fuel Standard (RFS2) has spurred development of the first commercial scale cellulosic ethanol biorefineries. These efforts have also revived interest in the development of dedicated energy crops selected for biomass productivity and for properties that facilitate conversion of biomass to liquid fuels. While many aspects of developing these feedstocks are compatible with current agricultural activities, improving biomass productivity may provide opportunities to expand the potential for biofuel production beyond the classical research objectives associated with improving traditional food and feed crops.

Simplified Earthquake Analysis of Structures with Foundation Uplift

Simplified analysis procedures are developed to consider the beneficial effects of foundation‐mat uplift in computing the earthquake response of structures, which respond essentially as single‐degree‐of‐freedom systems in their fixed‐base condition. These analysis procedures are presented for structures attached to a rigid foundation mat, which is supported on rigid foundation soil or flexible foundation soil modeled as two spring‐damper elements. Winkler foundation with distributed spring‐damper elements, or a viscoelastic half‐space. In these analysis procedures, the maximum earthquake‐induced base shear and deformation for an uplifting structure are computed directly from the earthquake response spectrum. It is demonstrated that the simplified analysis procedures provide results for the maximum base shear and deformation to a useful degree of accuracy for practical structural design.

Water harvesting from air with metal-organic frameworks powered by natural sunlight

Solar heat helps harvest humidity Atmospheric humidity and droplets constitute a huge freshwater resource, especially at the low relative humidity (RH) levels typical of arid environments. Water can be adsorbed by microporous materials such as zeolites, but often, making these materials release the water requires too much energy to be practical. Kim et al. used a metal-organic framework (MOF) material that has a steep increase in water uptake over a narrow RH range to harvest water, using only ambient sunlight to heat the material. They obtained 2.8 liters of water per kilogram of MOF daily at 20% RH. Science , this issue p. 430

Long-fiber reinforced thermoplastic composite lattice structures: Fabrication and compressive properties

Recyclable lightweight materials with advanced processing techniques are essential for the sustainable development of future transportation. Thermoplastic composites lattice structures were developed to meet this demand. An additive manufacturing method is presented here to fabricate such lattice structures by reversible assembly of several long-fiber reinforced thermoplastic composite parts (LFTCPs) which were economically processed by injection molding. The resulting thermoplastic lattice structures (density of 30kg·m−3) assembled with different sequences and connections are structurally evaluated and compared. Out-of-plane compression tests revealed that their mechanical properties were more sensitive to the presence of the connections rather than their assembly sequence, although their structural failure mode was always brake of inclined struts followed by fracture of the horizontal struts. Potential solutions to the problem of internal stresses, induced during assembly, are also explored by designing novel LFTCPs. The novel fabrication route for thermoplastic lattice structures will improve the prospects for their industrial application.