79297 research outputs found
Sort by
Learning Neural Point Processes for Long Event Sequences
Get PDFThis research presents a comprehensive series of studies aimed at advancing learning neural point processes for long event sequences, with applications spanning disaster resilience, crime forecasting, and healthcare. Structured around three interconnected studies, this work addresses core challenges in temporal point process (TPP) modeling, efficient handling of long event sequences, and improving accuracy over extended forecasting horizons by reinforcement learning.
The first study proposes the Sparse Transformer Hawkes Process (STHP) to model long asynchronous event sequences. Traditional neural network-based TPPs struggle with long event sequences due to computational inefficiencies. To address this, the STHP model combines two components: a temporal sparse self-attention mechanism that focuses on short-term dependencies and a second transformer applied to aggregated event counts for long-term dependency extraction. By integrating these components, STHP models the conditional intensity of point processes efficiently, improving prediction performance for long sequences without high computational costs.
The second study proposes a debiased imitation learning (DIL) framework for Modulated Temporal Point Processes to handle biased event sequences encountered in applications like disaster resilience, criminology, and healthcare. In real-world settings, temporal events often suffer from unknown biases due to external factors, which lead to misspecifications in TPPs when learned using conventional maximum likelihood estimation (MLE). To address these biases, the DIL framework explicitly models biased sequences through additional unknown thinning processes, mitigating the impact of biased data. By leveraging a sequence-level reward function derived from historical embeddings, the DIL framework enhances prediction robustness and accuracy.
The third study introduces the Inflence-Guided Reinforcement Learning Spatio-Temporal Point Process (IGPO) framework, a novel framework for modeling spatio-temporal event sequences. Spatio-temporal data presents unique challenges due to dependencies across both spatial and temporal dimensions, and neural network-based spatio-temporal point processes provide a sophisticated modeling framework for these data. Conventional MLE may lead to inaccurate predictions because of model misspecification and compounding errors. Alternatively, reinforcement learning, which treats event generation as actions to mimic observed event patterns, can address the training and testing discrepancy but often suffers from poor exploration efficiency.
Together, these studies contribute a cohesive set of advancements in long event sequence prediction, addressing key challenges in efficiency, biased data, model misspecification, and compounding errors
SPECTRAL ANALYSIS OF THE NEUMANN–POINCARÉ OPERATOR FOR THIN DOUBLY CONNECTED DOMAINS
No full textWe analyze the spectrum of the Neumann–Poincaré (NP) operator for a doubly connected domain lying between two level curves defined by a conformal mapping, where the inner boundary of the domain is of general shape. The analysis relies on an infinite-matrix representation of the NP operator involving the Grunsky coefficients of the conformal mapping and an application of the Gershgorin circle theorem. As the thickness of the domain shrinks to zero, the spectrum of the doubly connected domain approaches the interval [-1/2,1/2] in the Hausdorff distance and the density of eigenvalues approaches that of a thin circular annulus
On the oriented diameter of near planar triangulations
No full textIn this paper, we show that the oriented diameter of any n-vertex 2-connected near triangulation is at most [Formula presented] (except for seven small exceptions), and the upper bound is tight. This extends a result of Wang et al. (2021) [29] on the oriented diameter of maximal outerplanar graphs, and improves an upper bound of n/2+O(n) on the oriented diameter of planar triangulations by Mondal et al. (2024) [24]
A UNIFIED PROXIMAL GRADIENT METHOD FOR NONCONVEX COMPOSITE OPTIMIZATION WITH EXTRAPOLATION
No full textIn this paper, we propose a unified proximal gradient method with extrapolation (UPG-E) to solve a class of nonconvex and nonsmooth composite optimization. UPG-E provides a unified treatment to both convex and nonconvex problems, and adaptively estimates the nonconvexity modulus of the possibly nonconvex component function in the objective function. It is shown that without restarting the extrapolation, UPG-E achieves the optimal convergence rate of the first-order methods for solving convex composite optimization. In the case that the problem is nonconvex, the method performs as a proximal gradient method with extrapolation and guaranteed global convergence. Moreover, a linear convergence rate can be achieved by UPG-E under proper additional regularity assumptions. Our numerical experiments show the performance of UPG-E is very promising compared with other well-established proximal gradient methods in the literature
Technology of the Oppressed: Inequity and the Digital Mundane in Favelas of Brazil by David Nemer (review)
No full textNew layered quaternary Zintl pnictide oxides Ba2Zn2Pn2O (Pn = Sb, Bi): Discovery, crystal structures, band engineering, and transport properties
No full textThree new heteroanionic oxypnictides, Ba2Zn2Sb2O, Ba2Zn2Bi2O, and the solid solution Ba2Zn2Sb2−xBixO (x ≈ 1.1–1.6), have been synthesized and structurally characterized. They are isostructural with their Mn-bearing analog, adopting the Ba2Mn2Sb2O-type structure (space group P63/mmc, No. 194), and feature a double-layered 2D 2∞ [Zn2Pn2O]2- substructure (Pn = Sb, Bi, Sb/Bi) composed of corner-sharing, distorted tetrahedral ZnPn3O units. Electronic structure calculations reveal a systematic progression from semiconducting Ba2Zn2Sb2O to metallic Ba2Zn2Bi2O as Bi content increases. These trends are corroborated by transport property measurements, with Ba2Zn2Sb0.9(1)Bi1.1O exhibiting relatively low electrical resistivity, high Hall mobilities of ∼160 cm2/V·s, and large Seebeck coefficients from 69 to 132 μV K−1 over the 300–600 K temperature range. Comparison with structurally related Zintl pnictides, such as SrIn2As2 and PrZn3As3 phases, situates Ba2Zn2Pn2O (Pn = Sb, Bi) within a broader family of heteroanionic oxypnictide Zintl compounds, highlighting their structural flexibility and amenability to band engineering. Electronic structure and bonding considerations point to tunable semiconducting behavior and underscore the relevance of these materials for thermoelectric and topological applications
Machine learning enabled measurements of astrophysical (p,n) reactions with the SECAR recoil separator
Get PDFThe synthesis of heavy elements in supernovae is affected by low-energy (n,p) and (p,n) reactions on unstable nuclei, yet experimental data on such reaction rates are scarce. The SECAR (SEparator for CApture Reactions) recoil separator at FRIB (Facility for Rare Isotope Beams) was originally designed to measure astrophysical reactions that change the mass of a nucleus significantly. We used a novel approach that integrates machine learning with ion-optical simulations to find an ion-optical solution for the separator that enables the measurement of (p,n) reactions, despite the reaction leaving the mass of the nucleus nearly unchanged. A new measurement of the Fe58(p,n)Co58 reaction in inverse kinematics with a 3.66±0.12 MeV/nucleon Fe58 beam (corresponding to 3.69±0.12 MeV proton energy in normal kinematics) yielded a cross-section of 20.3±6.3 mb and served as a proof of principle experiment for the new technique demonstrating its effectiveness in achieving the required performance criteria. This novel approach paves the way for studying astrophysically important (p,n) reactions on unstable nuclei produced at FRIB
The Kangaroo’s First Hop: The Early Fast Cooling Phase of EP250108a/SN 2025kg
No full textFast X-ray transients are a rare and poorly understood population of events. Previously difficult to detect in real time, the launch of the Einstein Probe with its Wide-field X-ray Telescope has led to a rapid expansionof the sample and allowed the exploration of these transients across the electromagnetic spectrum. EP250108a is a recently detected example linked to an optical counterpart, SN 2025kg, or “the kangaroo.” Together with a companion Letter we present our observing campaign and analysis of this event. In this letter, we focus on the early evolution of the optical counterpart over the first 6 days, including our measurement of the redshift of z = 0.17641. We compare to other supernovae and fast transients showing similar features, finding significant similarities with SN 2006aj and SN 2020bvc, and show that the source is well modelled by a rapidly expanding cooling blackbody. We show the observed X-ray and radio properties are consistent with a collapsar-powered jet that is low energy (≲1051 erg) and/or fails to break out of the dense material surrounding it. While we examine the possibility that the optical emission emerges from the shock produced as the supernova ejecta expand into a dense shell of circumstellar material, due to our X-ray and radio inferences, we favour a model where it arises from a shocked cocoon resulting from a trapped jet. This makes SN 2025 one of the few examples of this currently observationally rare event
Prompt Gamma-Ray Burst Recognition through Waterfalls and Deep Learning
No full textGamma-ray bursts (GRBs) are one of the most energetic phenomena in the cosmos, whose study can probe physics extremes beyond the reach of laboratories on Earth. Our quest to unravel the origin of these events and understand their underlying physics is far from complete. Central to this pursuit is the rapid classification of GRBs to guide follow-up observations and analysis across the electromagnetic spectrum and beyond. Here, we introduce a compelling approach that can set a milestone toward a new and robust GRB prompt classification method. Leveraging self-supervised deep learning, we pioneer a previously unexplored data product to approach this task: GRB waterfalls
Assessing Wintertime Export Fluxes in the Labrador Sea using ²³⁴Th-²³⁸U Disequilibria and a Mechanistic Particle Sinking Model.
No full textThe export efficiency of biogenic particles from the upper ocean is a crucial determinant of ocean carbon sequestration. High-latitude regions, such as the Labrador Sea, have been reported to be quite efficient at transporting carbon to the deep ocean, but the magnitude and efficiency of this carbon export vary widely between seasons. In this study, we combined observational measurements of export flux with a particle sinking model to constrain a seasonal carbon flux during winter, a time of year when the Labrador Sea is characterized by deepening convection and persistent storms. Samples were collected from 8 stations between December 2 and 23, following a Lagrangian sampling scheme that tracked an SF6-tagged water mass. POC export fluxes estimated using a steady-state 234Th-238U disequilibria approach and ranged from 1.3 mmol C m−2 d−1 to 8.6 mmol C m−2 d−1 at 100 m. However, this estimate is likely a lower limit due to the influences of physical and non-steady-state processes. The export efficiency, ratio of POC export to net primary production (NPP), varied from 17 % to 99 %, which revealed that the Labrador Sea can be quite efficient at exporting carbon in winter. This apparently higher export efficiency could be due to a time lag and seasonal decoupling of export flux and NPP. The mechanistic particle sinking model further supported this notion by revealing that smaller, slowly sinking particles play a significant role in efficiently transporting carbon in winter, which can result in such decoupling. This is the first study to show the importance of the Labrador Sea in exporting significant amounts of carbon during winter, and further research needs to be done to better constrain the overall biological carbon pump (BCP) in higher latitude seas. Significant seasonal gaps reveal the need to further constrain how the BCP functions on an annual basis. This will improve the predictive capabilities of how the BCP will respond to future climate change