Search for: All records

Abstract The tectonic stress field induces surface deformation. At long wavelengths, both lithospheric heterogeneity (changes in the thickness and density of crust and lithospheric mantle) and basal tractions from mantle convection contribute to the stress field. Here, we analyze the global alignment of principal horizontal tectonic stresses, fault traces, and river flow directions to infer whether and how deep subsurface stresses control geomorphic features. We find that fault trace orientations are consistent with predictions from Anderson's fault theory. River directions largely align with fault traces and partly with stresses. The degree of alignment depends on fault regime, the source of stress, and river order. Extensional faulting is best predicted by stresses from lithospheric structure variations, while compressive faulting is best predicted by stresses from mantle flow. We propose a metric to quantify the relative influence of mantle flow or lithospheric heterogeneity on surface features, which provides a proxy for lithospheric strength. 

Vision-language models are integral to computer vision research, yet many high-performing models remain closed-source, obscuring their data, design and training recipe. The research community has responded by using distillation from black-box models to label training data, achieving strong benchmark results, at the cost of measurable scientific progress. However, without knowing the details of the teacher model and its data sources, scientific progress remains difficult to measure. In this paper, we study building a Perception Language Model (PLM) in a fully open and reproducible framework for transparent research in image and video understanding. We analyze standard training pipelines without distillation from proprietary models and explore large-scale synthetic data to identify critical data gaps, particularly in detailed video understanding. To bridge these gaps, we release 2.8M human-labeled instances of fine-grained video question-answer pairs and spatio-temporally grounded video captions. Additionally, we introduce PLM-VideoBench, a suite for evaluating challenging video understanding tasks focusing on the ability to reason about "what", "where", "when", and "how" of a video. We make our work fully reproducible by providing data, training recipes, code & models. 

Abstract Landslides are notoriously difficult to predict because numerous spatially and temporally varying factors contribute to slope stability. Artificial neural networks (ANN) have been shown to improve prediction accuracy but are largely uninterpretable. Here we introduce an additive ANN optimization framework to assess landslide susceptibility, as well as dataset division and outcome interpretation techniques. We refer to our approach, which features full interpretability, high accuracy, high generalizability and low model complexity, as superposable neural network (SNN) optimization. We validate our approach by training models on landslide inventories from three different easternmost Himalaya regions. Our SNN outperformed physically-based and statistical models and achieved similar performance to state-of-the-art deep neural networks. The SNN models found the product of slope and precipitation and hillslope aspect to be important primary contributors to high landslide susceptibility, which highlights the importance of strong slope-climate couplings, along with microclimates, on landslide occurrences. 

Abstract We investigated the doping and temperature evolutions of the optical response of Sr₃(Ir_1−xMn_x)₂O₇single crystals with 0 ≤ x ≤ 0.36 by utilizing infrared spectroscopy. Substitution of 3dtransition metal Mn ions into Sr₃Ir₂O₇is expected to induce an insulator-to-metal transition via the decrease in the magnitude of the spin–orbit coupling and the hole doping. In sharp contrast, our data reveal the resilience of the spin–orbit coupling and the incoherent character of the charge transport. Upon Mn substitution, an incoherent in-gap excitation at about 0.25 eV appeared with the decrease in the strength of the optical transitions between the effective total angular momentumJ_effbands of the Ir ions. The resonance energies of the optical transitions between theJ_effbands which are directly proportional to the magnitude of the spin–orbit coupling hardly varied. In addition to these evolutions of the low-energy response, Mn substitution led to the emergence of a distinct high-energy optical excitation at about 1.2 eV which is larger than the resonance energies of the optical transitions between theJ_effbands. This observation indicates that the Mn 3dstates are located away from the Ir 5dstates in energy and that the large difference in the on-site energies of the transition metal ions is responsible for the incoherent charge transport and the robustness of the spin–orbit coupling. The effect of Mn substitution was also registered in the temperature dependence of the electronic response. The anomaly in the optical response of the parent compound observed at the antiferromagnetic transition temperature is notably suppressed in the Mn-doped compounds despite the persistence of the long-range antiferromagnetic ordering. The suppression of the spin-charge coupling could be related to charge disproportionation of the Ir ions. 

The motility mechanisms of microorganisms are critical virulence factors, enabling their spread and survival during infection. Motility is frequently characterized by qualitative analysis of macroscopic colonies, yet the standard quantification method has mainly been limited to manual measurement. Recent studies have applied deep learning for classification and segmentation of specific microbial species in microscopic images, but less work has focused on macroscopic colony analysis. Here, we advance computational tools for analyzing colonies of Proteus mirabilis, a bacterium that produces a macroscopic bullseye-like pattern via periodic swarming, a process implicated in its virulence. We present a dual-task pipeline for segmenting (1) the macroscopic colony including faint outer swarm rings, and (2) internal ring boundaries, unique features of oscillatory swarming. Our convolutional neural network for patch-based colony segmentation and U-Net with a VGG-11 encoder for ring boundary segmentation achieved test Dice scores of 93.28% and 83.24%, respectively. The predicted masks at times improved on the ground truths from our automated annotation algorithms. We demonstrate how application of our pipeline to a typical swarming assay enables ease of colony analysis and precise measurements of more complex pattern features than those which have been historically quantified. An implementation of our work can be found on https://github.com/daninolab/proteus-mirabilis. 

Abstract We report on optical spectroscopic study of the Sr 3 (Ir 1- x Ru x ) 2 O 7 system over a wide doping regime. We find that the changes in the electronic structure occur in the limited range of the concentration of Ru ions where the insulator–metal transition occurs. In the insulating regime, the electronic structure associated with the effective total angular momentum J eff  = 1/2 Mott state remains robust against Ru doping, indicating the localization of the doped holes. Upon entering the metallic regime, the Mott gap collapses and the Drude-like peak with strange metallic character appears. The evolution of the electronic structure registered in the optical data can be explained in terms of a percolative insulator–metal transition. The phonon spectra display anomalous doping evolution of the lineshapes. While the phonon modes of the compounds deep in the insulating and metallic regimes are almost symmetric, those of the semiconducting compound with x  = 0.34 in close proximity to the doping-driven insulator–metal transition show a pronounced asymmetry. The temperature evolution of the phonon modes of the x  = 0.34 compound reveals the asymmetry is enhanced in the antiferromagnetic state. We discuss roles of the S  = 1 spins of the Ru ions and charge excitations for the conspicuous lineshape asymmetry of the x  = 0.34 compound.