Search for: All records

Abstract AstroSage-Llama-3.1-8B is a domain-specialized natural-language AI assistant tailored for research in astronomy, astrophysics, cosmology, and astronomical instrumentation. Trained on the complete collection of astronomy-related arXiv papers from 2007 to 2024 along with millions of synthetically-generated question-answer pairs and other astronomical literature, AstroSage-Llama-3.1-8B demonstrates remarkable proficiency on a wide range of questions. AstroSage-Llama-3.1-8B scores 80.9% on the AstroMLab-1 benchmark, greatly outperforming all models—proprietary and open-weight—in the 8-billion parameter class, and performing on par with GPT-4o. This achievement demonstrates the potential of domain specialization in AI, suggesting that focused training can yield capabilities exceeding those of much larger, general-purpose models. AstroSage-Llama-3.1-8B is freely available, enabling widespread access to advanced AI capabilities for astronomical education and research. 

In this paper, we systematically study the evolution of the Universe within the framework of a modified loop quantum cosmological model (mLQC-I) using various inflationary potentials, including chaotic, Starobinsky, generalized Starobinsky, polynomials of the first and second kinds, generalized T-models and natural inflation. In all these models, the big bang singularity is replaced by a quantum bounce, and the evolution of the Universe, both before and after the bounce, is universal and weakly dependent on the inflationary potentials, as long as the evolution is dominated by the kinetic energy of the inflaton at the bounce. In particular, the pre-bounce evolution can be universally divided into three different phases: pre-bouncing, pre-transition, and pre-de Sitter. The pre-bouncing phase occurs immediately before the quantum bounce, during which the evolution of the Universe is dominated by the kinetic energy of the inflaton. Thus, the equation of state of the inflaton is about one, w(ϕ)≃1. Soon, the inflation potential takes over, so w(ϕ) rapidly falls from one to negative one. This pre-transition phase is very short and quickly turns into the pre-de Sitter phase, whereby the effective cosmological constant of Planck size takes over and dominates the rest of the contracting phase. Throughout the entire pre-bounce regime, the evolution of both the expansion factor and the inflaton can be approximated by universal analytical solutions, independent of the specific inflation potentials. 

Abstract Numerous eddies in the coastal ocean may experience distortion due to interactions with the ambient flow. Here we investigate how coastal submesoscale eddy distortion affects the cross‐shore and vertical tracer transport using a high‐resolution, wave‐current coupled model in the La Jolla Canyon region within the Southern California Bight. Model dye is released representing freshwater discharges. Model validations show that the coupled model has weaker stratification and weaker currents. Analyses primarily focus on an eddy‐induced cross‐shore dye transport event. The results show that, the coastal eddy is squeezed in the alongshore direction and extends in the cross‐shore direction, driving cross‐shore dye transport. Along a mid‐shelf boundary, the total cross‐shore transport is found to be dominated by the along‐boundary perturbation flow, which is linked to the eddy distortion. In addition, this coastal eddy also possesses vigorous vertical motions. The vertical velocity is more negative on the eddy northern side, favoring local dye subduction. This N‐S vertical velocity asymmetry may largely be induced by the topographic beta effect and the weaker modeled stratification may strengthen this effect. Overall, coastal eddy distortion contributes to the offshore tracer transport and induces spatially non‐uniform vertical dye flux. 

Vision-Language Models (VLMs) are trained on vast amounts of data captured by humans emulating our understanding of the world. However, known as visual illusions, human's perception of reality isn't always faithful to the physical world. This raises a key question: do VLMs have the similar kind of illusions as humans do, or do they faithfully learn to represent reality? To investigate this question, we build a dataset containing five types of visual illusions and formulate four tasks to examine visual illusions in state-of-the-art VLMs. Our findings have shown that although the overall alignment is low, larger models are closer to human perception and more susceptible to visual illusions. Our dataset and initial findings will promote a better understanding of visual illusions in humans and machines and provide a stepping stone for future computational models that can better align humans and machines in perceiving and communicating about the shared visual world. 

In this paper, we study analytical approximate solutions for second-order homogeneous differential equations with the existence of only two turning points (but without poles) by using the uniform asymptotic approximation (UAA) method. To be more concrete, we consider the Pöschl–Teller (PT) potential, for which analytical solutions are known. Depending on the values of the parameters involved in the PT potential, we find that the upper bounds of the errors of the approximate solutions in general are ≲0.15∼10% for the first-order approximation of the UAA method. The approximations can be easily extended to high orders, for which the errors are expected to be much smaller. Such obtained analytical solutions can be used to study cosmological perturbations in the framework of quantum cosmology as well as quasi-normal modes of black holes.