Search

Recently, large language and vision models (LLVMs) have received significant attention and development efforts due to their remarkable generalization performance across a wide range of tasks requiring perception and cognitive abilities. A key factor behind their success is their simple architecture, which consists of a vision encoder, a projector, and a large language model (LLM). Despite their achievements in advanced reasoning tasks, their performance on fundamental perception-related tasks (e.g., MMVP) remains surprisingly low. This discrepancy raises the question of how LLVMs truly perceive images and exploit the advantages of the vision encoder. To address this, we systematically investigate this question regarding several aspects: permutation invariance, robustness, math reasoning, alignment preserving and importance, by evaluating the most common LLVM's families (i.e., LLaVA) across 10 evaluation benchmarks. Our extensive experiments reveal several intriguing properties of current LLVMs: (1) they internally process the image in a global manner, even when the order of visual patch sequences is randomly permuted; (2) they are sometimes able to solve math problems without fully perceiving detailed numerical information; (3) the cross-modal alignment is overfitted to complex reasoning tasks, thereby, causing them to lose some of the original perceptual capabilities of their vision encoder; (4) the representation space in the lower layers (<25%) plays a crucial role in determining performance and enhancing visual understanding. Lastly, based on the above observations, we suggest potential future directions for building better LLVMs and constructing more challenging evaluation benchmarks., Comment: Code is available in https://github.com/passing2961/IP-LLVM

With numerical simulations that employ adaptive ray-tracing (ART) for radiative transfer at the same time as evolving gas magnetohydrodynamics, thermodynamics, and photochemistry, it is possible to obtain a high resolution view of ultraviolet (UV) fields and their effects in realistic models of the multiphase interstellar medium. Here, we analyze results from TIGRESS-NCR simulations, which follow both far-UV (FUV) wavelengths, important for photoelectric heating and PAH excitation, and the Lyman continuum (LyC), which photoionizes hydrogen. Considering two models, representing solar neighborhood and inner galaxy conditions, we characterize the spatial distribution and time variation of UV radiation fields, and quantify their correlations with gas. We compare four approximate models for the FUV to simulated values to evaluate alternatives when full ART is infeasible. By convolving FUV radiation with density, we produce mock maps of dust emission. We introduce a method to calibrate mid-IR observations, for example from JWST, to obtain high resolution gas surface density maps. We then consider the LyC radiation field, finding most of the gas exposed to this radiation to be in ionization-recombination equilibrium and to have a low neutral fraction. Additionally, we characterize the ionization parameter as a function of environment. Using a simplified model of the LyC radiation field, we produce synthetic maps of emission measure (EM). We show that the simplified model can be used to extract an estimate of the neutral fraction of the photoionized gas and mean free path of ionizing radiation from observed EM maps in galaxies., Comment: 38 pages, 26 figures, accepted for publication in ApJ

Electroencephalography (EEG) offers non-invasive, real-time mental workload assessment, which is crucial in high-stakes domains like aviation and medicine and for advancing brain-computer interface (BCI) technologies. This study introduces a customized ConvNeXt architecture, a powerful convolutional neural network, specifically adapted for EEG analysis. ConvNeXt addresses traditional EEG challenges like high dimensionality, noise, and variability, enhancing the precision of mental workload classification. Using the STEW dataset, the proposed ConvNeXt model is evaluated alongside SVM, EEGNet, and TSception on binary (No vs SIMKAP task) and ternary (SIMKAP multitask) class mental workload tasks. Results demonstrated that ConvNeXt significantly outperformed the other models, achieving accuracies of 95.76% for binary and 95.11% for multi-class classification. This demonstrates ConvNeXt's resilience and efficiency for EEG data analysis, establishing new standards for mental workload evaluation. These findings represent a considerable advancement in EEG-based mental workload estimation, laying the foundation for future improvements in cognitive state measurements. This has broad implications for safety, efficiency, and user experience across various scenarios. Integrating powerful neural networks such as ConvNeXt is a critical step forward in non-invasive cognitive monitoring., Comment: 8 Pages, 4 Figures, 1 Table

We present a new simulation suite for the star-forming interstellar medium (ISM) in galactic disks using the TIGRESS-NCR framework. Distinctive aspects of our simulation suite are: (1) sophisticated and comprehensive numerical treatments of essential physical processes including magnetohydrodynamics, self-gravity, and galactic differential rotation, as well as photochemistry, cooling, and heating coupled with ray-tracing UV radiation transfer and resolved supernova feedback and (2) wide parameter coverage including metallicity over $Z'\equiv Z/Z_\odot\sim0.1-3$, gas surface density $\Sigma_{\rm gas}\sim5-150 M_{\odot}{\rm pc^{-2}}$, and stellar surface density $\Sigma_{\rm star}\sim 1-50 M_{\odot}{\rm pc^{-2}}$. The range of emergent star formation rate surface density is $\Sigma_{\rm SFR}\sim 10^{-4}-0.5 M_{\odot}{\rm kpc^{-2}yr^{-1}}$ and ISM total midplane pressure is $P_{\rm tot}/k_B=10^3-10^6{\rm cm^{-3}K}$, with $P_{\rm tot}$ equal to the ISM weight $W$. For given $\Sigma_{\rm gas}$ and $\Sigma_{\rm star}$, we find $\Sigma_{\rm SFR} \propto Z'^{0.3}$. We provide an interpretation based on the pressure-regulated feedback-modulated (PRFM) star formation theory. We characterize feedback modulation in terms of the yield $\Upsilon$, defined as the ratio of each stress to $\Sigma_{\rm SFR}$. The thermal feedback yield varies sensitively with both weight and metallicity as $\Upsilon_{\rm th}\propto W^{-0.46}Z'^{-0.53}$, while the combined turbulent and magnetic feedback yield shows weaker dependence $\Upsilon_{\rm turb+mag}\propto W^{-0.22}Z'^{-0.18}$. The reduction in $\Sigma_{\rm SFR}$ at low metallicity is due mainly to enhanced thermal feedback yield, resulting from reduced attenuation of UV radiation. With the metallicity-dependent calibrations we provide, PRFM theory can be used for a new subgrid star formation prescription in cosmological simulations where the ISM is unresolved., Comment: Resubmitted to ApJ after minor revision

Bubbles driven by energy and mass injection from small scales are ubiquitous in astrophysical fluid systems and essential to feedback across multiple scales. In particular, O stars in young clusters produce high velocity winds that create hot bubbles in the surrounding gas. We demonstrate that the dynamical evolution of these bubbles is critically dependent upon the geometry of their interfaces with their surroundings and the nature of heat transport across these interfaces. These factors together determine the amount of energy that can be lost from the interior through cooling at the interface, which in turn determines the ability of the bubble to do work on its surroundings. We further demonstrate that the scales relevant to physical dissipation across this interface are extremely difficult to resolve in global numerical simulations of bubbles for parameter values of interest. This means the dissipation driving evolution of these bubbles in numerical simulations is often of a numerical nature. We describe the physical and numerical principles that determine the level of dissipation in these simulations; we use this, along with a fractal model for the geometry of the interfaces, to explain differences in convergence behavior between hydrodynamical and magneto-hydrodynamical simulations presented here. We additionally derive an expression for momentum as a function of bubble radius expected when the relevant dissipative scales are resolved and show that it still results in efficiently-cooled solutions as postulated in previous work., Comment: 41 pages, 20 figures, submitted to ApJ, comments welcome

Aerospace structures often experience high strain rate events such as ballistic impact, crash, or crush. A material model has been developed that enhances the capability to simulate the dynamic response of composite materials under these loading conditions. The material model has been implemented into the commercially available transient dynamic finite element code LS-DYNA as MAT213. The model can simulate the nonlinear deformation, damage, and failure that takes place in a composite under dynamic loading conditions. The specific goal of this work is to characterize the MAT213 input for the representative material. The specific composite material being examined consists of T700G unidirectional carbon fibers and a low-melt PolyArylEtherKetone (LMPAEK) thermoplastic resin system. It is formally referred to as Toray TC1225 LMPAEK T700G. As the initial part of this work, this paper is focused on characterizing the material parameters for the MAT213 deformation model based on results obtained from multi-scale experimentation. The effort concentrated on characterizing the in-plane material response suitable for use with thin shell elements. For shell elements within MAT213, tabulated stress-strain results from tension and compression tests in the longitudinal and transverse directions and in-plane shear tests are required. Due to the difficulty of measuring small strains in the transverse direction, a multi-scale testing method was developed. Macro-scale testing is performed per the typical ASTM methods while micro-scale testing uses a microscope along with smaller coupon sizes to obtain the smaller strains in the transverse direction of each test. For both testing methods, a VIC-2D camera and software for digital image correlation analysis are used. Using the DIC combined with each test fixture, reliable stress and strain data are collected.

We investigated the effect of Ni doping in a recently proposed quantum spin liquid (QSL) candidate Li$_4$CuTeO$_6$. We performed a comprehensive study on the structural and magnetic properties. We find that the anti-site disorder between Li$^+$ and Cu$^{2+}$ persists until 50\% Ni doping in which Ni and Cu occupy different crystallographic sites. As a result, while Cu sits in both triangular and honeycomb layers in Li$_4$CuTeO$_6$, Ni forms only honeycomb layer in Li$_4$NiTeO$_6$ and Li$_4$Cu$_{0.5}$Ni$_{0.5}$TeO$_6$. Our magnetic susceptibility measurements show that the Weiss temperature decreases from -145.68 K in Li$_4$CuTeO$_6$ to -6.15 K in Li$_4$NiTeO$_6$ as Ni doping increases, and find no hint of magnetic ordering or freezing down to 1.8 K. Our analysis implies the existence of abundant low energy excitations in these materials., Comment: 7 pages, 4 figures

This work is focused on the experimental characterization of traction-separation laws for cohesive modeling of mode-II interlaminar fracture in composites. For the experimental investigation, damage progression in end-notched flexure specimens under three-point bending was captured using microscopic and macroscopic Digital Image Correlation (DIC) techniques. The specimens were geometrically scaled with three scaling levels for a size effect study. The fracture energy of the material was analyzed based on both linear elastic fracture mechanics and Bazant's type-II size effect law for comparison. For damage modeling, traction-separation laws were built from the experimental data, and the fracture process zones of the specimens were modeled using cohesive interactions. Difficulties in characterizing traction-separation laws merely relying on load-displacement curves (i.e., global behaviors) were demonstrated and discussed through extensive simulations. To address this issue, through-thickness deformation analysis and progressive DIC methods were proposed and implemented in this work. The observations and proposed methods herein will contribute to characterizing a single traction-separation law for a given composite material by capturing global and local fracture behaviors on any geometric scale.

This project is focused on investigating the structural performance of parts and structures produced using the latest additive manufacturing techniques. For additive manufacturing of test coupons, fused deposition modeling of fiber-reinforced polymer (RFP) and high-resolution low-force stereolithography (LFS) thermoset resin printing systems were employed. UV thermoset resin was used for LFS printing, while RFP printing adopted two different types of filaments: amorphous polycarbonate carbon fiber filaments and semi-crystalline Nylon 12 glass fiber filaments. For the experimental work, specimens were printed for tension, compression, and shear tests. Additionally, mode-II interlminar fracture in these specimens was explored. The elastic modulus and strength values of these specimens were compared with the data of oven-cured T700G/2510 composites. The experimental work herein will be extended to develop damage models for 3D-printed structural parts and structures for aerospace and space applications.

This project is focused on developing an experimental framework for characterizing non-associative plasticity flow rule coefficients through coupon-scale tests for the LS-DYNA MAT213 model. The main objective is to characterize these coefficients based on the multi-scale (i.e., both microscopic and macroscopic) full-field measurement of the evolution of strain and stress fields. This paper focuses on presenting the experimental work on characterizing the full-scale stress-strain curves of T700/LM-PAEK composites under tension, compression, and shear loads. The experimental data set was intended to build a deformation sub-model in the MAT213 model for the material. The strain data were collected using both microscopic and macroscopic digital image correlation techniques. The microscopic technique was particularly useful for fracture cases under small strains. A preliminary simulation result obtained from the MAT213 model is also presented in the paper. The experimental framework herein will be extended to characterize post-peak stress degradation in the composite material and to develop a damage sub-model for the material. This project will contribute to developing a simulation tool based on the MAT213 model for simulating the rate-dependent impact damages in composites under multi-axial loading.

This work is focused on investigating the impact of out-of-plane stitches on enhancing mode-II interlaminar fracture toughness (or energy) and characterizing damage progression and crack arrestment in stitched resin-infused composites. For the experimental work, End-Notched Flexure (ENF) quasi-isotropic specimens were manufactured using +/-45 non-crimp carbon-fiber fabrics through a resin-infusion process. Both stitched and unstitched specimen sets were designed for comparison. For a size effect study, the ENF specimens were geometrically scaled with three scaling levels. Based on the load-displacement data (i.e., global analysis), the fracture energy of the specimen material was analyzed using the compliance calibration method and a size effect theory. The fracture energy values were compared between the stitched and unstitched cases to characterize the enhanced fracture toughness of stitched composites. For local analysis, two types of digital image correlation (DIC) systems were employed: microscopic and macroscopic (i.e., coupon-scale) DIC systems. By analyzing in-plane displacement through the thickness, separation development was characterized along predicted fracture process zones. The impact of out-of-plane stitches on separation propagation along fracture process zones was discussed based on the DIC analysis. This work will contribute to developing a high-fidelity damage model for stitched resin-infused composites in the form of a traction-separation for high-speed aircraft applications.

Sentiment analysis (SA) is an emerging field in text mining. It is the process of computationally identifying and categorizing opinions expressed in a piece of text over different social media platforms. Social media plays an essential role in knowing the customer mindset towards a product, services, and the latest market trends. Most organizations depend on the customer's response and feedback to upgrade their offered products and services. SA or opinion mining seems to be a promising research area for various domains. It plays a vital role in analyzing big data generated daily in structured and unstructured formats over the internet. This survey paper defines sentiment and its recent research and development in different domains, including voice, images, videos, and text. The challenges and opportunities of sentiment analysis are also discussed in the paper. \keywords{Sentiment Analysis, Machine Learning, Lexicon-based approach, Deep Learning, Natural Language Processing}

In quantum machine learning, algorithms with parameterized quantum circuits (PQC) based on a hardware-efficient ansatz (HEA) offer the potential for speed-ups over traditional classical algorithms. While much attention has been devoted to supervised learning tasks, unsupervised learning using PQC remains relatively unexplored. One promising approach within quantum machine learning involves optimizing fewer parameters in PQC than in its classical counterparts, under the assumption that a sub-optimal solution exists within the Hilbert space. In this paper, we introduce the Variational Quantum Approximate Spectral Clustering (VQASC) algorithm - a NISQ-compatible method that requires optimization of fewer parameters than the system size, N, traditionally required in classical problems. We present numerical results from both synthetic and real-world datasets. Furthermore, we propose a descriptor, complemented by numerical analysis, to identify an appropriate ansatz circuit tailored for VQASC., Comment: 21 pages, 6 figures

Elemental phosphorus exhibits fascinating structural varieties and versatile properties. The unique nature of phosphorus bonds can lead to the formation of extremely complex structures, and detailed structural information on some phosphorus polymorphs is yet to be investigated. In this study, we investigated an unidentified crystalline phase of phosphorus, type-II red phosphorus (RP), by combining state-of-the-art structural characterization techniques. Electron diffraction tomography, atomic-resolution scanning transmission electron microscopy (STEM), powder X-ray diffraction, and Raman spectroscopy were concurrently used to elucidate the hidden structural motifs and their packing in type-II RP. Electron diffraction tomography, performed using individual crystalline nanowires, was used to identify a triclinic unit cell with volume of 5330 {\AA}^3, the largest unit cell for elemental phosphorus crystals up to now, which contains approximately 250 phosphorus atoms. Atomic-resolution STEM imaging, which was performed along different crystal-zone axes, confirmed that the twisted wavy tubular motif is the basic building block of type-II RP. Our study discovered and presented a new variation of building blocks in phosphorus, and it provides insights to clarify the complexities observed in phosphorus as well as other relevant systems., Comment: 16 pages, 5 figures, 1 table

Chemistry plays a key role in many aspects of astrophysical fluids. Atoms and molecules are agents for heating and cooling, determine the ionization fraction, serve as observational tracers, and build the molecular foundation of life. We present the implementation of a chemistry module in the publicly available magneto-hydrodynamic code Athena++. We implement several chemical networks and heating and cooling processes suitable for simulating the interstellar medium (ISM). A general chemical network framework in the KIDA format is also included, allowing the user to easily implement their own chemistry. Radiation transfer and cosmic-ray ionization are coupled with chemistry and solved with the simple six-ray approximation. The chemical and thermal processes are evolved as a system of coupled ODEs with an implicit solver from the CVODE library. We perform and present a series of tests to ensure the numerical accuracy and convergence of the code. Many tests combine chemistry with gas dynamics, including comparisons with analytic solutions, 1D problems of the photo-dissociation regions and shocks, and realistic 3D simulations of the turbulent ISM. We release the code with the new public version of Athena++, aiming to provide a robust and flexible code for the astrochemical simulation community.

As sharing images in an instant message is a crucial factor, there has been active research on learning an image-text multi-modal dialogue models. However, training a well-generalized multi-modal dialogue model remains challenging due to the low quality and limited diversity of images per dialogue in existing multi-modal dialogue datasets. In this paper, we propose an automated pipeline to construct a multi-modal dialogue dataset, ensuring both dialogue quality and image diversity without requiring minimum human effort. In our pipeline, to guarantee the coherence between images and dialogue, we prompt GPT-4 to infer potential image-sharing moments - specifically, the utterance, speaker, rationale, and image description. Furthermore, we leverage CLIP similarity to maintain consistency between aligned multiple images to the utterance. Through this pipeline, we introduce DialogCC, a high-quality and diverse multi-modal dialogue dataset that surpasses existing datasets in terms of quality and diversity in human evaluation. Our comprehensive experiments highlight that when multi-modal dialogue models are trained using our dataset, their generalization performance on unseen dialogue datasets is significantly enhanced. We make our source code and dataset publicly available., Comment: NAACL 2024

Vision Transformer (ViT) extracts the final representation from either class token or an average of all patch tokens, following the architecture of Transformer in Natural Language Processing (NLP) or Convolutional Neural Networks (CNNs) in computer vision. However, studies for the best way of aggregating the patch tokens are still limited to average pooling, while widely-used pooling strategies, such as max and GeM pooling, can be considered. Despite their effectiveness, the existing pooling strategies do not consider the architecture of ViT and the channel-wise difference in the activation maps, aggregating the crucial and trivial channels with the same importance. In this paper, we present Group Generalized Mean (GGeM) pooling as a simple yet powerful pooling strategy for ViT. GGeM divides the channels into groups and computes GeM pooling with a shared pooling parameter per group. As ViT groups the channels via a multi-head attention mechanism, grouping the channels by GGeM leads to lower head-wise dependence while amplifying important channels on the activation maps. Exploiting GGeM shows 0.1%p to 0.7%p performance boosts compared to the baselines and achieves state-of-the-art performance for ViT-Base and ViT-Large models in ImageNet-1K classification task. Moreover, GGeM outperforms the existing pooling strategies on image retrieval and multi-modal representation learning tasks, demonstrating the superiority of GGeM for a variety of tasks. GGeM is a simple algorithm in that only a few lines of code are necessary for implementation.

Massive, young stars are the main source of energy that maintains multiphase structure and turbulence in the interstellar medium (ISM), and without this "feedback" the star formation rate (SFR) would be much higher than is observed. Rapid energy loss in the ISM and efficient energy recovery by stellar feedback lead to co-regulation of SFRs and the ISM state. Realistic approaches to this problem should solve the dynamical evolution of the ISM, including star formation, and the input of feedback energy self-consistently and accurately. Here, we present the TIGRESS-NCR numerical framework, in which UV radiation, supernovae, cooling and heating processes, and gravitational collapse are modeled explicitly. We use an adaptive ray tracing method for UV radiation transfer from star clusters represented by sink particles, accounting for attenuation by dust and gas. We solve photon-driven chemical equations to determine the abundances of H (time-dependent) and C/O-bearing species (steady-state), which then set cooling and heating rates self-consistently. Applying these methods, we present high-resolution magnetohydrodynamics simulations of differentially rotating local galactic disks representing typical conditions of nearby star-forming galaxies. We analyze ISM properties and phase distributions and show good agreement with existing multiwavelength galactic observations. We measure midplane pressure components (turbulent, thermal, and magnetic) and the weight, demonstrating that vertical dynamical equilibrium holds. We quantify the ratios of pressure components to the SFR surface density, which we call the feedback yields. The TIGRESS-NCR framework will allow for a wide range of parameter exploration, including low metallicity system., Comment: ApJ accepted

We present an efficient heating/cooling method coupled with chemistry and ultraviolet (UV) radiative transfer, which can be applied to numerical simulations of the interstellar medium (ISM). We follow the time-dependent evolution of hydrogen species (H$_2$, H, H$^+$), assume carbon/oxygen species (C, C$^+$, CO, O, and O$^+$) are in formation-destruction balance given the non-steady hydrogen abundances, and include essential heating/cooling processes needed to capture thermodynamics of all ISM phases. UV radiation from discrete point sources and the diffuse background is followed through adaptive ray tracing and a six-ray approximation, respectively, allowing for H$_2$ self-shielding; cosmic ray (CR) heating and ionization are also included. To validate our methods and demonstrate their application for a range of density, metallicity, and radiation field, we conduct a series of tests, including the equilibrium curves of thermal pressure vs. density, the chemical and thermal structure in photo-dissociation regions, H I-to-H$_2$ transitions, and the expansion of H II regions and radiative supernova remnants. Careful treatment of photochemistry and CR ionization is essential for many aspects of ISM physics, including identifying the thermal pressure at which cold and warm neutral phases co-exist. We caution that many current heating and cooling treatments used in galaxy formation simulations do not reproduce the correct thermal pressure and ionization fraction in the neutral ISM. Our new model is implemented in the MHD code Athena and incorporated in the TIGRESS simulation framework, for use in studying the star-forming ISM in a wide range of environments., Comment: 57 pages, 22 figures; accepted for publication in ApJS

Magnetic fields ($\textbf{B}$) are an important factor that controls the star formation process. The leading method to observe $\textbf{B}$ is using polarized thermal emission from dust grains aligned with $\textbf{B}$. However, in dense environments such as protostellar cores, dust grains may have inefficient alignment due to strong gas randomizations, so that using dust polarization to trace $\textbf{B}$ is uncertain. Hoang $\&$ Lazarian (2016) demonstrated that the grain alignment by RAdiative Torques is enhanced if dust grains contain embedded iron inclusions. Here we extend POLARIS code to study the effect of iron inclusions on grain alignment and thermal dust polarization toward a protostellar core, assuming uniform magnetic fields. We found that paramagnetic grains produce a low polarization degree of $p \sim 1\%$ in the envelope and negligible $p \ll 1\%$ in the central region due to the loss of grain alignment. In contrast, grains with a high level of iron inclusions can have perfect alignment and produce high $p \sim 40\%$ in the envelope and low $p \leq 10\%$ in the central region. Grains with a moderate level of iron inclusions induce the polarization flipping from $\textbf{P}$ $\parallel$ $\textbf{B}$ at millimeter to $\textbf{P}$ $\perp$ $\textbf{B}$ at submillimeter due to the change in the internal alignment caused by slow internal relaxation. The weak alignment of very large grains of $a \geq 10\mu m$ reduces the polarization by dichroic extinction at submillimeter wavelengths. We found a positive correlation between p and the level of iron inclusions, which opens a new window to constrain the abundance of irons locked in dust through dust polarimetry., Comment: 39 pages, 27 figures

Feature Pyramid Network (FPN) has been an essential module for object detection models to consider various scales of an object. However, average precision (AP) on small objects is relatively lower than AP on medium and large objects. The reason is why the deeper layer of CNN causes information loss as feature extraction level. We propose a new scale sequence (S^2) feature extraction of FPN to strengthen feature information of small objects. We consider FPN structure as scale-space and extract scale sequence (S^2) feature by 3D convolution on the level axis of FPN. It is basically scale invariant feature and is built on high-resolution pyramid feature map for small objects. Furthermore, the proposed S^2 feature can be extended to most object detection models based on FPN. We demonstrate the proposed S2 feature can improve the performance of both one-stage and two-stage detectors on MS COCO dataset. Based on the proposed S2 feature, we achieve upto 1.3% and 1.1% of AP improvement for YOLOv4-P5 and YOLOv4-P6, respectively. For Faster RCNN and Mask R-CNN, we observe upto 2.0% and 1.6% of AP improvement with the suggested S^2 feature, respectively.