Hasil untuk "Astrophysics"

K. Freeman, J. Bland-Hawthorn

▪ Abstract The formation and evolution of galaxies is one of the great outstanding problems of astrophysics. Within the broad context of hierachical structure formation, we have only a crude picture of how galaxies like our own came into existence. A detailed physical picture where individual stellar populations can be associated with (tagged to) elements of the protocloud is far beyond our current understanding. Important clues have begun to emerge from both the Galaxy (near-field cosmology) and the high redshift universe (far-field cosmology). Here we focus on the fossil evidence provided by the Galaxy. Detailed studies of the Galaxy lie at the core of understanding the complex processes involved in baryon dissipation. This is a necessary first step toward achieving a successful theory of galaxy formation.

Alex Simpson, M. Visser

So-called “regular black holes” are a topic currently of considerable interest in the general relativity and astrophysics communities. Herein we investigate a particularly interesting regular black hole spacetime described by the line element This spacetime neatly interpolates between the standard Schwarzschild black hole and the Morris-Thorne traversable wormhole; at intermediate stages passing through a black-bounce (into a future incarnation of the universe), an extremal null-bounce (into a future incarnation of the universe), and a traversable wormhole. As long as the parameter a is non-zero the geometry is everywhere regular, so one has a somewhat unusual form of “regular black hole”, where the “origin” r=0 can be either spacelike, null, or timelike. Thus this spacetime generalizes and broadens the class of “regular black holes” beyond those usually considered.

B. Ludwig, M. R. Drout, Y. Götberg et al.

Most massive stars (∼8–25 M _⊙ ) interact with a binary companion during their lifetimes. These interactions can remove the hydrogen-rich envelope, producing intermediate-mass (∼2–8 M _⊙ ) and helium-rich stars. These “stripped stars” are predicted to emit predominantly in the ultraviolet (UV) and can therefore be identified via a UV excess—provided they are not outshone by their companion. However, despite their importance to binary evolution, supernovae, and ionizing feedback, few stripped stars have been confirmed. This is likely due to the scarcity of wide-field, high angular-resolution, UV surveys of stellar populations with reliable distances and extinction estimates. To address this, we present the Stripped-Star Ultraviolet Magellanic Clouds Survey catalog. We use the Tractor forward modeling software to perform point-spread function photometry on 2420 Swift UVOT images of the LMC and SMC. The resulting catalog contains 734,862 sources in three UV filters to a depth of ∼20 Vega mag. We perform validation tests on the photometry pipeline and highlight the catalog’s broad applicability. We then identify sources with excess UV light compared to main-sequence stars and apply a series of quality cuts. From this, we identify 522 candidate stripped stars in the LMC and 298 in the SMC. We assess the potential contamination from other UV excess systems and argue the dominant uncertainty to be dust: early main-sequence stars can mimic the colors of stripped-star binaries when extinction is overcorrected. This survey lays the groundwork for the first systematic census of stripped stars and opens new windows into binary evolution and massive star populations.

Yoshiaki Ono, Masami Ouchi, Yuichi Harikane et al.

We homogeneously investigate the morphological properties of 169 galaxies at z  ∼ 10–16 with deep JWST NIRCam images employing our established techniques of GALFIT modeling and uncertainty evaluation (systematics+statistics). We obtain effective radii r _e ranging 20–500 pc, with a distribution significantly broader than the scatter made by the uncertainties. We find that the r _e distribution is well described by a log-normal distribution with a mean of ${r}_{{{\rm{e}}}}=13{3}_{-12}^{+13}$ pc and a standard deviation of ${\sigma }_{{\mathrm{ln}}\,{r}_{{{\rm{e}}}}}=0.52\pm 0.08$ . The standard deviation is comparable to that of local galaxies, indicating no significant evolution over z  ∼ 0–10. We also find that the axis ratio distribution is nearly uniform, statistically similar to that of local spirals. We estimate the virial radius r _vir from the stellar masses via the star formation main sequence and stellar-to-halo mass relation, obtaining a stellar-to-halo size ratio ${r}_{{{\rm{e}}}}/{r}_{{\mathrm{vir}}}=0.01{5}_{-0.005}^{+0.015}$ , which is comparable to those of star-forming galaxies in the local and low- z Universe. Our results of (1) the log-normal r _e distribution, (2) the uniform axis ratio distribution, and (3) a mean radial profile consistent with an exponential profile ( n  = 1.3 ± 0.6) suggest that galaxies at z  ∼ 10–16 generally follow the classical galaxy disk formation scenario with a specific disk angular momentum fraction of j _d / m _d  ∼ 0.5–1. Interestingly, we identify two remarkable outliers GN-z11 ( z _spec  = 10.60) and GHZ2 ( z _spec  = 12.34) with ${r}_{{{\rm{e}}}}=5{5}_{-6}^{+5}$ pc and 39 ± 11 pc, respectively, that may not be explained by disk structures but by active galactic nucleus or compact star-forming galaxies merging underway in short periods of time, as reproduced in numerical simulations.

S. Tulasi Ram, Deeksha Rai, B. Nilam et al.

Abstract The intensity of storm‐time disturbance in the ground magnetic field varies significantly at different longitudes due to the magnetic local time (MLT) dependent contributions from different magnetospheric and ionospheric currents. Local geomagnetic field disturbances at low‐to‐mid latitudes often deviate considerably from the global depression represented by symmetric geomagnetic storm indices (such as Dst/SymH/SMR). In this study, we quantitatively investigated the geomagnetic horizontal field depressions (ΔH) at different local time sectors, compared to the longitudinally averaged SuperMAG Ring current (SMR), at eleven low‐latitude stations during a large number (665) of geomagnetic storms that occurred from 1996 to 2024. The relative disturbances (i.e., ΔH‐SMR) exhibit significant asymmetry with respect to MLT, which further varies with storm evolution, intensity, and phase. The MLT asymmetry of ΔH grows rapidly in the early main phase and then grows gradually with storm intensity. Further, the MLT sector of weakest/strongest ΔH depression shifts from post‐dawn/post‐dusk to pre‐dawn/pre‐dusk periods as storm intensity increases. Finally, an empirical model is derived that can quantitatively represent the MLT variations in the low‐latitude ΔH disturbances during geomagnetic storms. This model is very useful in estimating the low‐latitude geomagnetic field disturbances at different longitudes/MLT sectors from the global SMR index and can have significant applications in space weather studies.

Haixiang Gao

This paper investigates the problem of computation offloading and resource allocation in an integrated space–air–sea network based on unmanned aerial vehicle (UAV) and low Earth orbit (LEO) satellites supporting Maritime Internet of Things (M-IoT) devices. Considering the complex, dynamic environment comprising M-IoT devices, UAVs and LEO satellites, traditional optimization methods encounter significant limitations due to non-convexity and the combinatorial explosion in possible solutions. A multi-agent deep deterministic policy gradient (MADDPG)-based optimization algorithm is proposed to address these challenges. This algorithm is designed to minimize the total system costs, balancing energy consumption and latency through partial task offloading within a cloud–edge-device collaborative mobile edge computing (MEC) system. A comprehensive system model is proposed, with the problem formulated as a partially observable Markov decision process (POMDP) that integrates association control, power control, computing resource allocation, and task distribution. Each M-IoT device and UAV acts as an intelligent agent, collaboratively learning the optimal offloading strategies through a centralized training and decentralized execution framework inherent in the MADDPG. The numerical simulations validate the effectiveness of the proposed MADDPG-based approach, which demonstrates rapid convergence and significantly outperforms baseline methods, and indicate that the proposed MADDPG-based algorithm reduces the total system cost by 15–60% specifically.

Demosthenes Ellinas, Giorgio Kaniadakis

A novel approach to the quantum version of <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>κ</mi></semantics></math></inline-formula>-entropy that incorporates it into the conceptual, mathematical and operational framework of quantum computation is put forward. Various alternative expressions stemming from its definition emphasizing computational and algorithmic aspects are worked out: First, for the case of canonical Gibbs states, it is shown that <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>κ</mi></semantics></math></inline-formula>-entropy is cast in the form of an expectation value for an observable that is determined. Also, an operational method named “the two-temperatures protocol” is introduced that provides a way to obtain the <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>κ</mi></semantics></math></inline-formula>-entropy in terms of the partition functions of two auxiliary Gibbs states with temperatures <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>κ</mi></semantics></math></inline-formula>-shifted above, the hot-system, and <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>κ</mi></semantics></math></inline-formula>-shifted below, the cold-system, with respect to the original system temperature. That protocol provides physical procedures for evaluating entropy for any <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>κ</mi></semantics></math></inline-formula>. Second, two novel additional ways of expressing the <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>κ</mi></semantics></math></inline-formula>-entropy are further introduced. One determined by a non-negativity definite quantum channel, with Kraus-like operator sum representation and its extension to a unitary dilation via a qubit ancilla. Another given as a simulation of the <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>κ</mi></semantics></math></inline-formula>-entropy via the quantum circuit of a generalized version of the Hadamard test. Third, a simple inter-relation of the von Neumann entropy and the quantum <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>κ</mi></semantics></math></inline-formula>-entropy is worked out and a bound of their difference is evaluated and interpreted. Also the effect on the <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>κ</mi></semantics></math></inline-formula>-entropy of quantum noise, implemented as a random unitary quantum channel acting in the system’s density matrix, is addressed and a bound on the entropy, depending on the spectral properties of the noisy channel and the system’s density matrix, is evaluated. The results obtained amount to a quantum computational tool-box for the <inline-formula><math xmlns="http://www.w3.org/1998/Math/MathML" display="inline"><semantics><mi>κ</mi></semantics></math></inline-formula>-entropy that enhances its applicability in practical problems.

Michaela Vítková, Rafael Brahm, Trifon Trifonov et al.

We present a joint analysis of transit timing variations (TTVs) and Doppler data for the transiting exoplanet system TOI-4504. TOI-4504 c is a warm Jupiter-mass planet that exhibits the largest known TTVs, with a peak-to-node amplitude of ∼2 days, the largest value ever observed, and a superperiod of ~930 days. TOI-4504 b and c were identified in public Transiting Exoplanet Survey Satellite (TESS) data, while the TTVs observed in TOI-4504 c, together with radial velocity (RV) data collected with FEROS, allowed us to uncover a third, nontransiting planet in this system, TOI-4504 d. We were able to detect transits of TOI-4504 b in the TESS data with a period of 2.4261 ±  0.0001 days and derive a radius of 2.69 ±  0.19 R _⊕ . The RV scatter of TOI-4504 was too large to constrain the mass of TOI-4504 b, but the RV signals of TOI-4504 c and d were sufficiently large to measure their masses. The TTV+RV dynamical model we apply confirms TOI-4504 c as a warm Jupiter planet with an osculating period of 82.54 ±  0.02 days, a mass of 3.77 ±  0.18 M _J , and a radius of 0.99 ±  0.05 R _J , while the nontransiting planet TOI-4504 d has an orbital period of 40.56 ±  0.04 days and a mass of 1.42 ${}_{-0.06}^{+0.07}$ M _J . We present the discovery of a system with three exoplanets: a hot sub-Neptune and two warm Jupiter planets. The gas giant pair is stable and likely locked in a first-order 2:1 mean-motion resonance (MMR). The TOI-4504 system is an important addition to MMR pairs, whose increasing occurrence supports a smooth migration into a resonant configuration during the protoplanetary disk phase.

Rubens E. G. Machado, Ricardo C. Volert, Richards P. Albuquerque et al.

Abell 1758 ( z ∼ 0.278) is a galaxy cluster composed of two structures, A1758N and A1758S, separated by ∼2.2 Mpc. The northern cluster is itself a dissociative merging cluster that has already been modeled by dedicated simulations. Recent radio observations revealed the existence of a previously undetected bridge connecting A1758N and A1758S. New simulations are now needed to take into account the presence of A1758S. We wish to evaluate which orbital configuration would be compatible with a bridge between the clusters. Using N -body hydrodynamical simulations that build upon the previous model, we explore different scenarios that could have led to the current observed configuration. Five types of orbital approaches were tested: radial, tangential, vertical, postapocentric, and outgoing. We found that the incoming simulated scenarios are generally consistent with mild enhancements of gas density between the approaching clusters. The mock X-ray images exhibit a detectable bridge in all cases. Compared to measurements of Chandra data, the amplitude of the X-ray excess is overestimated by a factor of ∼2–3 in the best simulations. The scenario of tangential approach proved to be the one that best matches the properties of the profiles of X-ray surface brightness. The scenarios of radial approach of vertical approach are also marginally compatible.

H. Abe, K. Abe, S. Abe et al.

The Cherenkov Telescope Array (CTA) is a next-generation ground-based observatory for gamma-ray astronomy at very high energies. The Large-Sized Telescope prototype (LST-1) is located at the CTA-North site, on the Canary Island of La Palma. LSTs are designed to provide optimal performance in the lowest part of the energy range covered by CTA, down to ≃20 GeV. LST-1 started performing astronomical observations in 2019 November, during its commissioning phase, and it has been taking data ever since. We present the first LST-1 observations of the Crab Nebula, the standard candle of very-high-energy gamma-ray astronomy, and use them, together with simulations, to assess the performance of the telescope. LST-1 has reached the expected performance during its commissioning period—only a minor adjustment of the preexisting simulations was needed to match the telescope’s behavior. The energy threshold at trigger level is around 20 GeV, rising to ≃30 GeV after data analysis. Performance parameters depend strongly on energy, and on the strength of the gamma-ray selection cuts in the analysis: angular resolution ranges from 0.°12–0.°40, and energy resolution from 15%–50%. Flux sensitivity is around 1.1% of the Crab Nebula flux above 250 GeV for a 50 hr observation (12% for 30 minutes). The spectral energy distribution (in the 0.03–30 TeV range) and the light curve obtained for the Crab Nebula agree with previous measurements, considering statistical and systematic uncertainties. A clear periodic signal is also detected from the pulsar at the center of the Nebula.

L. D. Matthews, N. R. Evans, M. P. Rupen

We report the detection of 15 GHz radio continuum emission associated with the classical Cepheid variable star δ Cephei ( δ Cep) based on observations with the Karl G. Jansky Very Large Array. Our results constitute the first probable detection of radio continuum emission from a classical Cepheid. We observed the star at pulsation phase ϕ ≈ 0.43 (corresponding to the phase of maximum radius and minimum temperature) during three pulsation cycles in late 2018 and detected statistically significant emission (>5 σ ) during one of the three epochs. The observed radio emission appears to be variable at a ≳10% level on timescales of days to weeks. We also present an upper limit on the 10 GHz flux density at pulsation phase ϕ = 0.31 from an observation in 2014. We discuss possible mechanisms that may produce the observed 15 GHz emission, but cannot make a conclusive identification from the present data. The emission does not appear to be consistent with originating from a close-in, late-type dwarf companion, although this scenario cannot yet be strictly excluded. Previous X-ray observations have shown that δ Cep undergoes periodic increases in X-ray flux during pulsation phase ϕ ≈ 0.43. The lack of radio detection in two out of three observing epochs at ϕ ≈ 0.43 suggests that either the radio emission is not linked with a particular pulsation phase, or else that the strength of the generated radio emission in each pulsation cycle is variable.

Shingo Hirano, Masahiro N. Machida, Shantanu Basu

Intermediate-mass black holes (with ≥10 ^5 M _⊙ ) are promising candidates for the origin of supermassive black holes (with ∼10 ^9 M _⊙ ) in the early universe (redshift z ∼ 6). Chon & Omukai first pointed out direct collapse black hole (DCBH) formation in metal-enriched atomic-cooling halos (ACHs), which relaxes the DCBH formation criterion. On the other hand, Hirano et al. showed that magnetic effects promote DCBH formation in metal-free ACHs. We perform a set of magnetohydrodynamical simulations to investigate star formation in magnetized ACHs with metallicities Z / Z _⊙ = 0, 10 ^−5 , and 10 ^−4 . Our simulations show that the mass accretion rate onto the protostars becomes lower in metal-enriched ACHs than in metal-free ACHs. However, many protostars form from gravitationally and thermally unstable metal-enriched gas clouds. Under such circumstances, the magnetic field rapidly increases as magnetic field lines wind up due to the spin of protostars. The region with the amplified magnetic field expands outwards due to the orbital motion of protostars and the rotation of the accreting gas. The amplified magnetic field extracts angular momentum from the accreting gas, promotes the coalescence of low-mass protostars, and increases the mass growth rate of the primary protostar. We conclude that the magnetic field amplification is always realized in metal-enriched ACHs regardless of the initial magnetic field strength, which affects the DCBH formation criterion. In addition, we find a qualitatively different trend from the previous unmagnetized simulations in that the mass growth rate is maximal for extremely metal-poor ACHs with Z / Z _⊙ = 10 ^−5 .

D. Alberton, V. Lattanzi, C. Endres et al.

Cyanides, ranging from three carbon atoms to polycyclic aromatic hydrocarbons, and alkenyl compounds are abundant in the interstellar medium. Aminoacrylonitrile (3-Amino-2-propenenitrile, H _2 N–CH=CH–C≡N), an alkenyl cyanide, thus represents a promising candidate for new interstellar detection. A comprehensive spectroscopic laboratory investigation of aminoacrylonitrile in its rotational ground vibrational state has been herein performed. The measurements carried out up to the THz regime made it possible to generate a precise set of reliable rest frequencies for its search in space up to submillimeter wavelengths. The Z -aminoacrylonitrile ( Z -apn) isomer spectrum has been recorded employing a source-modulated submillimeter spectrometer, from 80 GHz to 1 THz. A combination of Doppler and sub-Doppler measurement regimes allowed one to record 600 new lines. The collected data have enabled the characterization of a set of spectroscopic parameters up to decic centrifugal distortion constants. The catalog generated from the improved spectral data has been used for the search of Z -apn in the spectral survey of the G+0.693-0.027 molecular cloud located in the central molecular zone, in the proximity of the Galactic center.

Wen-Di Guo, Shao-Wen Wei, Yu-Xiao Liu

Abstract Seeking singularity free solutions are important for further understanding black holes in quantum level. Recently, a five-dimensional singularity free black hole/topological star was constructed (Bah and Heidmann in Phys Rev Lett 126:151101, 2021). Through the Kaluza–Klein reduction, an effective four-dimensional static spherically symmetric charged black hole with scalar hair can be obtained. In this paper, we study shadow of this charged black hole with scalar hair in terms of four kinds of observers, i.e., static observers, surrounding observers, freely falling observers, and escaping observers in four-dimensional spacetime. For a spherically symmetric black hole, the shadow is circular for any observer, but the shadow size depends on the motion status of the observer. On the other hand, the effect of plasma is also investigated by a simple model. The radius of the photon sphere depends on the plasma model. Most importantly, we find that the shadow sizes do not monotonically decrease with r in some cases.

Keiichi Namizaki, Kosuke Namekata, Hiroyuki Maehara et al.

Active M-type stars are known to often produce superflares on the surface. Radiation from stellar (super)flares is important for exoplanet habitability, but the mechanisms are not well understood. In this paper, we report simultaneous optical spectroscopic and photometric observations of a stellar superflare on an active M dwarf, YZ Canis Minoris, with the 3.8 m Seimei telescope and the Transiting Exoplanet Survey Satellite. The flare bolometric energy is ${1.3}_{-0.6}^{+1.6}\times {10}^{34}\,\mathrm{erg}$ and the H α energy is ${3.0}_{-0.1}^{+0.1}\times {10}^{32}\,\mathrm{erg}$ . The H α emission line profile shows red asymmetry throughout the flare, with a duration of 4.6–5.1 hr. The velocity of the red asymmetry is ∼200–500 km s ^–1 and the line width of H α broadens up to 34 ± 14 Å. The redshifted velocity and line width of H α line decay more rapidly than the equivalent width, and their time evolutions are correlated with that of the white-light emission. This indicates the possibility of the white light, the H α red asymmetry, and the H α line broadening originating from nearly the same site, i.e., the dense chromospheric condensation region, heated by nonthermal electrons. On the other hand, the flux ratio of the redshifted excess components to the central components is enhanced one hr after the flare’s onset. This may be due to the main source of the red asymmetry changing to post-flare loops in the later phase of the flare.

Guangqiang Teng, Boping Tian, Yuanyuan Zhang et al.

The optimal subsampling is an statistical methodology for generalized linear models (GLMs) to make inference quickly about parameter estimation in massive data regression. Existing literature only considers bounded covariates. In this paper, the asymptotic normality of the subsampling M-estimator based on the Fisher information matrix is obtained. Then, we study the asymptotic properties of subsampling estimators of unbounded GLMs with nonnatural links, including conditional asymptotic properties and unconditional asymptotic properties.

Anand N. Vidyashankar, Jeffrey F. Collamore

Hellinger distance has been widely used to derive objective functions that are alternatives to maximum likelihood methods. While the asymptotic distributions of these estimators have been well investigated, the probabilities of rare events induced by them are largely unknown. In this article, we analyze these rare event probabilities using large deviation theory under a potential model misspecification, in both one and higher dimensions. We show that these probabilities decay exponentially, characterizing their decay via a “rate function” which is expressed as a convex conjugate of a limiting cumulant generating function. In the analysis of the lower bound, in particular, certain geometric considerations arise that facilitate an explicit representation, also in the case when the limiting generating function is nondifferentiable. Our analysis involves the modulus of continuity properties of the affinity, which may be of independent interest.

Yutian Cao, Dandan Niu, Jun Cui et al.

The launch of Tianwen 1 marks the brand-new stage of the deep space exploration of China, which calls for the foundation of planetary science education in China. In planetary science, the ionosphere is considered to be the region of an atmosphere where a significant number of free thermal electrons and ions exist. These electrons and ions are produced via ionization of the neutral particles both by extreme ultraviolet radiation from the Sun and by collisions with energetic particles that penetrate the atmosphere. The planetary ionosphere, where the matters and energies outside the planet exchanges with the planet itself, is extremely important to the planetary upper atmosphere and the surface environment. In this study, we briefly review the studies of Venusian and Martian ionospheres based on the measurements starting from last century to provide the basis for comparative studies on planetary ionospheres.

Gonen Singer, Matan Marudi

In this research, we develop ordinal decision-tree-based ensemble approaches in which an objective-based information gain measure is used to select the classifying attributes. We demonstrate the applicability of the approaches using AdaBoost and random forest algorithms for the task of classifying the regional daily growth factor of the spread of an epidemic based on a variety of explanatory factors. In such an application, some of the potential classification errors could have critical consequences. The classification tool will enable the spread of the epidemic to be tracked and controlled by yielding insights regarding the relationship between local containment measures and the daily growth factor. In order to benefit maximally from a variety of ordinal and non-ordinal algorithms, we also propose an ensemble majority voting approach to combine different algorithms into one model, thereby leveraging the strengths of each algorithm. We perform experiments in which the task is to classify the daily COVID-19 growth rate factor based on environmental factors and containment measures for 19 regions of Italy. We demonstrate that the ordinal algorithms outperform their non-ordinal counterparts with improvements in the range of 6–25% for a variety of common performance indices. The majority voting approach that combines ordinal and non-ordinal models yields a further improvement of between 3% and 10%.

Zhipeng Lin, Yuhua Tang, Yongjun Zhang

The Recommender System (RS) has obtained a pivotal role in e-commerce. To improve the performance of RS, review text information has been extensively utilized. However, it is still a challenge for RS to extract the most informative feature from a tremendous amount of reviews. Another significant issue is the modeling of user&#8315;item interaction, which is rarely considered to capture high- and low-order interactions simultaneously. In this paper, we design a multi-level attention mechanism to learn the usefulness of reviews and the significance of words by Deep Neural Networks (DNN). In addition, we develop a hybrid prediction structure that integrates Factorization Machine (FM) and DNN to model low-order user&#8315;item interactions as in FM and capture the high-order interactions as in DNN. Based on these two designs, we build a Multi-level Attentional and Hybrid-prediction-based Recommender (MAHR) model for recommendation. Extensive experiments on Amazon and Yelp datasets showed that our approach provides more accurate recommendations than the state-of-the-art recommendation approaches. Furthermore, the verification experiments and explainability study, including the visualization of attention modules and the review-usefulness prediction test, also validated the reasonability of our multi-level attention mechanism and hybrid prediction.