arXiv subjects

Xilu Wang

Publications and source records attributed to Xilu Wang.

At least 37 records · Page 2Linked to original sources

arXiv2026

Proton-rich production of lanthanides: the $νi$ process

The astrophysical origin of the lanthanides is an open question in nuclear astrophysics. Besides the widely studied $s$, $i$, and $r$ processes in moderately-to-strongly neutron-rich environments, an intriguing alternative site for lanthanide production could in fact be robustly $\textit{proton-rich}$ matter outflows from core-collapse supernovae under specific conditions -- in particular, high-entropy winds with enhanced neutrino luminosity and fast dynamical timescales. In this environment, excess protons present after charged particle reactions have ceased can continue to be converted to neutrons by (anti-)neutrino interactions, producing a neutron capture reaction flow up to A~200. This scenario, christened the $νi$ process in a recent paper, has previously been discussed as a possibility. Here, we examine the prospects for $νi$ process through the lens of stellar abundance patterns, bolometric lightcurves, and galactic chemical evolution models, with a particular focus on hypernovae as candidate sites. We identify specific lanthanide signatures for which the $νi$ process can provide a credible alternative to $r$/$i$ processes.

astro-ph.HE↗

arXiv2025

Observatory Science with eXTP

Scheduled for launch in 2030, the enhanced X-ray Timing and Polarization (eXTP) telescope is a Chinese space-based mission aimed at studying extreme conditions and phenomena in astrophysics. eXTP will feature three main payloads: Spectroscopy Focusing Arrays (SFAs), Polarimetry Focusing Arrays (PFAs), and a Wide-field Camera (W2C). This white paper outlines observatory science, incorporating key scientific advances and instrumental changes since the publication of the previous white paper [1]. We will discuss perspectives of eXTP on the research domains of flare stars, supernova remnants, pulsar wind nebulae, cataclysmic variables, X-ray binaries, ultraluminous X-ray sources, AGN, and pulsar-based positioning and timekeeping.

astro-ph.IM↗

arXiv2025

OWLed: Outlier-weighed Layerwise Pruning for Efficient Autonomous Driving Framework

The integration of Large Language Models (LLMs) into autonomous driving systems offers promising enhancements in environmental understanding and decision-making. However, the substantial computational demands of deploying LLMs locally on vehicles render this approach unfeasible for real-world automotive applications. To address this challenge, we introduce OWLed, the Outlier-Weighed Layerwise Pruning for Efficient Autonomous Driving Framework that leverages outlier-weighted layerwise sparsity for model compression. Our method assigns non-uniform sparsity ratios to different layers based on the distribution of outlier features, significantly reducing the model size without the need for fine-tuning. To ensure the compressed model adapts well to autonomous driving tasks, we incorporate driving environment data into both the calibration and pruning processes. Our empirical studies reveal that the encoder component is more sensitive to pruning than the LLM, highlighting its critical role in the system. Experimental results demonstrate that OWLed outperforms existing methods in perception, action prediction, and language understanding while substantially lowering computational requirements. These findings underscore the potential of combining advanced pruning techniques with LLMs to develop efficient and robust autonomous driving systems capable of handling complex scenarios. Code will be made publicly available.

cs.LG↗

arXiv2025

LOST: Low-rank and Sparse Pre-training for Large Language Models

While large language models (LLMs) have achieved remarkable performance across a wide range of tasks, their massive scale incurs prohibitive computational and memory costs for pre-training from scratch. Recent studies have investigated the use of low-rank parameterization as a means of reducing model size and training cost. In this context, sparsity is often employed as a complementary technique to recover important information lost in low-rank compression by capturing salient features in the residual space. However, existing approaches typically combine low-rank and sparse components in a simplistic or ad hoc manner, often resulting in undesirable performance degradation compared to full-rank training. In this paper, we propose \textbf{LO}w-rank and \textbf{S}parse pre-\textbf{T}raining (\textbf{LOST}) for LLMs, a novel method that ingeniously integrates low-rank and sparse structures to enable effective training of LLMs from scratch under strict efficiency constraints. LOST applies singular value decomposition to weight matrices, preserving the dominant low-rank components, while allocating the remaining singular values to construct channel-wise sparse components to complement the expressiveness of low-rank training. We evaluate LOST on LLM pretraining ranging from 60M to 7B parameters. Our experiments show that LOST achieves competitive or superior performance compared to full-rank models, while significantly reducing both memory and compute overhead. Moreover, Code is available at \href{https://github.com/JiaxiLi1/LOST-Low-rank-and-Sparse-Training-for-Large-Language-Models}{LOST Repo}

cs.LG↗

arXiv2025

CLIP Brings Better Features to Visual Aesthetics Learners

Image Aesthetics Assessment (IAA) is a challenging task due to its subjective nature and expensive manual annotations. Recent large-scale vision-language models, such as Contrastive Language-Image Pre-training (CLIP), have shown their promising representation capability for various downstream tasks. However, the application of CLIP to resource-constrained and low-data IAA tasks remains limited. While few attempts to leverage CLIP in IAA have mainly focused on carefully designed prompts, we extend beyond this by allowing models from different domains and with different model sizes to acquire knowledge from CLIP. To achieve this, we propose a unified and flexible two-phase CLIP-based Semi-supervised Knowledge Distillation (CSKD) paradigm, aiming to learn a lightweight IAA model while leveraging CLIP's strong generalization capability. Specifically, CSKD employs a feature alignment strategy to facilitate the distillation of heterogeneous CLIP teacher and IAA student models, effectively transferring valuable features from pre-trained visual representations to two lightweight IAA models, respectively. To efficiently adapt to downstream IAA tasks in a low-data regime, the two strong visual aesthetics learners then conduct distillation with unlabeled examples for refining and transferring the task-specific knowledge collaboratively. Extensive experiments demonstrate that the proposed CSKD achieves state-of-the-art performance on multiple widely used IAA benchmarks. Furthermore, analysis of attention distance and entropy before and after feature alignment shows the effective transfer of CLIP's feature representation to IAA models, which not only provides valuable guidance for the model initialization of IAA but also enhances the aesthetic feature representation of IAA models. Code will be made publicly available.

cs.CV↗

arXiv2025

Gamma rays as a signature of r-process producing supernovae: remnants and future Galactic explosions

We consider the question of whether core-collapse supernovae (CCSNe) can produce rapid neutron capture process (r-process) elements and how future MeV gamma-ray observations could address this. Rare types of CCSNe characterized by substantial magnetic fields and rotation, known as magnetorotational supernovae (MR-SNe), are theoretically predicted to produce these elements, although direct observational evidence is lacking. We suggest that this critical question be addressed through the study of some of the eleven CCSN remnants located within 10 kpc, as well as through the detection of gamma-ray emission from a future Galactic supernova. We use a two-dimensional MR-SN model to estimate the expected gamma flux stemming from nuclear decays in the range of a few tens of keV to a few MeV. Our results indicate that an observation of Sn-126 (Sb-126) in a remnant stands out as a signature of an r-process-producing supernova. Since the neutron-rich conditions that lead to the production of the r-process could also enhance the production of Fe-60, the detection of substantial Fe-60 (Co-60) would be indicative of favorable conditions for the r-process. In the case of a future supernova explosion, when the evolution of the spectrum is studied over ten days to a few years, a rich picture emerges. At various epochs, second peak r-process isotopes such as Sb-125, I-131, Te-132, I-132 and La-140 produce gamma-ray signals that emerge above the background from explosive burning products and electron-positron annihilation. The weak r-process isotopes Nb-95, Ru-103, Rh-106 also have periods of prominence. While MR-SNe are predicted to have a relatively small main r-process contribution, third peak isotopes like Ir-194 could still be above next-generation MeV gamma instrument sensitivities.

astro-ph.HE↗

arXiv2024

Implication of Jet Physics from MeV Line Emission of GRB 221009A

Ultrarelativistic jets are believed to play an important role in producing prompt emission and afterglow of gamma-ray bursts (GRBs), but the nature of the jet is poorly known owing to the lack of decisive features observed in the prompt emission. The discovery of an emission line evolving from about 37 to 6 MeV in the brightest-of-all-time GRB 221009A provides an unprecedented opportunity to probe GRB jet physics. The time evolution of the central energy of the line with power-law index $-1$ is naturally explained by the high-latitude curvature effect. Under the assumption that the line emission is generated in the prompt emission by $e^\pm$ pair production, cooling, and annihilation in the jet, we can strictly constrain jet physics with observed line emission properties. We find that the radius of the emission region is $r\gtrsim10^{16}$ cm. The narrow line width of $\sim10\%$ requires that the line emission occurs within $\sim10\%$ of the dynamical time, which further implies short timescales of pair cooling to the nonrelativistic state and pair annihilation, as well as a slightly clumpy emission region. If the jet's Lorentz factor is $Γ\gtrsim400$, the fast cooling requirement needs an energy density of magnetic field in the jet much larger than that of prompt gamma rays, i.e., a magnetically dominated jet. The temporal behavior of line flux suggests some angle dependence of line emission. We also discuss the difficulties of other scenarios for the observed emission line.

astro-ph.HE↗

arXiv2024

Collective neutrino oscillations and heavy-element nucleosynthesis in supernovae: exploring potential effects of many-body neutrino correlations

In high-energy astrophysical processes involving compact objects, such as core-collapse supernovae or binary neutron star mergers, neutrinos play an important role in the synthesis of nuclides. Neutrinos in these environments can experience collective flavor oscillations driven by neutrino-neutrino interactions, including coherent forward scattering and incoherent (collisional) effects. Recently, there has been interest in exploring potential novel behaviors in collective oscillations of neutrinos by going beyond the one-particle effective or "mean-field" treatments. Here, we seek to explore implications of collective neutrino oscillations, in the mean-field treatment and beyond, for the nucleosynthesis yields in supernova environments with different astrophysical conditions and neutrino inputs. We find that collective oscillations can impact the operation of the $νp$-process and $r$-process nucleosynthesis in supernovae. The potential impact is particularly strong in high-entropy, proton-rich conditions, where we find that neutrino interactions can nudge an initial $νp$ process neutron rich, resulting in a unique combination of proton-rich low-mass nuclei as well as neutron-rich high-mass nuclei. We describe this neutrino-induced neutron capture process as the "$ν$i process". In addition, nontrivial quantum correlations among neutrinos, if present significantly, could lead to different nuclide yields compared to the corresponding mean-field oscillation treatments, by virtue of modifying the evolution of the relevant one-body neutrino observables.

astro-ph.HE↗

arXiv2023

MeV Astrophysical Spectroscopic Surveyor (MASS): A Compton Telescope Mission Concept

We propose a future mission concept, the MeV Astrophysical Spectroscopic Surveyor (MASS), which is a large area Compton telescope using 3D position sensitive cadmium zinc telluride (CZT) detectors optimized for emission line detection. The payload consists of two layers of CZT detectors in a misaligned chessboard layout, with a total geometric area of 4096 cm$^2$ for on-axis observations. The detectors can be operated at room-temperature with an energy resolution of 0.6\% at 0.662 MeV. The in-orbit background is estimated with a mass model. At energies around 1 MeV, a line sensitivity of about $10^{-5}$ photons cm$^{-2}$ s$^{-1}$ can be obtained with a 1 Ms observation. The main science objectives of MASS include nucleosynthesis in astrophysics and high energy astrophysics related to compact objects and transient sources. The payload CZT detectors weigh roughly 40 kg, suggesting that it can be integrated into a micro- or mini-satellite. We have constructed a pathfinder, named as MASS-Cube, to have a direct test of the technique with 4 detector units in space in the near future.

astro-ph.IM↗

arXiv2023

Thallium-208: a beacon of in situ neutron capture nucleosynthesis

We demonstrate that the well-known 2.6 MeV gamma-ray emission line from thallium-208 could serve as a real-time indicator of astrophysical heavy element production, with both rapid (r) and intermediate (i) neutron capture processes capable of its synthesis. We consider the r process in a Galactic neutron star merger and show Tl-208 to be detectable from ~12 hours to ~10 days, and again ~1-20 years post-event. Detection of Tl-208 represents the only identified prospect for a direct signal of lead production (implying gold synthesis), arguing for the importance of future MeV telescope missions which aim to detect Galactic events but may also be able to reach some nearby galaxies in the Local Group.

nucl-th↗

arXiv2023

Could a Kilonova Kill: a Threat Assessment

Binary neutron star mergers (BNS) produce high-energy emissions from several physically different sources, including a gamma-ray burst (GRB) and its afterglow, a kilonova, and, at late times, a remnant many parsecs in size. Ionizing radiation from these sources can be dangerous for life on Earth-like planets when located too close. Work to date has explored the substantial danger posed by the GRB to on-axis observers: here we focus instead on the potential threats posed to nearby off-axis observers. Our analysis is based largely on observations of the GW 170817/GRB 170817A multi-messenger event, as well as theoretical predictions. For baseline kilonova parameters, we find that the X-ray emission from the afterglow may be lethal out to $\sim 5$ pc and the off-axis gamma-ray emission may threaten a range out to $\sim 4$ pc, whereas the greatest threat comes years after the explosion, from the cosmic rays accelerated by the kilonova blast, which can be lethal out to distances up to $\sim 11$ pc. The distances quoted here are typical, but the values have significant uncertainties and depend on the viewing angle, ejected mass, and explosion energy in ways we quantify. Assessing the overall threat to Earth-like planets, have a similar kill distance to supernovae, but are far less common. However, our results rely on the scant available kilonova data, and multi-messenger observations will clarify the danger posed by such events.

astro-ph.HE↗

arXiv2023

Neutrinos and Heavy Element Nucleosynthesis

This chapter discusses three nucleosynthesis processes involved in producing heavy nuclei beyond the iron group that are influenced or shaped by neutrino interactions: the v process, the vp process and the r process. These processes are all related to explosive events involving compact objects, such as core-collapse supernovae and binary neutron star mergers, where an abundant amount of neutrinos are emitted. The interactions of the neutrinos with nucleons and nuclei through both charged-current and neutral-current reactions play a crucial role in the nucleosynthesis processes. During the propagation of neutrinos inside the nucleosynthesis sites, neutrinos may undergo flavor oscillations that can also potentially affect the nucleosynthesis yields. Here we provide a general overview of the possible effects of neutrinos and neutrino flavor conversions on these three heavy-element nucleosynthesis processes.

astro-ph.HE↗

arXiv2023

Flares from merged magnetars: their prospects as a new population of gamma-ray counterparts of binary neutron star mergers

Long-lived massive magnetars are expected to be remnants of some binary neutron star (BNS) mergers. In this paper, we argue that the magnetic powered flaring activities of these merged magnetars would occur dominantly in their early millisecond-period-spin phase, which is in the timescale of days. Such flares endure significant absorption by the ejecta from the BNS collision, and their detectable energy range is from 0.1-10 MeV, in a time-lag of $\sim$ days after the merger events indicated by the gravitational wave chirps. We estimate the rate of such flares in different energy ranges, and find that there could have been ~0.1-10 cases detected by Fermi/GBM. A careful search for $\sim10$ milliseconds spin period modulation in weak short gamma-ray bursts (GRBs) may identify them from the archival data. The next generation MeV detectors could detect them at a mildly higher rate. The recent report on the Quasi-Period-Oscillation found in two BASTE GRBs should not be considered as cases of such flares, for they were detected in a lower energy range and with a much shorter period spin modulation.

astro-ph.HE↗

arXiv2023

Proposed Lunar Measurements of $r$-Process Radioisotopes to Distinguish Origin of Deep-sea 244Pu

244Pu has recently been discovered in deep-sea deposits spanning the past 10 Myr, a period that includes two 60Fe pulses from nearby supernovae. 244Pu is among the heaviest $r$-process products, and we consider whether it was created in the supernovae, which is disfavored by nucleosynthesis simulations, or in an earlier kilonova event that seeded 244Pu in the nearby interstellar medium that was subsequently swept up by the supernova debris. We discuss how these possibilities can be probed by measuring 244Pu and other $r$-process radioisotopes such as 129I and 182Hf, both in lunar regolith samples returned to Earth by missions such as Chang'e and Artemis, and in deep-sea deposits.

astro-ph.HE↗

arXiv2022

Recent Advances in Bayesian Optimization

Bayesian optimization has emerged at the forefront of expensive black-box optimization due to its data efficiency. Recent years have witnessed a proliferation of studies on the development of new Bayesian optimization algorithms and their applications. Hence, this paper attempts to provide a comprehensive and updated survey of recent advances in Bayesian optimization and identify interesting open problems. We categorize the existing work on Bayesian optimization into nine main groups according to the motivations and focus of the proposed algorithms. For each category, we present the main advances with respect to the construction of surrogate models and adaptation of the acquisition functions. Finally, we discuss the open questions and suggest promising future research directions, in particular with regard to heterogeneity, privacy preservation, and fairness in distributed and federated optimization systems.

cs.LG↗

arXiv2022

Near-Earth Supernovae in the Past 10 Myr: Implications for the Heliosphere

We summarize evidence that multiple supernovae exploded within 100 pc of Earth in the past few Myr. These events had dramatic effects on the heliosphere, compressing it to within ~20 au. We advocate for cross-disciplinary research of nearby supernovae, including on interstellar dust and cosmic rays. We urge for support of theory work, direct exploration, and study of extrasolar astrospheres.

astro-ph.IM↗

arXiv2022

Alleviating Search Bias in Bayesian Evolutionary Optimization with Many Heterogeneous Objectives

Multi-objective optimization problems whose objectives have different evaluation costs are commonly seen in the real world. Such problems are now known as multi-objective optimization problems with heterogeneous objectives (HE-MOPs). So far, however, only a few studies have been reported to address HE-MOPs, and most of them focus on bi-objective problems with one fast objective and one slow objective. In this work, we aim to deal with HE-MOPs having more than two black-box and heterogeneous objectives. To this end, we develop a multi-objective Bayesian evolutionary optimization approach to HE-MOPs by exploiting the different data sets on the cheap and expensive objectives in HE-MOPs to alleviate the search bias caused by the heterogeneous evaluation costs for evaluating different objectives. To make the best use of two different training data sets, one with solutions evaluated on all objectives and the other with those only evaluated on the fast objectives, two separate Gaussian process models are constructed. In addition, a new acquisition function that mitigates search bias towards the fast objectives is suggested, thereby achieving a balance between convergence and diversity. We demonstrate the effectiveness of the proposed algorithm by testing it on widely used multi-/many-objective benchmark problems whose objectives are assumed to be heterogeneously expensive.

cs.NE↗

arXiv2021

r-Process Radioisotopes from Near-Earth Supernovae and Kilonovae

The astrophysical sites where r-process elements are synthesized remain mysterious: it is clear that neutron star mergers (kilonovae (KNe)) contribute, and some classes of core-collapse supernovae (SNe) are also likely sources of at least the lighter r-process species. The discovery of 60Fe on the Earth and Moon implies that one or more astrophysical explosions have occurred near the Earth within the last few million years, probably SNe. Intriguingly, 244Pu has now been detected, mostly overlapping with 60Fe pulse. However, the 244Pu flux may extend to before 12 Myr ago, pointing to a different origin. Motivated by these observations and difficulties for r-process nucleosynthesis in SN models, we propose that ejecta from a KN enriched the giant molecular cloud that gave rise to the Local Bubble, where the Sun resides. Accelerator mass spectrometry (AMS) measurements of 244Pu and searches for other live isotopes could probe the origins of the r-process and the history of the solar neighborhood, including triggers for mass extinctions, e.g., that at the end of the Devonian epoch, motivating the calculations of the abundances of live r-process radioisotopes produced in SNe and KNe that we present here. Given the presence of 244Pu, other r-process species such as 93Zr, 107Pd, 129I, 135Cs, 182Hf, 236U, 237Np and 247Cm should be present. Their abundances and well-resolved time histories could distinguish between the SN and KN scenarios, and we discuss prospects for their detection in deep-ocean deposits and the lunar regolith. We show that AMS 129I measurements in Fe-Mn crusts already constrain a possible nearby KN scenario.

astro-ph.HE↗