arXiv subjects

Junhao Liu

Publications and source records attributed to Junhao Liu.

At least 37 records · Page 2Linked to original sources

arXiv2026

WASD: Locating Critical Neurons as Sufficient Conditions for Explaining and Controlling LLM Behavior

Precise behavioral control of large language models (LLMs) is critical for complex applications. However, existing methods often incur high training costs, lack natural language controllability, or compromise semantic coherence. To bridge this gap, we propose WASD (unWeaving Actionable Sufficient Directives), a novel framework that explains model behavior by identifying sufficient neural conditions for token generation. Our method represents candidate conditions as neuron-activation predicates and iteratively searches for a minimal set that guarantees the current output under input perturbations. Experiments on SST-2 and CounterFact with the Gemma-2-2B model demonstrate that our approach produces explanations that are more stable, accurate, and concise than conventional attribution graphs. Moreover, through a case study on controlling cross-lingual output generation, we validated the practical effectiveness of WASD in controlling model behavior.

cs.CL↗

arXiv2026

Mott-Derived Local Moments and Kondo Hybridization in a d-electron Kagome lattice

Unlike canonical Kondo lattices in f-electron systems, where localized f orbitalsnaturally provide local moments, d-electron Kondo lattices require a distinct mechanism for local-moment formation. However, the study of d-electron Kondo lattices in bulk materials remains far from settled, particularly with regard to the microscopic origin of the local moments. Here, we report a microscopic mechanism for this process in the bilayer kagome metal CsCr6Sb6, where strong correlations drive a Mott splitting of the kagome flat band to supply the requisite local moments. By combining STM/STS and ARPES, we resolve a spectroscopic hierarchy between high-energy correlation effects and low temperature hybridization. Low-temperature STS reveals a robust asymmetric suppression of the density of states near EF that is well captured phenomenologically by a Fano-type lineshape, while ARPES detects a sharp quasiparticlepeak near EF. These low-energy signatures evolveon the same temperature scale and disappear upon warming, consistent with the onset of Kondo hybridization. At the same time, STS resolves symmetric humps at approximately +-50 mV and ARPES identifies a weakly dispersive feature around 50 meV below EF; unlike the near-EF hybridization signatures, these features persist to substantially higher temperatures. This separation of energy and temperature scales supports a two-stage picture in which a kagome flat band first undergoes correlation-driven splitting into lower and upper Hubbard bands, and the occupied lower Hubbard band supplies the local moments that later hybridize with itinerant electrons at lower temperature. Our results therefore move beyond the phenomenology of a kagome Kondo lattice candidate and instead provide a microscopic spectroscopic picture linking Mottness to Kondo hybridization in a frustrated d-electron system.

cond-mat.str-el↗

arXiv2026

Beyond Attribution: Unified Concept-Level Explanations

There is an increasing need to integrate model-agnostic explanation techniques with concept-based approaches, as the former can explain models across different architectures while the latter makes explanations more faithful and understandable to end-users. However, existing concept-based model-agnostic explanation methods are limited in scope, mainly focusing on attribution-based explanations while neglecting diverse forms like sufficient conditions and counterfactuals, thus narrowing their utility. To bridge this gap, we propose a general framework UnCLE to elevate existing local model-agnostic techniques to provide concept-based explanations. Our key insight is that we can uniformly extend existing local model-agnostic methods to provide unified concept-based explanations with large pre-trained model perturbation. We have instantiated UnCLE to provide concept-based explanations in three forms: attributions, sufficient conditions, and counterfactuals, and applied it to popular text, image, and multimodal models. Our evaluation results demonstrate that UnCLE provides explanations more faithful than state-of-the-art concept-based explanation methods, and provides richer explanation forms that satisfy various user needs.

cs.LG↗

arXiv2026

SC-Arena: A Natural Language Benchmark for Single-Cell Reasoning with Knowledge-Augmented Evaluation

Large language models (LLMs) are increasingly applied in scientific research, offering new capabilities for knowledge discovery and reasoning. In single-cell biology, however, evaluation practices for both general and specialized LLMs remain inadequate: existing benchmarks are fragmented across tasks, adopt formats such as multiple-choice classification that diverge from real-world usage, and rely on metrics lacking interpretability and biological grounding. We present SC-ARENA, a natural language evaluation framework tailored to single-cell foundation models. SC-ARENA formalizes a virtual cell abstraction that unifies evaluation targets by representing both intrinsic attributes and gene-level interactions. Within this paradigm, we define five natural language tasks (cell type annotation, captioning, generation, perturbation prediction, and scientific QA) that probe core reasoning capabilities in cellular biology. To overcome the limitations of brittle string-matching metrics, we introduce knowledge-augmented evaluation, which incorporates external ontologies, marker databases, and scientific literature to support biologically faithful and interpretable judgments. Experiments and analysis across both general-purpose and domain-specialized LLMs demonstrate that (i) under the Virtual Cell unified evaluation paradigm, current models achieve uneven performance on biologically complex tasks, particularly those demanding mechanistic or causal understanding; and (ii) our knowledge-augmented evaluation framework ensures biological correctness, provides interpretable, evidence-grounded rationales, and achieves high discriminative capacity, overcoming the brittleness and opacity of conventional metrics. SC-Arena thus provides a unified and interpretable framework for assessing LLMs in single-cell biology, pointing toward the development of biology-aligned, generalizable foundation models.

cs.AI↗

arXiv2026

Digging into the Interior of Hot Cores with ALMA (DIHCA). VI. The Formation of Low-mass Multiple Systems in High-mass Cluster-forming Regions

Most stars form in multiple systems, with profound implications in numerous astronomical phenomena intrinsically linked to multiplicity. However, our knowledge about the process on how multiple stellar systems form is incomplete and biased toward nearby molecular clouds forming only low-mass stars, which are unrepresentative of the stellar population in the Galaxy. Most stars form within dense cores in clusters alongside high-mass stars (>8 M$_{\odot}$), as likely the Sun did. Here we report deep ALMA 1.33 mm dust continuum observations at ~160 au spatial resolution, revealing 72 low-mass multiple systems embedded in 23 high-mass cluster-forming regions, as part of the Digging into the Interior of Hot Cores with ALMA (DIHCA) survey. We find that the companion separation distribution presents a distinct peak at ~1200 au, in contrast to the one at ~4000 au observed in nearby low-mass regions. The shorter fragmentation scale can be explained by considering the higher pressure exerted by the surrounding medium, which is higher than the one in low-mass regions, due to the larger turbulence and densities involved. Because the peak of the companion separation distribution occurs at much larger scales than the expected disk sizes, we argue that the observed fragmentation is produced by turbulent core fragmentation. Contrary as predicted, the multiplicity fraction remains constant as the stellar density increases. We propose that in the extremely dense environments where high-mass stars form, dynamical interactions play an important role in disrupting weakly bound systems.

astro-ph.GA↗

arXiv2026

Focus-LIME: Surgical Interpretation of Long-Context Large Language Models via Proxy-Based Neighborhood Selection

As Large Language Models (LLMs) scale to handle massive context windows, achieving surgical feature-level interpretation is essential for high-stakes tasks like legal auditing and code debugging. However, existing local model-agnostic explanation methods face a critical dilemma in these scenarios: feature-based methods suffer from attribution dilution due to high feature dimensionality, thus failing to provide faithful explanations. In this paper, we propose Focus-LIME, a coarse-to-fine framework designed to restore the tractability of surgical interpretation. Focus-LIME utilizes a proxy model to curate the perturbation neighborhood, allowing the target model to perform fine-grained attribution exclusively within the optimized context. Empirical evaluations on long-context benchmarks demonstrate that our method makes surgical explanations practicable and provides faithful explanations to users.

cs.CL↗

arXiv2026

Magnetic threads and gravity: ALMA Observations of IRDC G14.225-0.506

During the star formation process, the interplay between gravity, turbulence, and B-fields is significant, with B-fields apparently serving a regulatory function. However, the extent to which B-fields are decisive relative to turbulence and gravity remains uncertain. This study aims to ascertain the role of B-fields in the fragmentation of molecular clouds. We examine the B-field observed with ALMA at core scales towards the infrared dark cloud G14.225-0.506, focusing on 3 regions with shared physical conditions, and juxtapose it with prior observations at the Hub-filament system scale. Our findings indicate a similar B-field strength and fragmentation level between the 2 hubs. However, distinct B-field morphologies are identified across the 3 regions where polarized emission is detected. In the region N, the large-scale B-field, which is perpendicular to the filamentary structure, persists at smaller scales in the southern half but becomes distorted near the more massive condensations in the northern half. Notably, these condensations exhibit signs of impending collapse, as evidenced by supercritical mass-to-flux values. In the region S, the B-field is considerably inhomogeneous among the detected condensations, and we do not observe a direct correlation between the field morphology and the condensation density. Lastly, in an isolated dust clump located within a southern filament of the northern hub, the B-field aligns parallel to the elongated emission, suggesting a transition in the field geometry. The B-field shows a clear evolution with spatial scales. We propose that the most massive condensations detected in the northern Hub are undergoing gravitational collapse, as revealed by the relative significance of the magnetic field and gravitational potential and mass-to-flux ratio. The distortion of the B-field could be a response to the flow of material due to the collapse.

astro-ph.GA↗

arXiv2025

Deep Andromeda JCMT-SCUBA2 Observations. The Submillimeter Maps and Giant Molecular Clouds

We have carried out unprecedentedly deep, nearly confusion-limited JCMT-SCUBA2 mapping observations on the nearest spiral galaxy, M31 (Andromeda). The 850 $μ$m image with a $\sim$50 pc resolution yields a comprehensive catalog of 383 giant molecular clouds (GMCs) that are associated with the spiral arms. In addition, it unveiled a population of 189 compact inter-arm GMCs in M31, which are mostly unresolved or marginally resolved. The masses of all these GMCs are in the range of 2$\times$10$^4$ -- 6$\times$10$^6$ $M_{\odot}$; the sizes are in the range of 30--130 pc. They follow a mass-size correlation, $M$ $\propto$ $R_{c}$$^{2.5}$. The inter-arm GMCs are systematically less massive, more diffuse, colder, and have lower star-forming efficiency (SFE) than on-arm GMCs. Moreover, within individual spatially resolved on-arm and off-arm M31 GMCs, the SFE is considerably lower than the SFE in molecular clouds in main sequence and green valley galaxies. Follow-up investigations on M31 GMCs may provide clues for how star formation may be quenched in galactic environments. Finally, we reconstrained the dust opacity spectral index $β$ in the M31 galaxy by combining our new JCMT observations with archival Herschel and Planck data and found that the radial variation of $β$ may not be as large as was proposed by previous studies.

astro-ph.GA↗

arXiv2025

Lattice design of a storage-ring-based light source for generating high-power fully coherent EUV radiation

We present the physical design and systematic optimization of a high-performance storage ring tailored for the generation of high-power coherent radiation, with particular emphasis on the extreme ultraviolet (EUV) regime. The proposed ring adopts a Double Bend Achromat (DBA) lattice configuration and integrates 12 superconducting wigglers to significantly enhance radiation damping and minimize the natural emittance. And a bypass line is adopted to generate high power coherent radiation. Comprehensive linear and nonlinear beam dynamics analyses have been conducted to ensure beam stability and robustness across the operational parameter space. The optimized design achieves a natural emittance of approximately 0.8 nm and a longitudinal damping time of around 1.4 ms, enabling the efficient buildup of coherent radiation. Three-dimensional numerical simulations, incorporating the previously proposed angular dispersion-induced microbunching (ADM) mechanism, further confirm the system's capability to generate high-power EUV coherent radiation, with output powers reaching the order of several hundred watts. These results underscore the strong potential of the proposed design for applications in coherent photon science and EUV lithography.

physics.acc-ph↗

arXiv2025

Magnetic Fields in Massive Star-forming Regions (MagMaR). VI. Magnetic Field Dragging in the Filamentary High-mass Star-forming Region G35.20--0.74N due to Gravity

We investigate the magnetic field orientation and strength in the massive star-forming region G35.20-0.74N (G35), using polarized dust emission data obtained with the Atacama Large Millimeter/submillimeter Array (ALMA) as part of the Magnetic fields in Massive star-forming Regions (MagMaR) survey. The G35 region shows a filamentary structure (a length of $\sim$0.1 pc) with six bright cores located along the filament's long axis. Magnetic field strengths across the G35 region range from 0.2 to 4.4 mG with a mean value of 0.8 $\pm$ 0.4 mG. The mass-to-flux ratio ($λ$) varies from 0.1 to 6.0 the critical value. The highest values are found locally around cores, whereas the remains of the filament are subcritical. A H$^{13}$CO$^+$ (3--2) velocity gradient of 29 km s$^{-1}$ pc$^{-1}$ is evident along the filament's long axis, aligned with the magnetic field direction. At larger scales ($\sim$0.1 pc), the magnetic field lines appear roughly perpendicular to the filament's long axis, in contrast to the smaller-scale structure ($\sim$0.003 pc) traced by ALMA. The magnetic field lines could be dragged along the filament as a result of the gas motion induced by the gravitational potential of the filament. Six cores in the filament have similar spacings between 0.02--0.04 pc. The initial filament fragmentation could have produced a core spacing of 0.06 pc, following filament fragmentation theory, and the current core spacing is the result of cores comoving with the gas along the filament. This core migration could occur in a few 10$^4$ years, consistent with high-mass star formation time scales.

astro-ph.GA↗

arXiv2025

Attosecond Waveform Synthesis through Echo-enabled Harmonic Generation Free-electron Lasers

Attosecond pulse trains (APTs) are indispensable for probing electron dynamics at their intrinsic timescales. High harmonic generation (HHG) has long been a successful and widely used technique in producing extreme ultraviolet APTs. While in the soft X-ray regime, HHG suffers from low conversion efficiency and lacking flexibility in the waveform and spectrum control. Here in this study, based on the waveform synthesis of the echo-enabled harmonic generation (EEHG) free-electron laser (FEL), we propose a novel method that can generate soft X-ray APTs with high temporal and spectral tunability. EEHG scheme is well-known in forming phase-locked high harmonic bunching combs down to the soft X-ray regime. And We can facilitate successive FEL lasing in several different harmonic numbers and then perform the coherent waveform synthesis to generate soft X-ray APTs. Three-dimensional simulation results indicate that reproducible generation of coherent APTs in the soft X-ray regime with peak power of 3.5 GW and micropulse duration of 160 as can be achieved by five-harmonic synthesis. And the micropulse duration together with the spectral components of the APTs can be adjusted easily according to the programmable high harmonic bunching combs in the EEHG scheme. This method has the potential to establish a robust platform for attosecond sciences in the soft X-ray regime, which can enable unprecedented studies of electron dynamics in physical and chemical reactions, biological systems and quantum materials.

physics.acc-ph↗

arXiv2025

Detection of a septuple stellar system in formation via disk fragmentation

Stellar multiple systems play a pivotal role in cluster dynamics and stellar evolution, leading to intense astronomical phenomena like X-ray binaries, gamma-ray bursts, Type Ia supernova, and stellar mergers, which are prime sources of gravitational waves. However, their origin remains poorly understood. Here we report the discovery of a septuple protostellar system embedded in a Keplerian disk within the high-mass star-forming region NGC\,6334IN, with close separations of 181-461 AU. The stability analysis reveals that the disk surrounding the septuple system is dynamically unstable, indicating that the septuple system formed via disk fragmentation. Previous studies have typically found only 2--3 members forming via disk fragmentation in both low- and high-mass star-forming regions. Our findings provide compelling observational evidence that the fragmentation of a gravitationally unstable disk is a viable mechanism for the formation of extreme high-order multiplicity, confirming what was previously only a theoretical concept. The results shed new light on the formation of extreme high-order multiplicity in cluster environments.

astro-ph.GA↗

arXiv2025

Impact of gravity on changing magnetic field orientations in a sample of massive protostellar clusters observed with ALMA

The magnetic field is integral to our understanding of the formation and dynamical evolution of molecular clouds and star formation within. We present a polarimetric survey of 17 massive protostellar cluster forming clumps, covered in 34 pointings in the 230-GHz window using the Atacama Large Millimeter/submillimeter Array (ALMA). The two array configurations, C43-1 and C43-4, probe linearly polarized dust emission, hence the plane-of-the-sky orientation of magnetic fields, at resolutions of 1\arcsec\ and 0\arcsec.4 that correspond to approximately 0.01pc core and $10^3$ au envelope scales, respectively. The relative orientations (ROs) of the magnetic field probed at two spatial scales are analyzed for the entire protostellar cluster sample and for a subset of objects in NGC 6334. We found a bimodal distribution of ROs with peaks at 0° (parallel) and 90°(orthogonal) for the entire sample combined as well as for NGC 6334. We investigate the physical origin of this bimodal distribution through a projected Rayleigh statistic (PRS) analysis in relation to column densities and local gravity in NGC 6334. We found an excess of parallel magnetic fields at column densities $> 10^{23}$ \cmm. The underlying cause of the RO distribution of the magnetic field is gravitational collapse at higher gas densities, which drags and reorients the magnetic field as shown in the alignment between the magnetic field and the direction of gravitational forces. The distribution of ROs observed here is consistent with the evolution of relative orientations of an initially sub-Alvénic cloud that becomes magnetically super-critical and super-Alvénic as the cloud collapses to form stars.

astro-ph.GA↗

arXiv2025

From Thinking to Output: Chain-of-Thought and Text Generation Characteristics in Reasoning Language Models

Recently, there have been notable advancements in large language models (LLMs), demonstrating their growing abilities in complex reasoning. However, existing research largely overlooks a thorough and systematic comparison of these models' reasoning processes and outputs, particularly regarding their self-reflection pattern (also termed "Aha moment") and the interconnections across diverse domains. This paper proposes a novel framework for analyzing the reasoning characteristics of four cutting-edge large reasoning models (GPT-o1, DeepSeek-R1, Kimi-k1.5, and Grok-3) using keywords statistic and LLM-as-a-judge paradigm. Our approach connects their internal thinking processes with their final outputs. A diverse dataset consists of real-world scenario-based questions covering logical deduction, causal inference, and multi-step problem-solving. Additionally, a set of metrics is put forward to assess both the coherence of reasoning and the accuracy of the outputs. The research results uncover various patterns of how these models balance exploration and exploitation, deal with problems, and reach conclusions during the reasoning process. Through quantitative and qualitative comparisons, disparities among these models are identified in aspects such as the depth of reasoning, the reliance on intermediate steps, and the degree of similarity between their thinking processes and output patterns and those of GPT-o1. This work offers valuable insights into the trade-off between computational efficiency and reasoning robustness and provides practical recommendations for enhancing model design and evaluation in practical applications. We publicly release our project at: https://github.com/ChangWenhan/FromThinking2Output

cs.CL↗

arXiv2025

The JCMT BISTRO-3 Survey: Variation of magnetic field orientations on parsec and sub-parsec scales in the massive star-forming region G28.34+0.06

Magnetic fields play a significant role in star-forming processes on core to clump scales. We investigate magnetic field orientations and strengths in the massive star-forming clump P2 within the filamentary infrared dark cloud G28.34+0.06 using dust polarization observations made using SCUBA-2/POL-2 on the James Clerk Maxwell Telescope as part of the B-field In STar-forming Region Observations (BISTRO) survey. We compare the magnetic field orientations at the clump scale of ~2 parsecs from these JCMT observations with those at the core scale of ~0.2 parsecs from archival ALMA data, finding that the magnetic field orientations on these two different scales are perpendicular to one another. We estimate the distribution of magnetic field strengths, which range from 50 to 430 μG over the clump. The region forming the core shows the highest magnetic field strength. We also obtain the distribution of mass-to-flux ratios across the clump. In the region surrounding the core, the mass-to-flux ratio is larger than 1, which indicates the magnetic field strength is insufficient to support the region against gravitational collapse. Therefore, the change in the magnetic field orientation from clump to core scales may be the result of gravitational collapse, with the field being pulled inward along with the flow of material under gravity.

astro-ph.GA↗

arXiv2025

First Lasing and Stable Operation of a Direct-Amplification Enabled Harmonic Generation Free-Electron laser

Seeded free-electron lasers (FELs) capable of operating at repetition rates up to the MHz level are in high demand for advanced time-resolved spectroscopies, which require both full longitudinal coherence and high average photon flux in the extreme ultraviolet (EUV) and x-ray regimes. However, conventional external-seed laser systems cannot sustain MHz operation with sufficient hundreds of megawatts peak power requirement due to their limited total power. Here, we report the first lasing and stable operation of a direct-amplification-enabled harmonic generation FEL driven by a weak seed laser with MW-level peak power. Beginning with an ultraviolet seed laser with only 0.75 μJ pulse energy, we demonstrate its direct amplification to over 10 μJ within an 8-meter-long modulator. We observe coherent harmonic generation up to the 12th harmonic of the seed and achieve saturation of the 7th harmonic in the radiator. These results represent a crucial milestone toward the realization of MHz-class, fully coherent EUV and x-ray light sources.

physics.acc-ph↗

arXiv2025

Characterizing Magnetic Properties of Young Protostars in Orion

The {\em B}-field Orion Protostellar Survey (BOPS) recently obtained polarimetric observations at 870 ${\rm μm}$ towards 61 protostars in the Orion molecular clouds with $\sim 1^{\prime\prime}$ spatial resolution using the Atacama Large Millimeter/submillimeter Array. From the BOPS sample, we selected the 26 protostars with extended polarized emission within a radius of $\sim 6^{\prime\prime}$ (2400~au) around the protostar. This allows to have sufficient statistical polarization data to infer the magnetic field strength. The magnetic field strength is derived using the Davis-Chandrasekhar-Fermi method. The underlying magnetic field strengths are approximately 2.0~mG for protostars with a standard hourglass magnetic field morphology, which is higher than the values derived for protostars with rotated hourglass, spiral, and complex magnetic field configurations ($\lesssim1.0$~mG). This suggests that the magnetic field plays a more significant role in envelopes exhibiting a standard hourglass field morphology, and a value of $\gtrsim2.0$ mG would be required to maintain such a structure at these scales. Furthermore, most protostars in the sample are slightly supercritical, with mass-to-flux ratios $\lesssim3.0$. In particular, the mass-to-flux ratios for all protostars with a standard hourglass magnetic field morphology are lower than 3.0. However, these ratios do not account for the contribution of the protostellar mass, which means they are likely significantly underestimated.

astro-ph.GA↗

arXiv2025

Dense Cores in IRDC G14.225-0.506 revealed by ALMA Observations

Dense cores in massive, parsec-scale molecular clumps are sites that harbor protocluster formation. We present results from observations towards a hub-filament structure of a massive Infrared Dark Cloud (IRDC) G14.225-0.506 using the Atacama Large Millimeter/submillimeter Array (ALMA). The dense cores are revealed by the 1.3 mm dust continuum emission at an angular resolution of $\sim$ 1.5'' and are identified through the hierarchical Dendrogram technique. Combining with the N$_2$D$^+$ 3-2 spectral line emission and gas temperatures derived from a previous NH$_3$ study, we analyze the thermodynamic properties of the dense cores. The results show transonic and supersonic-dominated turbulent motions. There is an inverse correlation between the virial parameter and the column density, which implies that denser regions may undergo stronger gravitational collapse. Molecular outflows are identified in the CO 2-1 and SiO 5-4 emission, indicating active protostellar activities in some cores. Besides these star formation signatures revealed by molecular outflows in the dense cores, previous studies in the infrared, X-ray, and radio wavelengths also found a rich and wide-spread population of young stellar objects (YSOs), showing active star formation both inside and outside of the dense cloud.

astro-ph.GA↗