arXiv subjects

Jiachen Jiang

Publications and source records attributed to Jiachen Jiang.

At least 19 recordsLinked to original sources

arXiv2026

Radio flares and X-ray hardening embedded in the long soft state of 4U 1543-475

We present a comprehensive multi-wavelength study of the black hole X-ray binary 4U 1543-475 during its 2021 outburst, focusing on radio flaring episodes that are commonly interpreted as signatures of episodic jet production and are embedded within states when the X-ray emission was dominated by an accretion disk component. The radio monitoring reveals at least two discrete flares that coincide with periods of enhanced Comptonized X-ray emission. Broadband spectral modelling shows a significant decrease in the reflection-to-disk flux ratio (by a factor of ~3-4) during these episodes, consistent with a temporary change in the geometry of the inner accretion flow, although the data do not allow the causal sequence to be firmly established. Optical photometry exhibits variability that broadly tracks the reflection fraction, consistent with changes in the illuminating component. The accompanying spectral hardening indicates that the radio flares were associated with short-lived excursions toward a "harder" state, departing from the soft state. X-ray timing analysis suggests that the radio flares may be associated with changes in the fractional rms variability; however, no consistent or unified pattern can be firmly established across different events. These results provide a multi-wavelength observational example of radio flaring activity in a black hole binary and highlight the complex interplay between accretion flow geometry, coronal emission, and jet-related phenomena.

astro-ph.HE↗

arXiv2026

Spin Constraints on 4U 1630-47 via combined Continuum Fitting and Reflection methods: a comparative study using Frequentist and Bayesian statistics

We present a comprehensive Bayesian spectral analysis of the black hole X-ray binary 4U 1630-47 during its 2022 outburst, using simultaneous NICER and NuSTAR observations. Using the traditional frequentist approach, we build our model combining reflection spectroscopy with continuum fitting techniques and analyze the data. In the Bayesian framework, we jointly constrain the black hole's spin, mass, and inclination within a unified framework. Employing nested sampling, we capture parameter degeneracies and rigorously propagate both statistical and systematic uncertainties. Our results yield robust evidence for a high spin, with the exact value still subject to model-dependent systematic uncertainties from both approaches. Our Bayesian analysis infers spin $a^* = 0.96_{-0.03}^{+0.02}$, mass $M_{\rm BH} = 12.19_{-1.46}^{+1.22} \, M_\odot$, and inclination angle $i = 55.75_{-1.37}^{+1.60}$ degrees, obtained via relxillNS flavor of our model. It also demonstrates the power of Bayesian inference in fetching valuable insights into the complex physics of black hole accretion and enabling high-confidence measurements of fundamental parameters.

astro-ph.HE↗

arXiv2026

Dense, multi-phase accretion disk atmosphere in the low-luminosity state of black hole transientV4641 Sgr

We present soft X-ray spectroscopy of the black-hole X-ray binary V4641~Sgr with the \textit{XMM-Newton} Reflection Grating Spectrometer (RGS). The RGS spectrum shows narrow emission features from N\,\textsc{vi--vii} and O\,\textsc{vii--viii} superimposed on a partially covered disk blackbody continuum. A blind Gaussian search confirms the presence of significant lines at the expected rest wavelengths. He-like triplet ratios (high $G$, low $R$) and full photoionization modelling both indicate a dense, photoionized plasma. Small redshifted velocities of $\sim 540$--$720\ \mathrm{km\ s^{-1}}$ are suggested, which are consistent with quasi-static or slowly flowing gas away from the observer after accounting for systematics. Photoionization modelling requires two \textsc{xstar} components with an intermediate ionization parameter ($\logξ\simeq 3.1$) and a low ionization parameter ($\logξ\simeq 0.36$), respectively. The simultaneous EPIC-pn spectrum suggests highly ionized Fe emission structures, hinting at an additional, more highly ionized component. These results imply the existence of a radially extended, multiphase, and dense disk atmosphere in the source. We compare the source with other X-ray binaries showing similar emission lines. V4641~Sgr shares a similarly high inclination with other sources; however, the presence of low ionization emission lines distinguishes it from the rest.

astro-ph.HE↗

arXiv2026

Assessing systematic uncertainties from spectral re-analysis of Cyg X-1 with different coronal geometries

In this work, we carry out a new spectral reanalysis of NuSTAR and Suzaku observations of the disk reflection spectra in the stellar-mass black hole X-ray binary Cyg~X-1. We compare three types of models: a broken power-law disk emissivity profile with no assumption about the coronal shape used in the previous work of the same observations, a compact lamppost corona, and an extended disk-like corona motivated by recent X-ray polarization results. Our goal is to measure the systematic uncertainties caused by the assumed geometry, with a focus on key parameters such as the black hole spin and the inclination of the inner accretion disk. We find that the disk-like corona gives a fit that is statistically similar to the broken power-law and lamppost models, but it leads to more physically reasonable results, such as a lower inclination angle of about $30^{\circ}$. By using a variable disk density model, we measure the disk density to be $n_{\rm e}\approx10^{20}$\,cm$^{-3}$, which is similar to earlier results. While the extended corona model infers a wider allowed parameter space for black hole spin and the inner radius of the disk-shaped coronal region, this reflects the additional physical freedom of the model. Even so, the disk-like corona remains a strong and physically well-motivated candidate for explaining the X-ray emission from Cyg~X-1.

astro-ph.HE↗

arXiv2026

SMCEvolve: Principled Scientific Discovery via Sequential Monte Carlo Evolution

LLM-driven program evolution has emerged as a powerful tool for automated scientific discovery, yet existing frameworks offer no principled guide for designing their individual components and provide no guarantee that the search converges. We introduce SMCEvolve, which recasts program search as sampling from a reward-tilted target distribution and approximates it with a Sequential Monte Carlo (SMC) sampler. From this view, three core mechanisms emerge as principled components: adaptive parent resampling, mixture of mutation with acceptance, and automatic convergence control. We further provide a finite-sample complexity analysis that bounds the LLM-call budget required to reach a target approximation error. Across math, algorithm efficiency, symbolic regression, and end-to-end ML research benchmarks, SMCEvolve surpasses state-of-the-art evolving systems while using fewer LLM calls under self-determined termination. The code is available at https://github.com/kongwanbianjinyu/SMCEvolve.

cs.AI↗

arXiv2026

Searching for Isolated Black Hole Candidates within 15 pc of the Solar System in Gaia DR3

Theoretical models predict that the Galaxy hosts $10^8$-$10^9$ black holes formed from the complete gravitational collapse of heavy stars and that most of these black holes are isolated, without any companion. Within 15 pc of the Solar System ($\sim 50$ ly), there may be a few black holes. If located inside one of the Local Interstellar Clouds - which occupy 5-20% of this local volume - an isolated black hole could produce detectable electromagnetic emission via accretion from the interstellar medium, given the capabilities of current or near-future observatories. However, precise predictions remain challenging due to large uncertainties in the expected accretion spectra. Outside these clouds, the accretion rate would be too low; according to our models, the resulting electromagnetic flux is well below the detection thresholds of current and near-future observational facilities. While astrometric detection via gravitational perturbation of nearby stars is conceivable, the local stellar density is too low for this method to be realistically successful. We have searched the Gaia DR3 catalog for candidate isolated black holes accreting from the interstellar medium and identified five sources. All candidates lie close to the Galactic plane, making them likely spurious astrometric solutions, for instance caused by unmodelled background sources (crowding) and/or unmodelled binarity. Our search for infrared and radio emission from these sources further suggests that they are unlikely to be black holes accreting from the interstellar medium.

astro-ph.HE↗

arXiv2026

On the Convergence of Gradient Descent on Learning Transformers with Residual Connections

Transformer models have emerged as fundamental tools across various scientific and engineering disciplines, owing to their outstanding performance in diverse applications. Despite this empirical success, the theoretical foundations of Transformers remain relatively underdeveloped, particularly in understanding their training dynamics. Existing research predominantly examines isolated components--such as self-attention mechanisms and feedforward networks--without thoroughly investigating the interdependencies between these components, especially when residual connections are present. In this paper, we aim to bridge this gap by analyzing the convergence behavior of a structurally complete yet single-layer Transformer, comprising self-attention, a feedforward network, and residual connections. We demonstrate that, under appropriate initialization, gradient descent exhibits a linear convergence rate, where the convergence speed is determined by the minimum and maximum singular values of the output matrix from the attention layer. Moreover, our analysis reveals that residual connections serve to ameliorate the ill-conditioning of this output matrix, an issue stemming from the low-rank structure imposed by the softmax operation, thereby promoting enhanced optimization stability. We also extend our theoretical findings to a multi-layer Transformer architecture, confirming the linear convergence rate of gradient descent under suitable initialization. Empirical results corroborate our theoretical insights, illustrating the beneficial role of residual connections in promoting convergence stability.

cs.LG↗

arXiv2026

Learning to Adapt: In-Context Learning Beyond Stationarity

Transformer models have become foundational across a wide range of scientific and engineering domains due to their strong empirical performance. A key capability underlying their success is in-context learning (ICL): when presented with a short prompt from an unseen task, transformers can perform per-token and next-token predictions without any parameter updates. Recent theoretical efforts have begun to uncover the mechanisms behind this phenomenon, particularly in supervised regression settings. However, these analyses predominantly assume stationary task distributions, which overlook a broad class of real-world scenarios where the target function varies over time. In this work, we bridge this gap by providing a theoretical analysis of ICL under non-stationary regression problems. We study how the gated linear attention (GLA) mechanism adapts to evolving input-output relationships and rigorously characterize its advantages over standard linear attention in this dynamic setting. To model non-stationarity, we adopt a first-order autoregressive process and show that GLA achieves lower training and testing errors by adaptively modulating the influence of past inputs -- effectively implementing a learnable recency bias. Our theoretical findings are further supported by empirical results, which validate the benefits of gating mechanisms in non-stationary ICL tasks.

cs.LG↗

arXiv2026

LLMs Struggle with Abstract Meaning Comprehension More Than Expected

Understanding abstract meanings is crucial for advanced language comprehension. Despite extensive research, abstract words remain challenging due to their non-concrete, high-level semantics. SemEval-2021 Task 4 (ReCAM) evaluates models' ability to interpret abstract concepts by presenting passages with questions and five abstract options in a cloze-style format. Key findings include: (1) Most large language models (LLMs), including GPT-4o, struggle with abstract meaning comprehension under zero-shot, one-shot, and few-shot settings, while fine-tuned models like BERT and RoBERTa perform better. (2) A proposed bidirectional attention classifier, inspired by human cognitive strategies, enhances fine-tuned models by dynamically attending to passages and options. This approach improves accuracy by 4.06 percent on Task 1 and 3.41 percent on Task 2, demonstrating its potential for abstract meaning comprehension.

cs.CL↗

arXiv2026

Extraction method for response functions from X-ray light curves of AGN by optimization algorithm

We introduce a numerical optimization method to extract the X-ray reverberation response functions from the multi-band light curves of the active galactic nuclei. This approach does not require prior assumptions about the accretion disc and corona geometry, provided that the light curves result from the superposition of direct and singly-convolved signals, consistently across all bands. By reformulating the light curve equations into the matrix form, the optimal response matrix is derived by minimizing the squared difference between the observed and reconstructed light curves using a gradient-based optimization algorithm. We demonstrate that the method can robustly accommodate up to two convolution processes, such as the reverberation and the propagation, simultaneously. When tested on the synthesized light curves, the method demonstrates robustness of the solutions to variations in the relative contributions of each light curve component as the recovered response kernel remains acceptably close to the ground truth, as evaluated by both the response geometry and the reconstructed light curves. The method's tolerance to random noise was also assessed. With appropriate denoising, the response kernel can be reliably recovered when the signal-to-noise ratio is at least $100$. We show, as a proof of concept, that the proposed method is geometrical-model independent and has the potential to offer a flexible complement to traditional approaches.

astro-ph.HE↗

arXiv2026

A clean broad iron line in GS 1354--64 as seen by XRISM

We present a spectroscopic analysis of XRISM and NuSTAR observations of the black hole X-ray binary GS~1354--64 during its 2026 outburst. A total number of 3.5 million photons are collected by the microcalorimeter Resolve on board XRISM, providing an unprecedented high-resolution view of the iron line profile. A clean broad iron line is found in the data, without significant narrow features. Modeling the broad iron line with relativistic reflection from the inner accretion disk suggests a rapidly spinning black hole (a>0.98) in the system. Measurements of the disk inclination angle from the reflection method are model-dependent. This work demonstrates the power of X-ray microcalorimeters in studying the inner accretion flow and constraining black hole parameters.

astro-ph.HE↗

arXiv2026

DAO: A New and Public Non-Relativistic Reflection Model

We present a new non-relativistic reflection model, DAO, designed to calculate reflection spectra in the rest frame of accretion disks in X-ray binaries and active galactic nuclei. The model couples the XSTAR code, which treats atomic processes, with the Feautrier method for solving the radiative transfer equation. A key feature of DAO is the incorporation of a high-temperature corrected cross section and an exact redistribution function to accurately treat Compton scattering. Furthermore, the model accommodates arbitrary illuminating spectra, enabling applications across diverse physical conditions. We investigate the spectral dependence on key physical parameters and benchmark the results against the widely used reflionx and xillver codes.

astro-ph.HE↗

arXiv2026

DeltaEvolve: Accelerating Scientific Discovery through Momentum-Driven Evolution

LLM-driven evolutionary systems have shown promise for automated science discovery, yet existing approaches such as AlphaEvolve rely on full-code histories that are context-inefficient and potentially provide weak evolutionary guidance. In this work, we first formalize the evolutionary agents as a general Expectation-Maximization framework, where the language model samples candidate programs (E-step) and the system updates the control context based on evaluation feedback (M-step). Under this view, constructing context via full-code snapshots constitutes a suboptimal M-step, as redundant implement details dilutes core algorithmic ideas, making it difficult to provide clear inspirations for evolution. To address this, we propose DeltaEvolve, a momentum-driven evolutionary framework that replaces full-code history with structured semantic delta capturing how and why modifications between successive nodes affect performance. As programs are often decomposable, semantic delta usually contains many effective components which are transferable and more informative to drive improvement. By organizing semantic delta through multi-level database and progressive disclosure mechanism, input tokens are further reduced. Empirical evaluations on tasks across diverse scientific domains show that our framework can discover better solution with less token consumption over full-code-based evolutionary agents.

cs.AI↗

arXiv2026

Constraining black hole spin in PG 1535+547 amidst complex multi-layered absorption

We present a spectroscopic analysis of XMM-Newton and NuSTAR observations of the 'complex' NLS1 PG 1535+547 at redshift $z=0.038$. These observations span three epochs: 2002 and 2006 with XMM-Newton alone, covering the $0.3-10$ keV energy range, and a coordinated XMM-Newton and NuSTAR observation in 2016, covering the $0.3-60$ keV energy range. The X-ray spectra across all epochs exhibit both neutral and ionized absorption, along with reflection features from the accretion disc, including a prominent Compton hump in the broadband data. Notably, the spectral shape varies across epochs. Our analysis suggests this variability is attributed to changes in both line-of-sight absorption and the intrinsic emission from PG 1535+547. The source is obscured by multiple layers of partially and/or fully covering neutral and ionized absorbers, with neutral column densities ranging from undetectable levels in the least obscured phase to $\sim0.3-5\times10^{23}\mathrm{cm^{-2}}$ in the most obscured phase. A clear warm absorber is revealed during the least obscured phase. The continuum remains fairly consistent ($Γ\approx 2.2\pm0.1$) during the first two observations, followed by a substantial flux decrease (by a factor of $\sim7$ in the $2-10$ keV band) in 2016 compared to 2006. The 2016 data indicates the source is in a reflection-dominated state during this epoch, with a reflection fraction of $R>7$ and an X-ray source located at a height $\leq 1.72r_g$. Simultaneous fitting of the multi-epoch data suggests a rapidly rotating black hole with a spin parameter, $a>0.99$. These findings imply that strong light-bending effects may account for the observed continuum flux reduction.

astro-ph.HE↗

arXiv2026

Modeling X-ray reflection spectra from returning radiation: application to 4U 1630$-$47

Returning radiation is thought to play a key role in disk illumination of black hole X-ray binaries in the high-soft state, yet it has not been fully incorporated into XSPEC reflection models. We present a new table model for reflection spectroscopy that, for the first time, self-consistently accounts for the returning radiation. To isolate this effect, we adopt the standard disk-corona configuration but disable the corona, allowing the reflection spectrum to be produced solely by self-irradiation of the disk. Applying our model to the black hole X-ray binary 4U 1630$-$47, we report a rapidly spinning black hole ($a_* \sim 0.99$), a disk inclination of $i \sim 53^\circ$, a mass accretion rate of $\dot{M}_{\rm BH} \sim 15\% \, {\rm \dot{M}_{Edd}}$, and an electron density of $n_{\rm e} \sim 10^{21} \,\mathrm{cm^{-3}}$ to reproduce the observed reflection features. The model also yields a source distance of $D\sim 8.2 \, {\rm kpc}$, slightly below the commonly adopted value of $10 \, {\rm kpc}$ in the literature. Compared to the widely used relxillNS, our model naturally produces a harder high-energy reflection spectrum, fitting the data without invoking a Comptonized component.

astro-ph.HE↗

arXiv2026

Evidence of mutually exclusive outflow forms from a black hole X-ray binary

Accretion onto black holes often leads to the launch of outflows that significantly influence their surrounding environments. The two primary forms of these outflows are X-ray disk winds-hot, ionized gases ejected from the accretion disk-and relativistic jets, which are collimated streams of particles often expelled along the rotational axis of the black hole. While previous studies have revealed a general association between spectral states and different types of outflows, the physical mechanisms governing wind and jet formation remain debated. Here, using coordinated NICER and MeerKAT observations of the recurrent black hole X-ray binary 4U 1630-472, we identify a clear anti-correlation between X-ray disk winds and jets: during three recent outbursts, only one type of outflow is detected at a time. Notably, this apparent exclusivity occurs even as the overall accretion luminosity remains within the range expected for a standard thin disk, characteristic of the canonical soft state. These results suggest a competition between outflow channels that may depend on how the accretion energy is partitioned between the disk and the corona. Our findings provide new observational constraints on jet and wind formation in X-ray binaries and offer a fresh perspective on the interplay between different modes of accretion-driven feedback.

astro-ph.HE↗

arXiv2026

Decoding Cygnus X-2: The Critical Role of Reflection in IXPE Data

We present a spectro-polarimetric re-analysis of the first IXPE observation of Cygnus X-2 which we determine to be mainly in the normal branch, from quasi-simultaneous observations with NuSTAR, NICER, and INTEGRAL. We measure the hard X-ray polarization angle and find it to be consistent with the previously measured position angle of the radio jet. Leveraging NuSTAR's detection of both the relativistic Fe K emission line and the Compton hump, we constrain the flux contribution of the reflected emission from the inner accretion disk to be 10% of the total X-ray flux in the IXPE energy band. Unlike previous studies that modeled only the Fe K emission line, we fit the full-band reflection spectrum using a fully relativistic disk model. There is strong degeneracy between the Comptonized and reflection components. Given that the Comptonized component is not expected to be highly polarized, a polarization degree of approximately 20% for the reflection component could explain the X-ray polarization data from IXPE. We also discuss the disk inclination angle inferred from our spectro-polarimetric modeling, as well as other possible explanations for the data.

astro-ph.HE↗

arXiv2025

Modeling reflection spectra of super-Eddington X-ray sources

Relativistic reflection is a common feature in the X-ray observations of accreting compact objects. We present reflux, a new X-ray reflection model for spectral analysis of super-Eddington sources. We develop two relativistic reflection frameworks for super-Eddington accretion: a slim-disk model that self-consistently accounts for disk thickening and self-shadowing, and an optically thick wind model that treats reflection off a funnel-shaped outflow. The slim-disk model offers a geometry where the inner disk thickness is proportional to radius, becoming thicker as the mass accretion rate increases. The wind model measures the opening angle of the funnel, the wind speed, and wind acceleration radius. The slim-disk profile reduces the brightness of the blue horn in the Fe K emission line for a fixed emissivity and significantly changes the intensity profile for a lamppost geometry. The wind model shows a blue-shifted iron line due to high velocity outflows. Both models assume a spherically symmetric spacetime. We apply the wind model to the XMM-Newton spectrum of the tidal disruption event Swift J1644+57, where the Fe K profile is expected to be shaped by scattering in an outflowing funnel. We constrain the opening angle of the funnel and find a high velocity of the wind.

astro-ph.HE↗