arXiv subjects

Zixian Wang

Publications and source records attributed to Zixian Wang.

13 recordsLinked to original sources

arXiv2026

The GECKOS survey: Assembly history of the lenticular galaxy NGC 3957

We analyse the assembly history of the edge-on lenticular galaxy NGC 3957 using deep integral-field spectroscopic MUSE data from the GECKOS survey. By applying a dust-corrected Multi-Gaussian Expansion and a population-orbit superposition model, we disentangle the galaxy's stellar kinematics, age, and metallicity. We dynamically decompose the galaxy and identify three distinct components: a dynamically-cold main disc, a compact Nuclear Stellar Disc (NSD), and a hot component. The NSD emerges as the youngest and most metal-rich component ($t = 6.9 \pm 0.4$ Gyr; $[Z/H] = 0.49 \pm 0.06$ dex), implying that the stellar bar is a long-lived structure that formed at least $\sim 7$ Gyr ago. The main stellar disc is dynamically cold ($σ_z \sim 20-30$ km/s), precluding any significant mergers over the last $\sim 8$ Gyr, and exhibits a strong positive age gradient (younger inside, older outside) beyond the bar radius. Synthesising these dynamical fossil records, NGC 3957 likely evolved as a `faded spiral' in a small-to-medium group environment. Its outer disc might passively fade due to mild gas starvation, while the bar fuelled prolonged central star formation. Comparison with S0s in the Fornax cluster reveals that this combination of internal secular evolution and mild starvation produces `outside-in' fading signatures that could mimic the environmental stripping typically seen in dense clusters.

astro-ph.GA↗

arXiv2026

SDSS-V Local Volume Mapper (LVM): The Integrated Light and Internal Rotation of Omega Centauri

The SDSS-V Local Volume Mapper (LVM) is a wide-field integral field spectroscopic survey of the Southern Milky Way plane, the Magellanic Clouds, and nearby Local Group galaxies. We use Early Science observations of the whole body of the nearest nuclear cluster, Omega Centauri, to extend the LVM beyond its primary interstellar-medium science case. The wide LVM field allows us to precisely map $ω$ Cen's line-of-sight rotation out to $\sim 3r_{HL}$ or $15^\prime$, reaching a maximum value of $(8.4 \pm 0.8)$ km s$^{-1}$ at $r \approx 4.7^\prime$. Within the central region, comparisons with existing VLT MUSE oMEGACat data show explicitly that the unresolved-light signal is dominated by a small number of bright stars, with an effective sample size of only $\sim$12 per resolution element. Using Gaia DR3 as an external reference, we verify that the SDSS-V's LVM reduction pipeline recovers integrated stellar fluxes to 1-4 % across six magnitudes of surface brightness. Our resulting total spectrum of $ω$ Cen is one of the highest S/N integrated spectrum for any globular or nuclear star cluster. We use it to test four widely-used SSP template libraries against resolved age-metallicity ground truth from oMEGACat. All templates recover an old, metal-poor population. But, even at S/N $\sim$1300, the inferred mean ages and mean [Fe/H] vary by $\sim$7 Gyr and $\sim$0.4 dex, respectively, across libraries and wavelength ranges, reflecting a systematic floor for integrated-light studies of old multi-population systems.

astro-ph.GA↗

arXiv2026

A Long Period Stellar-Mass Black Hole Binary in $ω$ Centauri

Modern simulations of stellar dynamics in globular clusters peg a dominant role for stellar-mass black holes, but direct evidence for black holes in clusters remains limited. We present the discovery of an astrometric stellar-mass black hole--main sequence star binary in $ω$ Centauri, the most massive Galactic globular cluster, using Hubble Space Telescope data from the oMEGACat project and additional JWST data that span a total of 23 years. The luminous companion to the black hole is a main-sequence turnoff star, and has a period of $94^{+63}_{-42}$ years, a semi-major axis of $31^{+15}_{-12}$ AU, and an eccentricity of $e=0.72^{+0.08}_{-0.13}$. Since we observe the binary during periastron, the mass of the black hole is well-constrained even though we only observe a partial orbit: the inferred black hole mass is $4.46^{+1.22}_{-1.01}$ M$_\odot$. We call this black hole oMEGACat BH-2. This is the first astrometric discovery of a stellar-mass black hole in a globular cluster, and is the longest period black hole binary system yet discovered. The low mass of this black hole is perhaps surprising given the low metallicity of the cluster, and shows that at least some low-mass black holes form at metallicity $Z<10^{-3}$. We find that the binary is almost certainly dynamically formed and is soft, with an expected binary disruption timescale of $\sim$800 Myr. While the total number of black hole binaries in $ω$ Centauri is uncertain, we show that existing surveys only cover a small area of parameter space, and that the presence of additional detectable black hole binaries is likely.

astro-ph.GA↗

arXiv2026

oMEGACat. X. Shedding light on the disrupted dwarf galaxy of Omega Centauri

Omega Centauri ($ω\,$Cen) is the most massive and chemically complex star cluster in the Milky Way and is widely regarded as the surviving nuclear star cluster of an accreted dwarf galaxy. However, its parent host remains uncertain. Here, we investigate a scenario in which Sequoia, Thamnos, and Gaia--Enceladus (GE) are debris from a single disrupted progenitor, the $ω\,$Dwarf, whose nucleus survives today as $ω\,$Cen. Using APOGEE and GALAH abundances together with Gaia astrometry, we reconstruct the chemical structure across this progenitor adopting orbital energy as a proxy for pre-merger radius. We find that the chemically evolved (younger Al-N-He-rich) population is strongly concentrated toward the inner regions, representing a population formed after/during the merger, while the primordial population represents a dwarf-galaxy-like population, supporting a common dwarf-galaxy origin for its components. The metallicity profile shows an inverted U-shaped gradient similar to those observed in present-day nucleated dwarf galaxies. At the same time, the inner regions ($ω\,$Cen+Thamnos) are more $α$-enhanced than the outskirts, pointing to shorter and more efficient star formation and indicating that the nucleus may have assembled through the merger of inspiraling globular clusters. Neutron-capture abundances reveal a Eu-rich, r-process-dominated outskirts and inner regions enhanced in [Ba/Eu] and [La/Eu], requiring delayed enrichment and more complex chemical evolution. Finally, our analysis shows that Sequoia and Thamnos naturally fit an outside-in stripping sequence around $ω\,$Cen, whereas the connection with GE remains unsure.

astro-ph.GA↗

arXiv2026

Studying the Soupability of Documents in State Space Models

We investigate whether hidden states from Structured State Space Models (SSMs) can be merged post hoc to support downstream reasoning. Inspired by model souping, we study document souping, a strategy where documents are encoded independently, and their representations are pooled, via simple operations like averaging, into a single context state. This approach enables modular encoding and reuse without reprocessing the full input for each query. We demonstrate that finetuned Mamba2 models with souped representations achieve competitive or superior performance across multi-hop QA, sparse retrieval, and long-document reasoning tasks compared to the standard monolithic encoding approach. For example, on the RACE and QuALITY benchmarks for long document question answering, this method substantially outperforms a traditional concatenation approach. Crucially, this modular design scales to hundreds of documents while delivering substantial savings in inference cost, unlocking new possibilities for large-scale corpus reasoning.

cs.CL↗

arXiv2026

Quiet Feature Learning in Algorithmic Tasks

We train Transformer-based language models on ten foundational algorithmic tasks and observe pronounced phase transitions in their loss curves that deviate from established power-law scaling trends. Over large ranges of compute, the validation loss barely improves, then abruptly decreases. Probing the models' internal representations reveals that quiet features are learned prior to any decrease in task loss. These quiet features represent intermediate algorithmic computations that do not by themselves improve the output loss. Ablation experiments demonstrate that individual quiet features are causally necessary for task performance. Our results demonstrate that substantial representational progress can remain hidden beneath an apparently flat loss curve, challenging the prevailing use of cross-entropy as a proxy for learning and motivating richer diagnostics for monitoring model training.

cs.LG↗

arXiv2025

Age, Chemistry, and Kinematics of the Inner Galaxy Revealed by MUSE

The bar/bulge and inner disk are fundamental building blocks of the Milky Way, containing a large fraction of its stellar mass. However, stars in these regions are faint, crowded, and have high extinction, which makes studying their formation and evolution challenging. Using the integral-field spectrograph MUSE with adaptive-optics on the Very Large Telescope, we overcome these limitations and measure accurate ages, chemical abundances, and line-of-sight velocities for 98 main-sequence turn-off and subgiant branch stars with $R_{gc}<3.5$ kpc in Baade's Window. We find that 17% stars have ages younger than 5 Gyr, and the age distribution reveals multiple peaks at 3.1, 4.8, 7.6, and 10.8 Gyr, indicating that star formation in the inner Galaxy occurred in multiple episodes. These stars are predominantly metal-rich but span a broad metallicity range ($-1.2<$[Fe/H]$<+0.6$). The [$α$/Fe]-[Fe/H] distribution shows both $α$-rich and $α$-poor sequences, with most stars being metal-rich and low-[$α$/Fe]. Our results demonstrate that IFUs enable reliable measurements of stellar parameters even in the most crowded regions of the Milky Way, opening a new pathway to study the chemodynamical evolution of the inner Galaxy.

astro-ph.GA↗

arXiv2025

Omniwise: Predicting GPU Kernels Performance with LLMs

In recent years, the rapid advancement of deep neural networks (DNNs) has revolutionized artificial intelligence, enabling models with unprecedented capabilities in understanding, generating, and processing complex data. These powerful architectures have transformed a wide range of downstream applications, tackling tasks beyond human reach. In this paper, we introduce Omniwise, the first end-to-end, self-supervised fine-tuning pipeline that applies large language models (LLMs) to GPU kernel performance prediction--a novel use case in performance profiling. Omniwise is model-agnostic and lightweight, achieving strong results even with a small 3B-parameter model. It can predict key performance metrics, including memory bandwidth, cache hit rates, GFLOPs, and arithmetic intensity, directly from kernel code without the need for code execution or profiling tools. Our approach achieves over 90% of predictions within 10% relative error on GPU kernels executed on AMD MI250 and MI300X architectures. In addition to the pipeline, we develop an online inference server and a Visual Studio Code plugin that seamlessly integrate LLM-based performance prediction into developers' workflows.

cs.LG↗

arXiv2025

oMEGACat. VI. Analysis of the overall kinematics of Omega Centauri in 3D: velocity dispersion, kinematic distance, anisotropy, and energy equipartition

Omega Centauri ($ω$ Cen) is the Milky Way's most massive globular cluster and is likely the stripped nucleus of an accreted dwarf galaxy. In this paper, we analyze $ω$ Cen's kinematics using data from oMEGACat, a comprehensive catalog of $ω$ Cen's central regions, including 1.4 million proper motion measurements and 300,000 spectroscopic radial velocities. Our velocity dispersion profiles and kinematic maps are consistent with previous work but improve on their resolution, precision, and spatial coverage. The cluster's 3D dispersion is isotropic in the core, with increasing radial anisotropy at larger radii. The 2D kinematic maps show an elongation of the velocity dispersion field comparable to the flattening observed photometrically. We find good agreement between proper motions and line-of-sight velocity dispersion and measure a kinematic distance of 5494$\pm$61 pc, the most precise kinematic distance to $ω$ Cen available. The subset of data with precise metallicity measurements shows no correlation between metallicity and kinematics, supporting the picture of well-mixed stellar populations within the half-light radius of $ω$ Cen. Finally, we study the degree of energy equipartition using a large range of stellar masses. We find partial energy equipartition in the center that decreases towards large radii. The spatial dependence of the radial energy equipartition is stronger than the tangential energy equipartition. Our kinematic observations can serve as a new reference for future dynamical modeling efforts that will help to further disentangle the complex mass distribution within $ω$ Cen.

astro-ph.GA↗

arXiv2024

Towards Understanding the Milky Way's Typicality: Assessing the Chemodynamics of M31's Bulge & Bar, Thick & Thin Discs

We describe a novel framework to model galaxy spectra with two cospatial stellar populations, such as may represent a bulge & bar or thick & thin disc, and apply it to APOGEE spectra in the inner $\sim$2 kpc of M31, as well as to stacked spectra representative of the northern and southern parts of M31's disc ($R\sim4-7$ kpc). We use a custom M31 photometric decomposition and A-LIST spectral templates to derive the radial velocity, velocity dispersion, metallicity, and $α$ abundance for both components in each spectrum. In the bulge, one component exhibits little net rotation, high velocity dispersion ($\sim$170 km s$^{-1}$), near-solar metallicity, and high $α$ abundance ([$α$/M] = 0.28), while the second component shows structured rotation, lower velocity dispersion ($\sim$121 km s$^{-1}$), and slightly higher abundances ([M/H] = 0.09, [$α$/M] = 0.3). We tentatively associate the first component with the classical bulge and the second with the bar. In the north disc we identify two distinct components: the first with hotter kinematics, lower metallicity, and higher $α$ abundance than the second ([M/H] = 0.1 and 0.39, [$α$/M] = 0.29 and 0.07). These discs appear comparable to the Milky Way's ''thick'' and ''thin'' discs, providing the first evidence that M31's inner disc has a similar chemodynamical structure. We do not identify two distinct components in the south, potentially due to effects from recent interactions. Such multi-population analysis is crucial to constrain galaxy evolution models that strive to recreate the complex stellar populations found in the Milky Way.

astro-ph.GA↗

arXiv2024

Validating full-spectrum fitting with a synthetic integral-field spectroscopic observation of the Milky Way

Ongoing deep IFS observations of disk galaxies provide opportunities for comparison with the Milky Way (MW) to understand galaxy evolution. However, such comparisons are marred by many challenges such as selection effects, differences in observations and methodology, and proper validation of full-spectrum fitting methods. In this study, we present a novel code GalCraft to address these challenges by generating mock IFS data cubes of the MW using simple stellar population models and a mock MW stellar catalog derived from E-Galaxia. We use the widely adopted full-spectrum fitting code pPXF to investigate the ability to recover kinematics and stellar populations for an edge-on mock MW IFS observation. We confirm that differences in kinematics, mean age, [M/H], and [$α$/Fe] between thin and thick disks can be distinguished. However, the age distribution is overestimated in the ranges between 2 - 4 and 12 - 14 Gyr compared to the expected values. This is likely due to the age spacing and degeneracy of SSP templates. We find systematic offsets in the recovered kinematics due to insufficient spectral resolution and the variation of line-of-sight velocity distribution with age and [M/H]. With future higher resolution and multi-[$α$/Fe] SSP templates, GalCraft will be useful to validate key signatures such as [$α$/Fe]-[M/H] distribution at different $R$ and $|z|$ and potentially infer radial migration and kinematic heating efficiency to study detailed chemodynamical evolution of MW-like galaxies.

astro-ph.GA↗

arXiv2022

Reliable stellar abundances of individual stars with the MUSE integral-field spectrograph

We present a novel approach to deriving stellar labels for stars observed in MUSE fields making use of data-driven machine learning methods. Taking advantage of the comparable spectral properties (resolution, wavelength coverage) of the LAMOST and MUSE instruments, we adopt the Data-Driven Payne (DD-Payne) model used on LAMOST observations and apply it to stars observed in MUSE fields. Remarkably, in spite of instrumental differences, according to the cross-validation of 27 LAMOST-MUSE common stars, we are able to determine stellar labels with precision better than 75K in $T_{\rm eff}$, 0.15 dex in $\log g$, and 0.1 dex in abundances of [Fe/H], [Mg/Fe], [Si/Fe], [Ti/Fe], [C/Fe], [Ni/Fe] and [Cr/Fe] for current MUSE observations over a parameter range of 3800<$T_{\rm eff}$<7000 K, -1.5<[Fe/H]<0.5 dex. To date, MUSE has been used to target 13,000 fields across the southern sky since it was first commissioned six years ago and it is unique in its ability to study dense star fields such as globular clusters or the Milky Way bulge. Our method will enable the automated determination of stellar parameters for all stars in these fields. Additionally, it opens the door for applications to data collected by other spectrographs having resolution similar to LAMOST. With the upcoming BlueMUSE and MAVIS, we will gain access to a whole new range of chemical abundances with higher precision, especially critical s-process elements such as [Y/Fe] and [Ba/Fe] that provide key age diagnostics for stellar targets.

astro-ph.SR↗

arXiv2021

A P2P Network Topology for Optimizing Partition Tolerance to Reach the CAP Guarantee Bound in Consortium Blockchains

Decentralized cryptocurrency systems, known as blockchains, have shown promise as infrastructure for mutually distrustful parties to agree on transactions safely. However, Bitcoin-derived blockchains and their variants suffer from the limitations of the CAP Trilemma, which is difficult to be solved through optimizing consensus protocols. Moreover, the P2P network of blockchains have problems in efficiency and reliability without considering the matching of physical and logical topologies. For the CAP Trilemma in consortium blockchains, we propose a physical topology based on the multi-dimensional hypercube with excellent partition tolerance in probability. On the other hand, the general hypercube has advantages in solving the mismatch problems in P2P networks. The general topology is further extended to a hierarchical recursive topology with more medium links or short links to balance the reliability requirements and the costs of physical network. We prove that the hypercube topology has better partition tolerance than the regular rooted tree topology and the ring lattice topology, and effectively fits the upper-layer protocols. As a result, blockchains constructed by the proposed topology can reach the CAP guarantee bound through adopting the proper transmission and the consensus protocols protocols that have strong consistency and availability.

cs.NI↗