betaFeedback

WorksPhysics and Astronomy

The Distant Milky Way Halo from the Southern Hemisphere: Spatially broad and closer stellar debris associated to the Magellanic Stream

Abstract

The Magellanic Stream (MS) is a gaseous structure extending over 150∘150^\circ across the sky, likely formed through interactions between the Large and Small Magellanic Clouds (LMC and SMC) as they fell into the Milky Way (MW). Its predicted stellar counterpart would provide a collisionless record of these interactions, but few candidate members have been identified. To expand this sample, we constructed a catalog of 87,763 distant K-giant candidates using DELVE DR2 and VHS DR5 photometry and Gaia DR3 astrometry, and analyzed VLT and Magellan spectra of 51 targets along the trailing MS. Combining radial velocities, spectrophotometric distances, and proper motions, we identify five new MS candidates at 58–95 kpc. Their metallicities are consistent with the outskirts of the SMC, while orbit integrations in a time-dependent MW–LMC–SMC potential support a possible Magellanic origin. Combined with previously identified candidates, these stars span a substantially broader range of Magellanic latitude than predicted by existing models, suggesting an extended debris field rather than a narrow stream. The candidates also lie predominantly closer than the simulated debris. More recent, less eccentric SMC orbits may help explain this distance discrepancy, although new simulations are required to test their implications for the stellar debris, along with larger spectroscopic samples.

Image not extracted

Figure *.

Corresponding author: Manuel Cavieres Email: [email protected]

Introduction

The LMC and SMC are the nearest example of an ongoing interaction between massive satellite galaxies and a Milky Way (MW)-mass host [8, 22]. Their proximity makes the Clouds uniquely valuable laboratories for studying satellite accretion, dwarf–dwarf interactions and gas removal [7, 8], as well as the dynamical response of a galaxy halo to a massive perturber (e.g., [32, 69, 28, 75]).

The proper motions of the Clouds indicate that they must have interacted recently [41, 91, 65]. During this galaxy–galaxy interaction, tidal theory demands the formation of tidal bridges and tails [83]. A gaseous bridge between the Clouds and a stream trailing them are both known to exist [52, 74, 63, 56]. Every simulation of the Clouds’ interaction history predicts the formation of a less prominent stellar tidal tail (the stellar MS), in addition to a gaseous tidal tail (e.g., [48, 33, 76, 20, 5, 49]).

Spectroscopic follow-up of distant halo samples have revealed stars potentially associated to the stellar MS [89], although their kinematics can also be associated to the GES merger [15]. More recently, through spectroscopic followup of K giants selected from WISE [88] and Gaia (Gaia Collaboration 2023), V. Chandra et al. [15] were able to identify 13 high-confidence members of the stellar counterpart to the gaseous trailing arm of the Stream (stellar MS). They combined spectroscopic distances, radial velocities, and metallicities with Gaia astrometry to obtain chemo-dynamical constraints. We refer to these thirteen stars as the Chandra sample (labeled Chandra+23 in the figures). This discovery suggests that a stellar MS may now be detected, but the existing spectroscopic samples remain small. A larger sample is therefore needed to constrain the kinematic structure of the debris, its extent across the sky, and its distance.

If stars can be confidently associated with the MS, their positions, distances, radial velocities, proper motions, and ages can provide collisionless constraints on the timing, geometry, and strength of the LMC–SMC interaction. The gas alone cannot provide these constraints because gas and stars respond to different physical processes (for a review, see [22]). Gas can be displaced by ram pressure [77], cooling and feedback [62], and interaction with the circumgalactic medium [30, 50], whereas stars are essentially collisionless tracers of the gravitational field [20, 5] and therefore provide critical constraints on the orbital geometry of the system.

In this paper, we present the results of a new search for stellar members of the MS. We selected distant K-giants using optical photometry from the DECam Local Volume Exploration Survey (DELVE, [24]) and near-infrared photometry from the VISTA Hemisphere Survey (VHS, [55]). This new optical–near-infrared selection reaches fainter magnitudes than earlier Gaia–WISE selections and enables the identification of candidate giants at heliocentric distances comparable to those expected for Magellanic debris.

We obtained medium-resolution spectra of a subsample of these K-giants to measure radial velocities and metallicities. These quantities provide the key tests of a Magellanic origin. Stars stripped from the SMC should occupy the appropriate region of phase space, follow a coherent velocity trend along the Stream, and have metallicities consistent with Magellanic stellar populations. More importantly, confirmed stellar debris can be used as a dynamical tracer of the SMC’s past orbit. Unlike the gas, these old stars retain a largely collisionless memory of the stripping event, making them especially useful for testing models of the latest LMC–SMC interaction.

The paper is organized as follows. Section 2 describes the photometric and astrometric data and the quality cuts applied. Section 3 presents the selection of distant halo K giants, and Section 4 describes their spectroscopic follow-up. Section 5 examines the stars’ association with the Clouds using their 3D positions and motions. We summarize our results in Section 6.

Survey data

In this section, we describe the photometric and astrometric surveys used for this study. The photometric data come mainly from the second data release of DELVE [25], which was a survey program on the 4-m Blanco Telescope at the Cerro Tololo Inter-American Observatory in Chile. DELVE covers ∼15,000 deg2\sim15{,}000\ \mathrm{deg}^{2} of the Southern sky at high Galactic latitude (∣b∣>10 deg\lvert b\rvert> 10\ \mathrm{deg}). It provides photometry in the gg, rr, ii, and zz bands to a 5σ5\sigma depth of ∼23.5\sim23.5 mag.

We cross-matched the second data release of the DELVE catalog with the 5th data release of VHS [54], which provides photometric coverage across the Southern celestial hemisphere in the JJ and KSK_S near-infrared bands using the VIRCAM instrument, formerly installed on the VISTA telescope. These combined catalogs encompass most of the Southern celestial hemisphere at J<17J < 17 mag, with the necessary photometric bands to select distant halo tracers such as the K giants used in this study. To remove clear foreground objects using Gaia parallaxes, we also cross-matched this combined catalog with the third data release of the Gaia mission (Gaia Collaboration 2023).

To ensure reliable measurements, the DELVE query required point-source classifications in the gg, ii, and zz bands and zero photometric flags in the gg, rr, ii, and zz bands. The final catalog additionally required J<17J < 17. Where overlapping VHS tiles provide more than one detection of the same Gaia source, we kept a single entry, preferring the detection flagged as primary in VHS (PRIMARY_SOURCE = 1). These selections are listed in Table 1.

SurveySelection criteria
DELVEextended_class_g = 0
DELVEextended_class_i = 0
DELVEextended_class_z = 0
DELVEflags_g = 0
DELVEflags_r = 0
DELVEflags_i = 0
DELVEflags_z = 0
VHSJAPERMAG4 < 17
VHSone detection per Gaia DR3 source

Table 1. Quality selections used to define the final 87 763-object catalog.

The observed DELVE and VHS magnitudes were used in the color–color and color–magnitude diagrams, and preliminary photometric-distance selections.

Halo K-giant selection

We selected K giants based on their position in the giKSgiK_S diagram, where the tip of the red giant branch (RGB) separates from the main stellar locus, as shown in Figure 1. The giKSgiK_S diagram was first presented by M. Cavieres et al. [14] as an effective tool to separate stellar sources from unresolved galaxies. It was developed based on (u−i)(u-i), (i−Ks)(i-K_s) color-color diagrams, extensively used to separate stars and galaxies at large distances (see [60, 64]), and the identification of K giant stars based on mid-infrared WISE photometry in C. Conroy et al. [17]. In this plane, metal-poor RGB stars are offset towards redder i−Ksi-K_s colors at fixed g−ig-i relative to main-sequence stars, which allows K giants to be selected using only gg, ii, and KsK_s bands (M. Cavieres et al. [14]). The K giants were selected to lie in the shaded gray region shown in Figure 1, corresponding to the interior of the polygon defined by:

giKs color–color diagram showing the selection region near the tip of the RGB as a gray-shaded area

Figure 1. giKsgiK_s color–color diagram showing the selection region near the tip of the RGB as a gray-shaded area. Spectroscopically selected K giants from L. Zhang et al. [90] are shown as red points. A PARSEC isochrone for a 10 Gyr, [M/H] = −0.9-0.9 population is shown in green for the main sequence and yellow for the RGB.

P(g−i, i−Ks)={(1.31, 2.15),(1.88, 2.43),(2.60, 3.26),(2.60, 3.46),(1.32, 2.47)}.(1)\mathcal{P}_{(g-i,\ i-K_s)} = \left\{ \begin{aligned} &(1.31,\ 2.15),\\ &(1.88,\ 2.43),\\ &(2.60,\ 3.26),\\ &(2.60,\ 3.46),\\ &(1.32,\ 2.47) \end{aligned} \right\}. \tag*{(1)}

This selection, motivated by the position of the RGB of a PARSEC 1.2S (A. Bressan et al. [12]) isochrone in this diagram (age = 10 Gyr, [M/H] = −0.9-0.9), allows stars at the tip of the RGB to be separated from dwarfs through their infrared excess, while limiting contamination from unresolved background galaxies, which dominate the counts at faint magnitudes.

The final DELVE-VHS-Gaia catalog contains 87 763 K-giant candidates distributed over an effective sky area of approximately 1.4×1041.4 \times10^4 deg2^2, corresponding to a mean surface density of 6.36.3 stars deg−2^{-2}. The footprint spans all Galactic longitudes and extends from b=−86.8∘b=-86.8^\circ to +58.4∘+58.4^\circ, with 93.9% of the sources located at ∣b∣>10∘|b|>10^\circ. The catalog covers 15.0<g<22.415.0<g<22.4 mag in DELVE PSF magnitudes, has a median magnitude of g=19.3g=19.3, and contains 23 989 sources satisfying the foreground-rejection criterion ϖ<0.1\varpi<0.1 mas, corresponding to ≈10\approx10 kpc. The full catalog, including the photometric distances of § 3.1, will be made publicly available on Zenodo2

Photometric distances

We estimated a preliminary photometric distance for each K-giant candidate by treating it as an RGB star. We linearly interpolated the RGB of a single 10 Gyr, [Fe/H] =−1.5=-1.5 PARSEC v1.2S isochrone (A. Bressan et al. [12]) in the (g−i,Mg)(g-i,M_g) plane and converted g−Mgg-M_g to a distance modulus. These point estimates, computed from the observed photometry, were used for target selection. Stars whose g−ig-i color lies outside the color range of the isochrone RGB have no photometric distance; this applies to 4 267 of the 23 989 sources with ϖ<0.1\varpi<0.1 mas.

Spectroscopic follow-up

In this section we describe how we selected targets from the K-giant catalog for spectroscopic follow-up of the stellar MS. We present the observations and the methods used to obtain stellar parameters, radial velocities, metallicities, and spectro-photometric distances.

Candidate selection for X-shooter follow-up

Follow-up targets were required to be distant, to lie along the trailing arm of the Stream, and to have a Gaia DR3 proper motion compatible with that of Magellanic debris. The criteria are collected in Table 2 and described below in the order in which they were applied. Foreground stars were removed with a parallax cut, ϖ<0.1\varpi< 0.1 mas, and the sample was restricted to the Southern Galactic hemisphere (b<0∘b < 0^\circ), where the trailing arm is found. Distant giants were then isolated directly in the color–magnitude diagram. We fitted a fifth-order polynomial to the RGB of a 10 Gyr, [M/H] = −0.9 PARSEC v1.2S isochrone (A. Bressan et al. [12]) in the (g−ig-i, gg) plane, shifted the resulting RGB to heliocentric distances of 40 and 80 kpc, and kept the 11 993 stars falling between the two shifted ridgelines. This band, shown in Figure 2, brackets the distance range in which Magellanic debris is expected along the trailing arm, while its faint edge keeps the selection above the magnitude regime where unresolved background galaxies dominate the counts.

Isochrone distance-based selection of K-giant stars for the spectroscopic follow-up

Figure 2. Isochrone distance-based selection of K-giant stars for the spectroscopic follow-up. The two ridgelines correspond to a 10 Gyr, [M/H] = −0.9 PARSEC v1.2S population placed at 40 and 80 kpc. Selecting in this distance regime retains distant objects without incorporating contaminants at fainter magnitudes.

Criterion
ϖ<0.1\varpi< 0.1 mas
b<0∘b < 0^\circ
Between the 40 and 80 kpc RGB ridgelines
−125∘<LMS<−25∘-125^\circ< L_{\mathrm{MS}} < -25^\circ
−50∘<BMS<20∘-50^\circ< B_{\mathrm{MS}} < 20^\circ
0<μL<30 < \mu_L < 3 mas yr−1\mathrm{yr}^{-1}
−0.5<μB<0.5-0.5 < \mu_B < 0.5 mas yr−1\mathrm{yr}^{-1}
d>50−0.3(LMS+25∘)d > 50 - 0.3 (L_{\mathrm{MS}} + 25^\circ) kpc
d<80−1.2(LMS+25∘)d < 80 - 1.2 (L_{\mathrm{MS}} + 25^\circ) kpc

Table 2. Selection criteria applied to the K-giant catalog to define the X-shooter target list, in the order in which they were applied. Here dd is the preliminary isochrone-based photometric distance described in § 3.1, obtained by interpolating a 10 Gyr, [Fe/H] = −1.5-1.5 PARSEC isochrone in the (g−i,Mg)(g-i, M_g) plane. The RGB ridgelines instead define the color–magnitude band using the [M/H] = −0.9-0.9 isochrone described in the text.

Sky positions and proper motions were then transformed into the Magellanic Stream coordinate system of D. L. Nidever et al. [63], and a deliberately generous window was applied around the trailing arm position, −125∘<LMS<−25∘-125^\circ< L_{\mathrm{MS}} < -25^\circ and −50∘<BMS<20∘-50^\circ< B_{\mathrm{MS}} < 20^\circ. The longitude range excludes the bodies of the Clouds themselves while covering the full extent of the trailing arm within the survey footprint, and the latitude range was chosen to be wide and asymmetric about the gaseous Stream, so as not to assume that the stellar debris follows strictly the spatial distribution of H i. This window contains 150 entries in the catalog, corresponding to 126 distinct Gaia sources; the 24 sources appearing twice, as a consequence of repeated DELVE or VHS detections of the same star in overlapping tiles, were discarded altogether to avoid any ambiguity in the target coordinates, leaving 102 candidates.

Two kinematic cuts were then imposed on the proper motions projected onto the Stream frame: 0<μL<30 < \mu_L < 3 mas yr−1^{-1}, which requires motion along the Stream in the sense of the trailing arm, and ∣μB∣<0.5|\mu_B| < 0.5 mas yr−1^{-1}, which requires that motion to be confined to the Stream plane. These cuts leave 97 and 85 candidates, respectively. Finally, we imposed a distance–longitude window following the distance gradient expected along the trailing arm, keeping stars with d>50−0.3(LMS+25∘)d > 50 - 0.3(L_{\mathrm{MS}} + 25^\circ) kpc and d<80−1.2(LMS+25∘)d < 80 - 1.2(L_{\mathrm{MS}} + 25^\circ) kpc, where dd is the preliminary photometric distance inferred from the 10 Gyr, [Fe/H] = −1.5 RGB isochrone described in § 3, we combine both distance estimates as one represents a general halo star age and metallicity, while the other assumes an SMC halo-like star. These two straight lines in the (LMS,d)(L_{\mathrm{MS}},d) plane generously bracket the distance gradient predicted for the trailing arm by the G. Besla et al. [8], hereafter B+12) models, widening from 50–80 kpc at LMS=−25∘L_{\mathrm{MS}} = -25^\circ to 80–200 kpc at LMS=−125∘L_{\mathrm{MS}} = -125^\circ. This last cut yields the final X-shooter target list of 51 stars, spanning −101.6∘<LMS<−28.6∘-101.6^\circ< L_{\mathrm{MS}} < -28.6^\circ, −43.1∘<BMS<16.7∘-43.1^\circ< B_{\mathrm{MS}} < 16.7^\circ, 57<d<11257 < d < 112 kpc and 17.2<g<18.817.2 < g < 18.8. The 12.8 h awarded to the program allowed us to observe 33 of them. These stars span 17.2<g<18.517.2 < g < 18.5 and 57<d<9957 < d < 99 kpc and are not a strictly magnitude-limited subset of the target list; they constitute the more targeted VLT search discussed in the rest of the paper.

X-shooter observations

We obtained medium-resolution spectra with the X-shooter instrument (J. Vernet et al. [85]), mounted on the Very Large Telescope. Échelle long-slit spectra were acquired for 33 stars over 12.8 h in program 115.28GR.001 (PI J. Chanamé), observed in Service Mode from July to September 2025. We used slit widths of 0\farcs8, 0\farcs7, and 0\farcs6 for the UVB, VIS, and NIR arms, respectively, yielding spectral resolutions of ∼\sim6 700, 11 400 and 8 100 in the respective arms. Exposure times ranged from 600 to 900 s, yielding a mean S/N of ∼\sim10, ∼\sim23, and ∼\sim27 for UVB, VIS, and NIR spectra.

The reduced, order-merged, and extracted 1D spectra were retrieved from the X-shooter ESO Phase 3 Data Products. The raw data were reduced with the ESO pipeline xshoo-2.3.12, which performs bias (UVB, VIS arms) and dark (NIR arm) correction, cosmic-ray removal, flat-field correction, sky subtraction, wavelength Note— LMSL_{\mathrm{MS}} and BMSB_{\mathrm{MS}} are MS coordinates (D. L. Nidever et al. [63]), μL\mu_L and μB\mu_B the corresponding proper-motion components.

calibration, and flux calibration. No telluric correction was performed; instead, regions affected by telluric absorption were masked, as explained in § 4.5.

Candidate selection for the MagE follow-up

A second target list was built to extend the coverage of the spectroscopic sample further along the trailing arm than the X-shooter selection reached. The criteria are collected in Table 3 and described below in the order in which they were applied. Starting from the 20 001 sources with ϖ<0.1\varpi< 0.1 mas whose VHS detection is the primary one (PRIMARY_SOURCE = 1), a subset of the 23 989 such sources in the final catalog (§ 3), we repeated the color–magnitude selection of § 4 with the RGB ridgeline placed at 40 and 160 kpc rather than 40 and 80 kpc, aiming to encompass the larger distances expected at the far end of the arm. A total of 15 452 stars fall between the two ridgelines. Neither the hemisphere cut nor the trailing-arm window of § 4 was imposed at this stage: we required only −135∘<LMS<80∘-135^\circ< L_{\mathrm{MS}} < 80^\circ and ∣BMS∣<50∘\lvert B_{\mathrm{MS}}\rvert< 50^\circ, leaving 10 764 stars. The kinematic criteria were likewise relaxed to −1<μL<3-1 < \mu_L < 3 mas yr−1\mathrm{yr}^{-1}, which admits motion along the Stream in either sense, and ∣μB∣<0.5\lvert\mu_B\rvert< 0.5 mas yr−1\mathrm{yr}^{-1}, leaving 10 411 and 6 870 candidates, respectively.

Criterion
ϖ<0.1\varpi< 0.1 mas
Between the 40 and 160 kpc RGB ridgelines
−135∘<LMS<80∘-135^\circ< L_{\mathrm{MS}} < 80^\circ
∣BMS∣<50∘\lvert B_{\mathrm{MS}}\rvert< 50^\circ
−1<μL<3-1 < \mu_L < 3 mas yr−1\mathrm{yr}^{-1}
∣μB∣<0.5\lvert\mu_B\rvert< 0.5 mas yr−1\mathrm{yr}^{-1}
LMS<−25∘L_{\mathrm{MS}} < -25^\circ

Table 3. Selection criteria applied to the K-giant catalog to define the MagE target list, in the order in which they were applied. The criteria deliberately relax those of Table 2 so as to reach further along the trailing arm; see § 4.3. The RGB ridgelines define the color–magnitude band using the same [M/H] = −0.9-0.9 isochrone as in Table 2, shifted to 40 and 160 kpc.

Restricting the sample to the trailing arm, LMS<−25∘L_{\mathrm{MS}} < -25^\circ, yields 136 targets, of which 27 had already been observed with X-shooter and were removed. The remaining 109 candidates span −103.7∘<LMS<−25.2∘-103.7^\circ< L_{\mathrm{MS}} < -25.2^\circ and 17.0<g<19.917.0 < g < 19.9 mag; of these, 28 were observ-

Note— LMSL_{\mathrm{MS}}, BMSB_{\mathrm{MS}}, μL\mu_L and μB\mu_B are as in Table 2. The longitude, latitude and proper-motion windows are wider than their X-shooter counterparts; in particular, the μL\mu_L range admits motion along the Stream in either sense.

able during the allocated time (§ 4.4) and were observed; 18 yielded reliable radial velocities and atmospheric parameters. The resulting MagE spectroscopic sample spans −95.9∘<LMS<−25.2∘-95.9^\circ< L_{\mathrm{MS}} < -25.2^\circ, −24.1∘<BMS<23.7∘-24.1^\circ< B_{\mathrm{MS}} < 23.7^\circ, 17.0<g<19.617.0 < g < 19.6 mag and 55<d<16155 < d < 161 kpc.

MagE observations

We carried out this second spectroscopic follow-up with the Magellan Echelle Spectrograph (MagE; J. L. Marshall et al. [51]) mounted on the 6.5 m Magellan Baade telescope at Las Campanas Observatory, as part of the program CN2025B-95 (PI J. Chanamé), over two observing nights (early November 2025 and early January 2026). We used a 0.7′′0.7^{\prime\prime} slit width, yielding a spectral resolution of ∼5850\sim5850, with magnitude-dependent exposure times ranging from 15 to 40 minutes to achieve a minimum S/N of ∼15\sim15 at 5100 Å.

The data were reduced with the PyPeIt 1.17.3 pipeline (J. Prochaska et al. [73]; Prochaska et al. [72]), which carries out bias and flat-field correction, wavelength calibration using paired ThAr arc exposures, and optimal extraction of each order. Following the standard PyPeIt procedure, the individual orders were flux-calibrated and merged into a continuous spectrum. The continuum was estimated using the procedure implemented in the iSpec package (S. Blanco-Cuaresma et al. [10]; S. Blanco-Cuaresma [9]), adapted to the MagE spectra by adjusting the continuum-fitting parameters, including the instrumental resolution and spline degree. A spline function was fitted to the observed spectrum, with strong absorption features automatically identified and excluded from the fit based on local flux variations. Finally, the normalized spectrum was obtained by dividing the observed flux by the fitted continuum.

Stellar parameters

We derived stellar atmospheric parameters and radial velocities from the X-shooter VIS and MagE spectra using the rvspecfit package (S. E. Koposov [44]). We compared each spectrum with PHOENIX v2.0 synthetic stellar spectra (T.-O. Husser et al. [39]), jointly fitting the effective temperature (TeffT_{\mathrm{eff}}), surface gravity (log⁡g\log g), metallicity ([Fe/H][\mathrm{Fe}/\mathrm{H}]), α-element abundance ([α/Fe][\alpha/\mathrm{Fe}]), and radial velocity. A smooth multiplicative polynomial was included to account for large-scale differences in continuum shape between the observations and models, ensuring that the fit was driven primarily by the absorption-line information.

Pixels with invalid uncertainties and the strong atmospheric O2_2 A-band were excluded from the X-shooter fits. Regions affected by telluric absorption were identified and masked out using TelFit (K. Gullikson et al. 2014). For the MagE data, we restricted our analysis to the spectral region between 5 000 and 6 700 Å, thereby avoiding regions of low S/N and less reliable wavelength calibration. Figure 3 highlights the agreement between the observed spectrum from XSHOOTER and its best-fitting PHOENIX model around Hα, the Ca II triplet, and the Mg I line at 880.7 nm. The resulting TeffT_{\mathrm{eff}}, log⁡g\log g, and [Fe/H][\mathrm{Fe}/\mathrm{H}] measurements were subsequently used as spectroscopic constraints in the distance determination described below.

Observed and PHOENIX-fit spectra around Hα, the Ca II triplet, and the Mg I line at 880.7 nm

Figure 3. Flux-calibrated spectra of Gaia DR3 689814983025950508, zoomed in on the Hα, Ca II triplet, and Mg I regions to show the quality of the rvspecfit fit.

The covariance matrix returned by rvspecfit is derived from the local Hessian of the likelihood (S. E. Koposov [44]; S. E. Koposov et al. [45]). The corresponding uncertainties therefore quantify only the formal, local statistical precision conditional on the adopted spectral model and do not account for correlated-pixel noise, continuum-placement errors, or imperfections and interpolation systematics in the PHOENIX templates (T. S. Li et al. (2019); A. Lançon et al. [46]; S. E. Koposov et al. [45]). External validation of rvspecfit and X-shooter full-spectrum analyses indicates atmospheric-parameter uncertainties of order 0.2–0.3 dex, with the largest discrepancies occurring for cool, low-gravity giants (T. S. Li et al. (2019); A. Arentsen et al. [1]; S. E. Koposov et al. [45]). We therefore adopt literature-informed systematic uncertainty floors of 0.30 dex in log⁡g\log g, 0.20 dex in [Fe/H][\mathrm{Fe}/\mathrm{H}], and 100 K in TeffT_{\mathrm{eff}} combined in quadrature with the formal uncertainties according to

σtot2=σHess2+σsys2\sigma_{\mathrm{tot}}^2 = \sigma_{\mathrm{Hess}}^2 + \sigma_{\mathrm{sys}}^2

For the MagE spectra, we additionally compared our results with independent measurements for a small number of targets in common with other samples (two stars with X-shooter analyses and one star, observed twice with MagE, from V. Chandra et al. [15]). Based on the scatter in [Fe/H][\mathrm{Fe}/\mathrm{H}] relative to these comparison stars, we adopt a larger systematic floor in [Fe/H][\mathrm{Fe}/\mathrm{H}] for the MagE sample (0.30 dex instead of 0.20 dex). For fits flagged with bad_hessian, for which the local covariance matrix could not be robustly estimated, we report the adopted systematic uncertainty alone. The resulting atmospheric parameters are shown in Table 4 and Table 5, for X-shooter and MagE data, respectively.

Gaia DR3 source IDInst.RA [deg]Dec. [deg]dd [kpc]vhelv_{\mathrm{hel}} [km s−1^{-1}]TeffT_{\mathrm{eff}} [K]log⁡g\log g [dex][Fe/H] [dex]
2622351759280470528MagE335.893366−7.425410-7.42541095−18+1295^{+12}_{-18}−103.68±0.46-103.68 \pm0.464298±1004298 \pm1000.40±0.300.40 \pm0.30−1.59±0.20-1.59 \pm0.20
2441369916261698304X-Sh359.169473−7.962814-7.96281490.2−2.0+1.890.2^{+1.8}_{-2.0}−169.58±0.04-169.58 \pm0.043660±1003660 \pm100−0.35±0.30-0.35 \pm0.30−1.75±0.20-1.75 \pm0.20
4849744515068988544MagE51.719178−41.285058-41.28505881.7−9.3+6.581.7^{+6.5}_{-9.3}−18.80±0.16-18.80 \pm0.164264±1004264 \pm100−0.06±0.30-0.06 \pm0.30−1.76±0.20-1.76 \pm0.20
6503888084644067712X-Sh341.691423−57.742600-57.74260057.9−4.2+3.257.9^{+3.2}_{-4.2}−64.92±0.07-64.92 \pm0.074403±1004403 \pm1000.25±0.300.25 \pm0.30−1.27±0.20-1.27 \pm0.20
4716496865562228736MagE19.178275−60.970305-60.97030557.7−11.4+8.057.7^{+8.0}_{-11.4}60.94±0.2760.94 \pm0.274130±1004130 \pm1000.34±0.300.34 \pm0.30−1.16±0.20-1.16 \pm0.20

Table 4. Coordinates and stellar parameters of the five MS candidates, from both the X-shooter and MagE samples. Metallicities are the spectroscopic values.

Gaia DR3 source IDInst.RADec.dd [kpc]vhelv_{\mathrm{hel}} [km s−1^{-1}]TeffT_{\mathrm{eff}} [K]log⁡g\log g [dex][Fe/H] [dex]
2349552547787827456X-Sh11.773665−22.08009283.5−8.4+6.883.5^{+6.8}_{-8.4}−129.66 ± 0.094485 ± 1000.34 ± 0.30−1.50 ± 0.20
2356964085088274688MagE12.627074−18.94761162.1−5.1+4.462.1^{+4.4}_{-5.1}45.97 ± 0.184157 ± 100−0.39 ± 0.30−1.79 ± 0.20
2408907694443864832X-Sh352.122466−13.94002119.9−2.3+2.419.9^{+2.4}_{-2.3}−191.60 ± 0.144614 ± 1002.00 ± 0.30−1.07 ± 0.20
2409939891048968576a^{a}MagE345.480647−15.07459350.4−9.9+10.750.4^{+10.7}_{-9.9}−250.48 ± 0.504558 ± 1001.03 ± 0.30−1.29 ± 0.20
2411032809902332160X-Sh344.615379−13.84506635.1−4.5+5.235.1^{+5.2}_{-4.5}−162.44 ± 0.234773 ± 1001.63 ± 0.30−1.00 ± 0.20
2413362606322074240MagE347.302020−10.98943535.8−7.7+9.035.8^{+9.0}_{-7.7}−133.25 ± 0.514613 ± 1071.50 ± 0.30−0.94 ± 0.20
2426397007431829760MagE6.926652−10.19294359.8−7.4+5.259.8^{+5.2}_{-7.4}17.01 ± 0.204249 ± 1000.01 ± 0.30−1.75 ± 0.20
2444786098888864256X-Sh1.763387−4.47566843.7−5.1+5.843.7^{+5.8}_{-5.1}−77.73 ± 0.124533 ± 1001.35 ± 0.30−1.00 ± 0.20
2481963537665282944a^{a}MagE19.829467−5.37349455.3−10.4+9.755.3^{+9.7}_{-10.4}−134.50 ± 0.264375 ± 1000.50 ± 0.30−1.65 ± 0.20
2529777385642667136X-Sh12.537191−2.09958746.6−5.1+6.246.6^{+6.2}_{-5.1}−87.31 ± 0.094674 ± 1000.94 ± 0.30−1.00 ± 0.20
2534734362017687680MagE18.063123+0.10128762.0−6.9+5.362.0^{+5.3}_{-6.9}−121.90 ± 0.164245 ± 100−0.00 ± 0.30−1.73 ± 0.20
2601835941752991872X-Sh338.639362−11.79016013.4−1.4+1.713.4^{+1.7}_{-1.4}−207.69 ± 0.164504 ± 1002.50 ± 0.30−0.50 ± 0.20
2602183761089670784X-Sh336.252480−12.13342752.5−5.8+5.852.5^{+5.8}_{-5.8}−139.10 ± 0.104557 ± 1010.66 ± 0.30−1.12 ± 0.20
2606713100456070912MagE345.204876−9.33777755.0−4.6+3.755.0^{+3.7}_{-4.6}−94.79 ± 0.244092 ± 100−0.20 ± 0.30−1.86 ± 0.20
2607354459332524416X-Sh345.698125−8.32847652.4−6.8+7.952.4^{+7.9}_{-6.8}−95.37 ± 0.104622 ± 1010.77 ± 0.31−1.00 ± 0.20
2609809183465911680X-Sh340.332831−7.52735455.1−5.1+4.755.1^{+4.7}_{-5.1}−42.10 ± 0.094442 ± 1000.71 ± 0.30−1.00 ± 0.20
2610146806550153216X-Sh342.566724−8.53336951.3−5.6+5.951.3^{+5.9}_{-5.6}−212.07 ± 0.114600 ± 1000.88 ± 0.30−1.16 ± 0.20
2614779835016325120X-Sh334.517977−10.59895756.0−5.3+4.856.0^{+4.8}_{-5.3}−242.17 ± 0.114600 ± 1000.69 ± 0.30−1.10 ± 0.20
2615394535031185920X-Sh336.402947−8.72836922.0−2.9+3.122.0^{+3.1}_{-2.9}−173.35 ± 0.144600 ± 1001.95 ± 0.30−1.00 ± 0.20
2627238263832329984MagE332.790557−4.17940844.5−7.3+8.244.5^{+8.2}_{-7.3}−293.47 ± 0.304500 ± 1000.80 ± 0.30−1.37 ± 0.20
2627493419249681280MagE335.764306−3.42582139.6−7.2+7.739.6^{+7.7}_{-7.2}−8.42 ± 0.204441 ± 1011.13 ± 0.30−1.41 ± 0.20
2680349107859540096MagE331.184038−0.44383740.5−7.5+10.640.5^{+10.6}_{-7.5}−278.19 ± 0.584653 ± 1001.70 ± 0.31−1.00 ± 0.20
4910361621900734208a^{a}X-Sh17.417055−57.53194239.4−5.2+6.239.4^{+6.2}_{-5.2}74.78 ± 0.144823 ± 1001.40 ± 0.30−1.00 ± 0.20
4944973693890618880MagE38.483418−47.73589529.4−6.1+5.429.4^{+5.4}_{-6.1}34.95 ± 0.324478 ± 1012.60 ± 0.300.10 ± 0.20
4977773229557246976MagE2.287267−48.37698043.0−7.2+10.343.0^{+10.3}_{-7.2}73.64 ± 0.284350 ± 1000.59 ± 0.30−1.34 ± 0.20
4979295743923207680a^{a}X-Sh9.507136−46.38998827.7−3.3+4.027.7^{+4.0}_{-3.3}2.06 ± 0.064459 ± 1001.30 ± 0.30−1.00 ± 0.20
4993221054391991040MagE9.057444−41.27132170−14+1070^{+10}_{-14}117.63 ± 0.514644 ± 1031.00 ± 0.30−1.27 ± 0.20
6458074626274350080X-Sh315.346833−55.19857359.6−8.2+10.659.6^{+10.6}_{-8.2}155.90 ± 0.064518 ± 1000.50 ± 0.30−1.18 ± 0.20
6518522240811816320X-Sh336.689384−47.02977052.3−5.5+5.552.3^{+5.5}_{-5.5}−39.50 ± 0.114376 ± 1000.88 ± 0.30−1.00 ± 0.20
6518847318297206528X-Sh336.442639−46.0933967.90−0.87+1.017.90^{+1.01}_{-0.87}6.85 ± 0.144723 ± 1003.00 ± 0.30−0.62 ± 0.20
6532434911193453952MagE354.412162−42.28448247.2−3.5+3.347.2^{+3.3}_{-3.5}187.63 ± 0.154068 ± 100−0.50 ± 0.30−1.84 ± 0.20
6536701153747802368X-Sh350.799829−40.74612257.2−9.7+6.857.2^{+6.8}_{-9.7}148.57 ± 0.084624 ± 1000.50 ± 0.30−1.36 ± 0.20
6544916189235036672X-Sh339.468720−42.52277473.1−6.0+4.373.1^{+4.3}_{-6.0}14.31 ± 0.074367 ± 1000.19 ± 0.30−1.15 ± 0.20
6548490220500114400a^{a}X-Sh349.620432−39.93386356.8−1.9+1.556.8^{+1.5}_{-1.9}−115.32 ± 0.053900 ± 100−0.50 ± 0.30−1.46 ± 0.20
6560803097275995904a^{a}X-Sh329.940187−48.46420838.9−5.8+7.638.9^{+7.6}_{-5.8}48.49 ± 0.204768 ± 1001.50 ± 0.30−1.16 ± 0.20
6579489610816557184X-Sh323.029483−40.58909413.2−1.5+1.213.2^{+1.2}_{-1.5}−84.01 ± 0.074238 ± 1002.27 ± 0.300.10 ± 0.20
6587583901886540288X-Sh330.281399−34.95426111.8−1.4+1.511.8^{+1.5}_{-1.4}182.03 ± 0.054235 ± 1002.27 ± 0.300.09 ± 0.20
6625431222418624896X-Sh335.874706−25.45175431.8−3.6+4.131.8^{+4.1}_{-3.6}−142.21 ± 0.114281 ± 1001.59 ± 0.30−0.59 ± 0.20
6625450601311209344MagE335.496690−25.31434554.3−7.7+3.754.3^{+3.7}_{-7.7}−120.88 ± 0.184527 ± 1000.14 ± 0.30−1.64 ± 0.20
6777274572785087360X-Sh316.510296−34.85923939.5−4.2+4.839.5^{+4.8}_{-4.2}−6.12 ± 0.174700 ± 1001.32 ± 0.30−1.32 ± 0.20
6824145855827406976X-Sh327.965006−19.98106329.4−3.9+4.329.4^{+4.3}_{-3.9}−223.60 ± 0.094533 ± 1001.50 ± 0.30−1.00 ± 0.20
6887945228483051392X-Sh312.819472−13.06086073.0−8.2+7.673.0^{+7.6}_{-8.2}−148.93 ± 0.104474 ± 1000.78 ± 0.30−0.99 ± 0.20
Gaia DR3 source IDInst.RADec.ddvhelv_{\mathrm{hel}}TeffT_{\mathrm{eff}}log⁡g\log g[Fe/H]
[deg][deg][kpc][km s−1^{-1}][K][dex][dex]
6897786884048378112X-Sh320.247233−7.64341315.5−1.9+2.315.5^{+2.3}_{-1.9}−248.45 ± 0.134866 ± 1002.29 ± 0.30−0.67 ± 0.20
68981498302595054083X-Sh321.745523−7.24119835.9−4.7+5.735.9^{+5.7}_{-4.7}−151.28 ± 0.074892 ± 1001.50 ± 0.30−1.05 ± 0.20
69086296879011601923X-Sh315.091530−8.88506069.1−7.8+12.769.1^{+12.7}_{-7.8}−76.43 ± 0.144658 ± 1000.79 ± 0.30−1.50 ± 0.20
6917207283052448768X-Sh315.392877−1.03062035.4−4.4+4.935.4^{+4.9}_{-4.4}−288.75 ± 0.164792 ± 1001.56 ± 0.30−1.00 ± 0.20

Table 5. Coordinates and stellar parameters of the spectroscopic targets not retained as MS candidates.

Spectro-photometric distances

We obtained spectro-photometric distances by fitting MESA Isochrones & Stellar Tracks (MIST, A. Dotter [23]; J. Choi et al. [16]; B. Paxton et al. [66, 67, 68]) to the spectroscopic constraints on effective temperature, surface gravity, and metallicity together with the photometry. We used the DELVE gg, rr, ii, and zz bands; the VHS JJ and KSK_{\mathrm{S}} bands, with HH where available; and the Gaia GG, BP, and RP bands. This gives nine photometric bands per star, or ten where HH is measured. The DELVE uu band is not used because no extinction coefficient is available for it, so it cannot be de-reddened consistently with the others. We performed the fits with the misfit package, which uses nested sampling and MIST models interpolated with minimint (S. Koposov [43]). For details of the fitting, see Section 4.1.2 of M. Cavieres et al. [13]. The implementation is publicly available4.

The resulting posterior from mistfit includes mass, age, metallicity, distance, and extinction. We adopted a mass prior based on a Salpeter IMF (E. E. Salpeter [78]), with allowed masses between 0.1 and 15 M⊙M_\odot; a flat prior for [Fe/H] within the range allowed by MIST; a flat prior on distance between 1 and 200 kpc; and a uniform prior on extinction from zero to five times the extinction reported in D. J. Schlegel et al. [80], using the recalibration given in E. F. Schlafly & D. P. Finkbeiner [79] and the corresponding relative extinction coefficients.

We further restricted the stellar age to 4–13 Gyr, consistent with the old populations expected in the Magellanic system, and sampled the isochrones in equivalent evolutionary phase rather than in mass and age, which covers the giant branch more uniformly. The Gaia parallax is evaluated in the likelihood rather than used to impose a prior, so that stars with low-significance or negative parallaxes are retained without truncation. An example posterior is shown in Figure 12.

MISfit posterior for Gaia DR3 6503888084644067712

Figure 12. Example misfit posterior for the MS candidate Gaia DR3 6503888084644067712. The five parameters shown are stellar mass, age, metallicity, heliocentric distance and extinction; panels below the diagonal show the joint distributions, with contours enclosing the 1σ1\sigma, 2σ2\sigma and 3σ3\sigma levels, and the histograms on the diagonal show the marginal distributions. Red lines mark the medians. Quoted values are the medians with the 16th and 84th percentiles: M⋆=0.87−0.06+0.10 M⊙M_\star= 0.87^{+0.10}_{-0.06}\,M_\odot, log⁡10(age/yr)=9.97−0.16+0.11\log_{10}(\mathrm{age}/\mathrm{yr}) = 9.97^{+0.11}_{-0.16}, [Fe/H] =−1.40±0.09= -1.40 \pm0.09, d⊙=57.9−4.2+3.2 kpcd_\odot= 57.9^{+3.2}_{-4.2}\,\mathrm{kpc} and E(B−V)=0.067−0.038+0.043E(B - V) = 0.067^{+0.043}_{-0.038}. The fit combines the spectroscopic TeffT_{\mathrm{eff}}, log⁡g\log g and [Fe/H] with nine photometric bands. The narrow anticorrelation between mass and age is the usual degeneracy along the giant branch, where a more massive star reaches the same luminosity at a younger age. The distance is essentially independent of it (r=0.05r = 0.05 against mass), which is why the spectro-photometric distance is well determined even though the age is not; the distance instead correlates with the extinction (r=−0.70r = -0.70), and this pair sets the achievable precision.

Results

We use the spectroscopic sample of 51 stars described in § 4 to identify stars consistent with the stellar MS and compare them with simulations and previous samples.

Using the full six-dimensional phase-space information for these stars, we compute their orbital angular momenta to assess their association with the Magellanic Clouds, presented in Figure 4, with error bars giving the 16th and 84th percentiles of the angular-momentum posteriors from 1000 samples over the proper-motion, distance and radial-velocity uncertainties. Following V. Chandra et al. [15], we use this angular-momentum space as an initial means of distinguishing potential MS members from MW halo stars. Figure 4 shows the angular momentum about the Galactocentric xx-axis, LxL_x, as a function of the angular momentum about the zz-axis, LzL_z, for our targets, together with stellar particles from B+12 Models 1 and 2. To isolate the simulated stellar MS from the main bodies of the Clouds, we mask particles with Magellanic Stream longitudes −20∘≤LMS≤0∘-20^\circ\leq L_{\mathrm{MS}} \leq0^\circ, retaining those with LMS<−20∘L_{\mathrm{MS}} < -20^\circ or LMS>0∘L_{\mathrm{MS}} > 0^\circ. The corresponding angular-momentum distribution for a representative sample of MW halo stars is constructed using stars from the DESI Milky Way Survey (S. E. Koposov et al. [45]) and is represented by the 1σ1\sigma and 2σ2\sigma contours in Figure 4. The MW halo is concentrated primarily at Lx≳−4×103L_x \gtrsim-4 \times10^3 kpc km s−1^{-1}. Eight stars pass our adopted central-value band, −45×103≤Lx≤−4×103-45 \times10^3 \leq L_x \leq-4 \times10^3 kpc km s−1^{-1}: two of the 33 X-shooter targets and six of the 18 MagE targets. Three further MagE targets have LxL_x below the lower bound and lie off the plotted panel.

Angular momentum of the spectroscopic samples in the Galactocentric x and z directions

Figure 4. Angular momentum of the spectroscopic samples in the Galactocentric xx and zz directions. Purple bars mark the 33 X-shooter targets and purple diamonds the 18 MagE targets. Pink circles show the Chandra sample (V. Chandra et al. [15]). Gray and light blue points are SMC star particles from the two B+12 models, with the Cloud main bodies removed. Orange contours trace the number density of DESI halo stars. The dashed lines bound the adopted central-value selection, −45×103≤Lx≤−4×103-45 \times10^3 \leq L_x \leq-4 \times10^3 kpc km s−1^{-1}. The inset enlarges the boxed region around Lz=0L_z = 0 near the upper selection boundary, using the same axis units as the main panel. Three MagE targets below the lower bound fall outside the plotted range. Eight stars pass this central-value band.

Three of the six MagE stars within this band (Gaia DR3 2426397007431829760, 2356964085088274688, and 4993221054391991040) have VLSRV_{\mathrm{LSR}} values that deviate by more than 100 km s−1^{-1} from the nearest local median velocity of the two B+12 models, evaluated within 3∘3^\circ in LMSL_{\mathrm{MS}} (Figure 5). In contrast, 12 of the 13 stars from V. Chandra et al. [15] lie within 100 km s−1^{-1} of the models, with the remaining star (Gaia DR3 2661637550258741376) also being a likely kinematic outlier. We therefore exclude the three MagE outliers from our stellar MS sample, while retaining the V. Chandra et al. [15] outlier for comparison. Sections 5.2 and 5.3 present further evidence against their MS membership. Notably, the 16th–84th percentile intervals of their LxL_x posteriors all cross the selection boundary, highlighting the ambiguity of a selection based solely on central values. However, requiring the full LxL_x posterior to satisfy the selection criterion would also exclude some of our five MS candidates. The radial velocities therefore provide the additional discrimination needed to identify these outliers.

Magellanic latitude, proper motions, and LSR radial velocity versus Magellanic longitude

Figure 5. Magellanic latitude, proper motions, and LSR radial velocity versus Magellanic longitude. Purple stars are the five MS candidates, blue-gray crosses the three radial-velocity outliers, and gray triangles the other 43 targets. Pink circles show the Chandra sample ([15]); gray and light-blue points are B+12 Models 1 and 2. Dashed segments connect each outlier to the nearer model median velocity within 3∘3^\circ in LMSL_{\mathrm{MS}}.

We retain five stars, two from X-shooter and three from MagE, which we refer to hereafter as the MS candidates. Together with the thirteen stars in the Chandra sample, they form the combined MS sample of 18 stars. The five MS candidates have spectro-photometric distances spanning 58–95 kpc, and their metallicities span −1.76≤[Fe/H]≤−1.16-1.76 \leq[\mathrm{Fe}/\mathrm{H}] \leq-1.16. The individual measurements of the five MS candidates are listed in Table 4. The other 46 analyzed stars, including the three velocity outliers, are listed in Table 5.

Because the LxL_x selection partly constrains proper motion, the agreement of the MS candidates with the model trends in the middle panels of Figure 5 is expected. It should therefore not be considered as independent evidence of membership, but rather a sanity check.

The spectro-photometric distances of the five MS candidates span 58–95 kpc. Their distance distribution is compared with the Cloud orbits in Figure 8. Our MagE target selection extends to 160 kpc by construction (§ 4.3), so the absence of MS candidates beyond ∼95\sim95 kpc is not imposed by the upper distance limit of that selection. With five stars, however, the full distance extent of the stellar MS remains uncertain. The 40 kpc lower bound used for target selection also prevents this sample from testing any nearer component predicted by S. Lucchini et al. [49] and S. Mishra et al. [57].

Magellanic longitude versus heliocentric distance

Figure 8. Magellanic longitude versus heliocentric distance. Gray and light-blue points are the stellar particles of B+12 Models 1 and 2, with their SMC center-of-mass orbits as the solid gray and dash-dotted light-blue curves. Given their origin as a tidal tail, the simulated stellar particles must reside at larger distances than the center-of-mass orbit of the progenitor SMC. The two orange curves show the SMC’s trajectory over the past Gyr in the Model C framework of B+K 2026 for a 1012 M⊙10^{12}\,M_\odot MW halo: solid for the H I-center and dashed for the photometric-center orbit solutions. The Chandra sample ([15]) is shown as pink circles and the five MS candidates of this work are purple stars. At the MS candidates’ longitudes, the B+12 simulated debris lies farther from the Sun than its corresponding SMC orbit. The H I-center and photometric-center orbits are less eccentric than the B+12 orbits, indicating that tidal debris from an SMC on the new orbits should reside at lower distances than the B+12 simulations, which adopted higher speeds for the Clouds than currently expected.

The five MS candidates span −22∘<BMS<+24∘-22^\circ< B_{\mathrm{MS}} < +24^\circ (Figure 5 and the left panel of Figure 6), with a median ∣BMS∣|B_{\mathrm{MS}}| of 20∘20^\circ, compared with 55–7∘7^\circ for the two B+12 models. All thirteen stars in the Chandra sample lie at negative BMSB_{\mathrm{MS}}. Together, the 18 stars of the combined MS sample span −37∘<BMS<+24∘-37^\circ< B_{\mathrm{MS}} < +24^\circ with a median ∣BMS∣|B_{\mathrm{MS}}| of 19∘19^\circ. The combined MS sample therefore occupies a qualitative broader band than either model predicts.

Figure 6. Left: Magellanic-latitude distributions of the two B+12 models (gray and light blue), the Chandra sample (pink line, V. Chandra et al. [15]), and the five MS candidates of this work (purple). The combined MS sample spans −37∘<BMS<+24∘-37^\circ< B_{\mathrm{MS}} < +24^\circ with median ∣BMS∣=19∘|B_{\mathrm{MS}}| = 19^\circ, compared with 55–7∘7^\circ for the simulations. Right: Normalized metallicity distribution function of the five MS candidates (purple), pooling two X-shooter and three MagE stars. The black histograms show all 33 X-shooter targets (solid) and the 15 MagE targets after omitting the three radial-velocity outliers (dashed). The SMC (teal, dot-dashed) and LMC (sage, solid) distributions from C. Muñoz et al. [59]. Every distribution is normalized to a peak of unity, so that samples differing by orders of magnitude in size remain comparable in shape.

Metallicities

Metallicity was not used in the final candidate selection and therefore provides an independent consistency check. The [Fe/H][\mathrm{Fe}/\mathrm{H}] distribution of the five MS candidates is shown in the right panel of Figure 6, alongside the LMC and SMC distributions from C. Muñoz et al. [59]. Their median metallicity is [Fe/H]=−1.59[\mathrm{Fe}/\mathrm{H}] = -1.59, and they span −1.76≤[Fe/H]≤−1.16-1.76 \leq[\mathrm{Fe}/\mathrm{H}] \leq-1.16. They are more

Note— X-Sh denotes X-shooter. The positions are the Gaia catalog coordinates used in the orbit calculations. Distances are posterior medians, with the 16th and 84th percentiles shown as asymmetric uncertainties. The quoted TeffT_{\mathrm{eff}} uncertainties include a systematic floor of 100 K added in quadrature to the formal uncertainties.

metal-poor than the peak of the SMC distribution at [Fe/H] ≃−1.06\simeq-1.06 and the LMC peak at [Fe/H] ≃−0.71\simeq-0.71. Such low metallicities are compatible with material stripped from the outer parts of the SMC. The SMC has a negative radial metallicity gradient, measured at −0.075±0.011-0.075 \pm0.011 dex deg−1^{-1} over its inner 5∘5^\circ from Ca ii-triplet spectroscopy of red giants (P. D. Dobbie et al. [21]). The SMC’s outer population is thus expected to be more metal-poor than the central population that dominates its distribution plotted in Figure 6. This makes the metallicities of the MS candidates consistent with an origin in the SMC outskirts (C. Muñoz et al. [59]), without imposing this metallicity restriction a priori.

The MS candidates also fall within the metallicity range of the MW halo, so their metallicities alone do not establish membership in the Stream. Five stars are insufficient to distinguish an SMC from an LMC origin, especially as the LMC outskirts can also be more metal-poor than its center. We identified no trend between [Fe/H] and position along the MS.

Velocity gradients

Once stars are tidally stripped, small offsets in their orbital energies and actions relative to the progenitor produce corresponding offsets in orbital frequency. These differences cause the debris from the Clouds to spread into leading and trailing tails. Consequently, within a dynamically young, unwrapped tidal arm, stars farther along the stream from the progenitor were generally stripped earlier, whereas recently stripped stars remain closer to the progenitor in phase space; this ordering nevertheless has intrinsic scatter (J. Bovy [11]; M. A. Fardal et al. [27]). This phase-space sorting produces coherent position–velocity gradients along tidal tails, including correlations between the component of stellar velocity parallel to the tail and distance from the progenitor (J. Klimentowski et al. [42]). Older stream segments have also had longer to respond to gravitational perturbations, although genuine velocity-space diffusion requires cumulative gravitational encounters (M. S. Delos & F. Schmidt [19]). We test for a related trend across the combined MS sample using each star’s present-day three-dimensional separation from the SMC and its relative velocity projected along the SMC’s direction of motion. These quantities are proxies for distance and velocity along a tidal tail, so a correlation would be suggestive rather than a unique signature of SMC stripping (J. D. Diaz & K. Bekki [20]; P. Zivick et al. [92]; V. Chandra et al. [15]).

Figure 7. Velocity relative to the SMC, projected along its direction of motion, against present-day distance from the SMC. Purple stars are the five MS candidates of this work; open gray circles are the other 46 analyzed targets, including the three radial-velocity outliers. Pink circles show the Chandra sample (V. Chandra et al. [15]). The dashed line is a least-squares fit to the 18 stars of the combined MS sample, and the gray band is its 1σ1\sigma bootstrap interval.

For this velocity-gradient calculation, we adopt the present-day SMC position and velocity from the photometric-center orbit solution of G. Besla & N. Kallivayalil [6], described in § 5.3. Let (Ri,vi)(R_i, v_i) be the Galactocentric position and velocity vectors of the ithi^{\mathrm{th}} star. Let (RSMC,vSMC)(R_{\mathrm{SMC}}, v_{\mathrm{SMC}}) be the present-day Galactocentric position and velocity vectors of the SMC. The relative position and velocity vectors of the star with respect to the SMC are given by:

Ri/SMC=Ri−RSMC(2)R_{i/\mathrm{SMC}} = R_i - R_{\mathrm{SMC}} \tag*{(2)}

and

vi/SMC=vi−vSMC(3)v_{i/\mathrm{SMC}} = v_i - v_{\mathrm{SMC}} \tag*{(3)}

Let vi/SMCprojv^{\mathrm{proj}}_{i/\mathrm{SMC}} be the component of the star’s velocity relative to the SMC, projected along the SMC’s direction of motion. We have:

vi/SMCproj=vi/SMC⋅vSMC∥vSMC∥(4)v^{\mathrm{proj}}_{i/\mathrm{SMC}} = v_{i/\mathrm{SMC}} \cdot\frac{v_{\mathrm{SMC}}}{\lVert v_{\mathrm{SMC}}\rVert} \tag*{(4)}

Figure 7 shows vi/SMCprojv^{\mathrm{proj}}_{i/\mathrm{SMC}} plotted against ∥Ri/SMC∥\lVert R_{i/\mathrm{SMC}}\rVert. A linear fit to the 18 stars of the combined MS sample gives a gradient of −1.82±0.66-1.82 \pm0.66 km s−1^{-1} kpc−1^{-1}, with Spearman ρ=−0.61\rho=-0.61 (p=7.5×10−3p=7.5\times10^{-3}; Pearson r=−0.56r=-0.56, p=0.015p=0.015). The five MS candidates alone also have a negative fitted gradient, −3.47±0.95-3.47\pm0.95 km s−1^{-1} kpc−1^{-1}, but this estimate is sensitive to individual points. This trend is compatible with position–velocity sorting of tidal debris; it does not independently establish membership. The other 46 analyzed targets are shown as open circles, including the three stars rejected by the line-of-sight velocity comparison despite passing the LxL_x cut.

Orbits

Figure 8 plots the observed heliocentric distances of the MS candidates and the Chandra sample (V. Chandra et al. [15]). Despite using a completely distinct data set and selection method, we find that our MS candidates reside at distances <100<100 kpc, even though our MagE target selection extends to 160 kpc. This further illustrates the significance of even five stars for understanding the properties of this system.

Figure 8 also shows the two B+12 stellar-debris models and the corresponding SMC center-of-mass orbits, together with the updated SMC trajectories from the recent review by G. Besla & N. Kallivayalil [6], hereafter B+K 2026. The B+K 2026 orbits are computed using their Model C moving center-of-mass framework with the H I-center velocities (N. Kallivayalil et al. [41]; P. Zivick et al. [91]) and photometric-center velocities of Gaia Collaboration (2023) (see their Table 2). The adopted MW model has a 1012M⊙10^{12}M_\odot NFW halo, disk, and bulge (the MW1 model in G. Besla & N. Kallivayalil [6]); the LMC model has a 1.8×1011M⊙1.8\times10^{11}M_\odot NFW halo and disk; and the SMC model has a 5×109M⊙5\times10^{9}M_\odot NFW halo. These calculations include dynamical friction from the MW on both Clouds and from the LMC on the SMC (see their Appendix F). The H I-center orbits are visualized in a Mollweide projection in Figure 9; the photometric-center orbits are similar in this projection. The distances to the B+12 simulated stellar debris are larger than those of most MS candidates. Interestingly, the B+K 2026 orbits are less eccentric than the B+12 orbits, which were based on higher speeds from the earlier measurements of N. Kallivayalil et al. [40]. Over the longitudes occupied by our MS candidates, the stellar particles in each B+12 simulation lie farther from the Sun than the SMC orbit that generated them; the orbit is therefore not itself the predicted debris track. The stellar MS (and tidal tails in general) is expected to lie at larger distances than the SMC’s orbit. We can use this dynamical fact to estimate the expected distances of MS stars removed from the SMC along the updated orbits from B+K 2026.

Mollweide projection in Galactic coordinates of the five MS candidates over the H I emission of the Magellanic system

Figure 9. Mollweide projection in Galactic coordinates of the five MS candidates over the H I emission of the Magellanic system. Thick purple tracks use their measured phase-space coordinates, faint tracks sample their uncertainties, and purple stars mark their present-day positions. The three radial-velocity outliers are excluded. Dashed green and orange lines show the B+K 2026 H I-center orbits of the LMC and SMC, respectively. The H I map is from T. Westmeier [87], and the grayscale background is the Gaia EDR3 all-sky flux map (ESA/Gaia/DPAC; A. Moitinho).

Around LMS≃−80∘L_{\mathrm{MS}} \simeq-80^\circ to −87∘-87^\circ, the median B+12 Model 1 debris is about 18–22 kpc farther away than its SMC orbit. The updated H I-center and photometric-center orbits lie at about 68–72 and 66–68 kpc, respectively, over the same LMSL_{\mathrm{MS}} range. Assuming a comparable debris–orbit separation based on B+12 Model 1, the expected distance of debris from an SMC following either new orbit is about 84–94 kpc. This estimated distance range is consistent with the two candidates at 90 and 95 kpc. In contrast, the original B+12 Model 1 debris medians are about 111–117 kpc at those longitudes. We therefore anticipate that full NN-body simulations using the B+K 2026 orbits will achieve better agreement with the observed distances (H. Rathore et al. 2027, in prep.).

We next attempt to confirm whether the MS candidates could have realistically originated from either the LMC or SMC by computing their trajectories in a time-evolving potential. Using the H I-center velocities and corresponding trajectories for the MW, LMC, and SMC (described above), we constructed a time-dependent gravitational potential with AGAMA (E. Vasiliev [84]). For the test-particle orbit integrations, the LMC and SMC are represented as moving, rigid Plummer potentials with masses of 1.8×1011 M⊙1.8 \times10^{11}\,M_\odot and 5×109 M⊙5 \times10^9\,M_\odot, and scale radii of 18 and 5 kpc, respectively. The MW is represented using the MilkyWayPotential2022 model of gala (A. M. Price-Whelan [71]), which comprises spherical nuclear and bulge components, three Miyamoto–Nagai disk components (M. Miyamoto & R. Nagai [58]) that together approximate an exponential disk, and an NFW dark-matter halo (J. F. Navarro et al. [61]). This model is calibrated to reproduce both the MW rotation curve measured by A.-C. Eilers et al. [26] and the vertical structure inferred from the solar-neighborhood phase spiral by E. Darragh-Ford et al. [18].

The B+K 2026 orbits provide the inertial phase-space coordinates of the centers of the three galaxies. We transformed the LMC and SMC orbits to a reference frame centered on the moving MW (i.e., non-inertial) by subtracting the interpolated position and velocity of the MW from each satellite at each time step. Because the stellar orbits were integrated in this accelerating Galactocentric frame, we included a spatially uniform, time-dependent fictitious acceleration associated with the motion of the MW. This term was calculated from the MW velocity trajectory as

afictitious(t)=−dvMWdt.(5)a_{\mathrm{fictitious}}(t)=-\frac{\mathrm{d}\mathbf{v}_{\mathrm{MW}}}{\mathrm{d}t}. \tag*{(5)}

The MW potential was held fixed at the origin of this reference frame, while the LMC and SMC potentials followed their prescribed MW-relative trajectories. The MS candidates were treated as massless test particles and therefore did not alter the trajectories or gravitational potentials of the three galaxies.

To initialize the orbit integrations, we transformed the sky positions, heliocentric distances, proper motions, and radial velocities of the MS candidates to the Galactocentric reference frame using Astropy (Astropy Collaboration et al. [2, 3, 4]). We assumed a distance to Sgr A* of 8.122 kpc, a solar Galactocentric velocity of (12.9,245.6,7.78)(12.9, 245.6, 7.78) km s−1^{-1}, and a solar height of 20.8 pc above the Galactic mid-plane for all coordinate transformations. Figure 9 shows the resulting stellar orbits for the first 300 Myr on the same time grid as the LMC and SMC orbits. In projection, the five tracks cross the region occupied by the gaseous trailing Stream. This spatial overlap is consistent with, but does not prove, a Magellanic origin.

The H i background in Figure 9 is based on the all-sky high-velocity-cloud column-density map of T. Westmeier [87], derived from the HI4PI 21-cm survey (HI4PI Collaboration et al. [37]). Although Galactic disk emission is excluded by the velocity selection, the map contains non-Magellanic high-velocity complexes. We therefore isolated the Magellanic system morphologically in the Magellanic coordinate frame of D. L. Nidever et al. [63]. After applying a threshold of log⁡10(NHI/cm−2)=18\log_{10}(N_{\mathrm{HI}}/\mathrm{cm}^{-2})=18, we retained connected components larger than 35 deg2^2 with centroids within ∣BMS∣<30∘\lvert B_{\mathrm{MS}}\rvert< 30^\circ and −140∘<LMS<90∘-140^\circ< L_{\mathrm{MS}} < 90^\circ. This selection recovers the Clouds, Bridge, trailing Stream, and Leading Arm while excluding unrelated high-velocity H i clouds for the purpose of visualization.

To test whether the MS candidates could have passed near either Cloud, we integrated 10,000 orbit realizations per star backward for approximately 2 Gyr, sampling their proper-motion, radial-velocity, and distance uncertainties. For these Monte Carlo integrations, we adopt the H i-center orbits of the LMC and SMC from B+K 2026 in the time-dependent potential described above. Figure 10 and Figure 11 show the resolved closest-approach locations relative to the LMC (blue) and SMC (orange), omitting realizations whose minimum separation occurs at an integration boundary. The probabilities quoted below are the fractions of all 10,000 realizations that enter the corresponding scale radius during the integration.

Figure 10. Posterior distributions of pericenter locations relative to the LMC center (blue) and SMC center (orange) for the two X-shooter MS candidates. The 1σ1\sigma and 2σ2\sigma contours are shown as solid and dashed lines, respectively. They are calculated from the resolved pericenter samples from 10,000 orbit realizations per star, with Gaussian smoothing of width 1.8 histogram bins. Realizations with a minimum separation at an integration boundary are excluded. The LMC and SMC Plummer scale radii are shown as blue and orange dotted circles, respectively [65].

Figure 11. As Figure 10, for the three MagE MS candidates. The three radial-velocity outliers are omitted.

Gaia DR3 6503888084644067712 and 4716496865562228736 each have more than 99% probability of passing inside the LMC’s 18 kpc Plummer scale radius; Gaia DR3 2441369916261698304 has about 34%. The former two have probabilities of about 14% and 23%, respectively, of passing inside the SMC’s 5 kpc scale radius. Gaia DR3 2622351759280470528 has about 3% probability of passing inside the LMC scale radius, while none of the realizations for Gaia DR3 4849744515068988544 enter it. These two candidates have median resolved closest approaches roughly 28–40 kpc from the Clouds; many Gaia DR3 4849744515068988544 realizations have no resolved minimum within the integration window.

These backward integrations depend on the adopted Cloud trajectories and do not uniquely establish an origin for any one star. In particular, the rejected star Gaia DR3 4993221054391991040 also has a close LMC passage in this potential, but its line-of-sight velocity is discrepant from both local Stream-model medians and its orbit reaches the inner MW.

This analysis builds confidence that our MS candidates did indeed originate in either the LMC or SMC. We acknowledge that these integrations treat the Clouds as rigid potentials and therefore do not capture deformations of their dark-matter distributions during mutual interactions, which can affect backward orbit integrations (H. R. Foote et al. [29]).

Conclusions

We have searched for the stellar counterpart of the MS using distant K giants selected from DELVE DR2 and VHS DR5 photometry combined with Gaia DR3 astrometry. Our initial catalog contains 87 763 K-giant candidates across the Southern sky, from which we selected hundreds of stars with consistent proper motions and spatial distribution as the MS. From them, we analyzed spectra of 51 targets selected by their proper motions and magnitudes to align with the B+12 predictions for the MS. Eight pass the adopted central-value LxL_x band. Three of these have radial velocities that differ by more than 100 km s−1^{-1} from both local B+12 Stream-model median velocities, suggesting significantly more radial trajectories than expected for the Stream. We therefore exclude these three stars, leaving five stars that we refer to as the MS candidates. Together with the thirteen stars in the Chandra sample (V. Chandra et al. [15]), they form the combined MS sample of 18 stars.

Our main results are as follows:

  • The combined MS sample spans −37∘<BMS<+24∘-37^\circ< B_{\mathrm{MS}} < +24^\circ, with a median ∣BMS∣=19∘\lvert B_{\mathrm{MS}}\rvert= 19^\circ, compared with 55–7∘7^\circ for the B+12 models. Our five MS candidates

lie on both sides of the gaseous Stream, whereas the stars in the Chandra sample all lie at negative BMSB_{\mathrm{MS}}. The observed distribution is broader than the simulated stellar-debris distributions, favoring an extended tidal debris field rather than a narrow stream.

  • Our five MS candidates have a mean heliocentric distance of 76 kpc, a median of 82 kpc, and a range of 58–95 kpc; the combined MS sample has a mean of 80 kpc. Over the longitudes spanned by our MS candidates, debris in the B+12 simulations lies farther from the Sun than most observed candidates. However, the updated B+K 2026 orbits, based on HST/Gaia DR3 proper motions, indicate that the SMC’s orbit is less eccentric than in B+12. At LMS≃−80∘L_{\mathrm{MS}} \simeq-80^\circ to −87∘-87^\circ, the median debris in B+12 Model 1 lies 18–22 kpc farther from the Sun than its progenitor SMC orbit. Applying a comparable debris–orbit offset to the new B+K 2026 orbits suggests that an SMC tidal tail would lie at distances of 84–94 kpc at these longitudes, consistent with the observed candidate distances. The updated SMC trajectories are therefore more promising for explaining the distances of the MS candidates, although a new debris simulation is needed for a quantitative test.

  • The five MS candidates have −1.76≤[Fe/H]≤−1.16-1.76 \leq[\mathrm{Fe}/\mathrm{H}] \leq-1.16, with a median [Fe/H]=−1.59[\mathrm{Fe}/\mathrm{H}] = -1.59. These values are compatible with material stripped from the metal-poor outskirts of the SMC. Our original target selection was designed to include a population resembling the Stream, without preferentially targeting stars by metallicity or selecting the metal-poor tail.

  • The five MS candidates show a negative correlation between their present-day distance from the SMC and their relative velocity projected along the SMC’s direction of motion. The fit to the combined MS sample has a slope of −1.82±0.66-1.82 \pm0.66 km s−1^{-1} kpc−1^{-1}. This pattern is compatible with tidal debris, although the five-star fit is sensitive to individual points. We do not use this correlation as a separate membership criterion.

  • In the adopted time-dependent MW–LMC–SMC potential, the MS candidates have orbits consistent with an origin in either the LMC or the SMC, with pericenter passages around the Clouds within the integration window. These close approaches depend on the adopted model and do not unambiguously associate individual stars with either Cloud. Nevertheless, the analysis strengthens the evidence for a Magellanic origin.

Taken together, the kinematics and spatial distribution of the five MS candidates are consistent with a spatially extended stellar component associated with the Magellanic system. Their broad distribution across the sky challenges the narrow latitude distributions of the two B+12 models. The distance comparison favors exploring debris generated along the newer, closer Cloud trajectories derived from the latest HST/Gaia DR3 proper motions, while accounting for the offset between the progenitor orbit and stripped stars expected from tidal stripping.

Even a sample as small as five stars contributes to our understanding of the stellar MS because the final candidate selection imposes no metallicity cut and our MagE target selection extends to 160 kpc. The absence of MS candidates beyond ∼95\sim95 kpc is therefore not imposed by the upper distance limit of our selection. Larger samples are essential to determine the full extent of the Magellanic stellar debris and distinguish between different orbital and stripping histories. The catalog presented here contains many additional distant K-giant candidates suitable for spectroscopic follow-up, including those in regions beyond the trailing arm explored in this work.

Mapping the stellar counterpart to the MS will become increasingly feasible with upcoming large spectroscopic programs such as 4MOST ([36]) and Via ([82]), which will provide radial velocities and metallicities for large samples of distant halo stars. Improved astrometric precision in future Gaia data releases, particularly DR4, will reduce uncertainties in the transverse motions and orbital histories of these distant tracers. Combining this astrometry with wide-area spectroscopy will enable us to map the stellar component of the Magellanic system with substantially larger samples and place stronger constraints on the masses, past orbits, and interaction history of the Clouds.

Acknowledgments

MC acknowledges support from the European Research Council (ERC) grant 101002511 (VEGA-P). JC acknowledges support from the Agencia Nacional de Investigación y Desarrollo (ANID) via Proyecto Fondecyt Regular 1231345, and by ANID BASAL project FB210003. We acknowledge A Dictionary of Colour Combinations (ISBN 978-4-86152-247-5), which proved a valuable resource for selecting color palettes for figures, particularly for a color-blind astronomer. This work was performed using the ALICE compute resources provided by Leiden University. This work has made use of data from the European Space Agency (ESA) mission Gaia (https://www.cosmos.esa.int/gaia), processed by the Gaia Data Processing and Analysis Consortium (DPAC, https://www.cosmos.esa.int/web/gaia/dpac/consortium). Funding for the DPAC has been provided by national institutions, in particular the institutions participating in the Gaia Multilateral Agreement.

Software: Astropy ([2, 3, 4]), NumPy ([35]), SciPy ([86]), Matplotlib ([38]), pandas ([53]), AGAMA ([84]), gala ([71]), dynesty ([81]), minimint ([43]), rvspecfit ([44]), PyPeIt ([73, 72]), iSpec ([10, 9]), Telfit (K. Gullikson et al. 2014)

Supplementary figures and tables

Note— X-Sh denotes X-shooter. The positions are the Gaia catalog coordinates. Distances are posterior medians, with the 16th and 84th percentiles shown as asymmetric uncertainties. The log⁡g\log g and [Fe/H] uncertainties include adopted systematic floors of 0.30 and 0.20 dex, respectively, added in quadrature to the formal Hessian uncertainties. Forty-three targets do not satisfy the central-value LxL_x band. The other three (Gaia DR3 2426397007431829760, 2356964085088274688, and 4993221054391991040) pass that band but are excluded from the MS candidate sample by the additional radial-velocity check described in § 5.

References

  1. [1]Arentsen, A., Prugniel, P., Gonneau, A., et al. 2019, A&A, 627, doi: 10.1051/0004-6361/201834273DOI
  2. [2]Astropy Collaboration, Robitaille, T. P., Tollerud, E. J., et al. 2013, A&A, 558, A33, doi: 10.1051/0004-6361/201322068
  3. [3]Astropy Collaboration, Price-Whelan, A. M., Sipőcz, B. M., et al. 2018, AJ, 156, 123, doi: 10.3847/1538-3881/aabc4f
  4. [4]Astropy Collaboration, Price-Whelan, A. M., Lim, P. L., et al. 2022, ApJ, 935, 167, doi: 10.3847/1538-4357/ac7c74DOI
  5. [5]Besla, G., Hernquist, L., & Loeb, A. 2013, MNRAS, 428, 2342, doi: 10.1093/mnras/sts192DOI
  6. [6]Besla, G., & Kallivayalil, N. 2026, arXiv e-prints, arXiv:2609.38277, doi: 10.48550/arXiv.2609.38277DOI
  7. [7]Besla, G., Kallivayalil, N., Hernquist, L., et al. 2007, ApJ, 668, 949, doi: 10.1086/521385DOI
  8. [8]Besla, G., Kallivayalil, N., Hernquist, L., et al. 2012, MNRAS, 421, 2109, doi: 10.1111/j.1365-2966.2012.20466.xDOI
  9. [9]Blanco-Cuaresma, S. 2019, MNRAS, 486, 2075, doi: 10.1093/mnras/stz549
  10. [10]Blanco-Cuaresma, S., Soubiran, C., Heiter, U., & Jofré, P. 2014, A&A, 569, A111, doi: 10.1051/0004-6361/201423945
  11. [11]Bovy, J. 2014, The Astrophysical Journal, 795, doi: 10.1088/0004-637X/795/1/95DOI
  12. [12]Bressan, A., Marigo, P., Girardi, L., et al. 2012, MNRAS, 427, 127, doi: 10.1111/j.1365-2966.2012.21948.xDOI
  13. [13]Cavieres, M., Koposov, S. E., Rossi, E. M., Penoyre, Z., & Verberne, S. 2026, A&A, 709, A117, doi: 10.1051/0004-6361/202659104DOI
  14. [14]Cavieres, M., Chanamé, J., Navarrete, C., et al. 2025, ApJ, 983, 83, doi: 10.3847/1538-4357/adbf08DOI
  15. [15]Chandra, V., Naidu, R. P., Conroy, C., et al. 2023, ApJ, 956, 110, doi: 10.3847/1538-4357/acf7bfDOI
  16. [16]Choi, J., Dotter, A., Conroy, C., et al. 2016, ApJ, 823, 102, doi: 10.3847/0004-637X/823/2/102DOI
  17. [17]Conroy, C., Naidu, R. P., Garavito-Camargo, N., et al. 2021, Nature, 592, 534, doi: 10.1038/s41586-021-03385-7DOI
  18. [18]Darragh-Ford, E., Hunt, J. A. S., Price-Whelan, A. M., & Johnston, K. V. 2023, ApJ, 955, 74, doi: 10.3847/1538-4357/acf1fc
  19. [19]Delos, M. S., & Schmidt, F. 2022, Monthly Notices of the Royal Astronomical Society, 513, 3682, doi: 10.1093/mnras/stac1022DOI
  20. [20]Diaz, J. D., & Bekki, K. 2012, ApJ, 750, 36, doi: 10.1088/0004-637X/750/1/36DOI
  21. [21]Dobbie, P. D., Cole, A. A., Subramaniam, A., & Keller, S. 2014, MNRAS, 442, 1680, doi: 10.1093/mnras/stu926DOI
  22. [22]D’Onghia, E., & Fox, A. J. 2016, ARA&A, 54, 363, doi: 10.1146/annurev-astro-081915-023251
  23. [23]Dotter, A. 2016, ApJS, 222, 8, doi: 10.3847/0067-0049/222/1/8DOI
  24. [24]Drlica-Wagner, A., Ferguson, P. S., Adamów, M., et al. 2022, ApJS, 261, 38, doi: 10.3847/1538-4365/ac78eb
  25. [25]Drlica-Wagner, A., Carlin, J. L., Nidever, D. L., et al. 2021, ApJS, 256, 2, doi: 10.3847/1538-4365/ac079d
  26. [26]Eilers, A.-C., Hogg, D. W., Rix, H.-W., & Ness, M. K. 2019, ApJ, 871, 120, doi: 10.3847/1538-4357/aaf648DOI
  27. [27]Fardal, M. A., Huang, S., & Weinberg, M. D. 2015, Monthly Notices of the Royal Astronomical Society, 452, 301, doi: 10.1093/mnras/stv1198DOI
  28. [28]Foote, H. R., Besla, G., Mocz, P., et al. 2023, ApJ, 954, 163, doi: 10.3847/1538-4357/ace533DOI
  29. [29]Foote, H. R., Rathore, H., Besla, G., et al. 2026, ApJ, 1001, 37, doi: 10.3847/1538-4357/ae4a9dDOI
  30. [30]Fox, A. J., Wakker, B. P., Barger, K. A., Hernandez, A. K., et al. 2014, ApJ, 787, 147, doi: 10.1088/0004-637X/787/2/147DOI
  31. [32]Garavito-Camargo, N., Besla, G., Laporte, C. F. P., et al. 2019, ApJ, 884, 51, doi: 10.3847/1538-4357/ab32eb
  32. [33]Gardiner, L. T., & Noguchi, M. 1996, MNRAS, 278, 191, doi: 10.1093/mnras/278.1.191DOI
  33. [35]Harris, C. R., Millman, K. J., van der Walt, S. J., et al. 2020, Nature, 585, 357, doi: 10.1038/s41586-020-2649-2DOI
  34. [36]Helmi, A., Irwin, M., Deason, A., et al. 2019, The Messenger, 175, 23, doi: 10.18727/0722-6691/5120DOI
  35. [37]HI4PI Collaboration, Ben Bekhti, N., Flöer, L., et al. 2016, A&A, 594, A116, doi: 10.1051/0004-6361/201629178
  36. [38]Hunter, J. D. 2007, Computing in Science and Engineering, 9, 90, doi: 10.1109/MCSE.2007.55
  37. [39]Husser, T.-O., Wende-von Berg, S., Dreizler, S., et al. 2013, Astronomy & Astrophysics, 553, A6, doi: 10.1051/0004-6361/201219058DOI
  38. [40]Kallivayalil, N., van der Marel, R. P., Alcock, C., et al. 2006, ApJ, 638, 772, doi: 10.1086/498972DOI
  39. [41]Kallivayalil, N., van der Marel, R. P., Besla, G., Anderson, J., & Alcock, C. 2013, ApJ, 764, 161, doi: 10.1088/0004-637X/764/2/161DOI
  40. [42]Klimentowski, J., Lokas, E. L., Kazantzidis, S., et al. 2009, Monthly Notices of the Royal Astronomical Society, 400, 2162, doi: 10.1111/j.1365-2966.2009.15626.xDOI
  41. [43]Koposov, S. 2023, segasaj/minimint: Minimint v0.4.1, v0.4.1, Zenodo, Zenodo, doi: 10.5281/zenodo.10437032
  42. [44]Koposov, S. E. 2019, RVSpectFit: Radial velocity and stellar atmospheric parameter fitting, Astrophysics Source Code Library, record ascl:1907.013 http://ascl.net/1907.013
  43. [45]Koposov, S. E., Allende Prieto, C., Cooper, A. P., et al. 2024, MNRAS, 533, 1012, doi: 10.1093/mnras/stae1842DOI
  44. [46]Lançon, A., Gonneau, A., Verro, K., et al. 2021, A&A, 649, doi: 10.1051/0004-6361/202039371DOI
  45. [48]Lin, D. N. C., Jones, B. F., & Klemola, A. R. 1995, ApJ, 439, 652, doi: 10.1086/175205DOI
  46. [49]Lucchini, S., D’Onghia, E., & Fox, A. J. 2021, ApJL, 921, L36, doi: 10.3847/2041-8213/ac3338DOI
  47. [50]Lucchini, S., D’Onghia, E., Fox, A. J., et al. 2020, Nature, 585, 203, doi: 10.1038/s41586-020-2663-4DOI
  48. [51]Marshall, J. L., Burles, S., Thompson, I. B., et al. 2008, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 7014, Ground-based and Airborne Instrumentation for Astronomy II, ed. I. S. McLean & M. M. Casali, 701454, doi: 10.1117/12.789972
  49. [52]Mathewson, D. S., Cleary, M. N., & Murray, J. D. 1974, ApJ, 190, 291, doi: 10.1086/152875DOI
  50. [53]McKinney, W. 2010, in Proceedings of the 9th Python in Science Conference, ed. S. van der Walt & J. Millman, 56–61, doi: 10.25080/Majora-92bf1922-00a
  51. [54]McMahon, R. G., Banerji, M., Gonzalez, E., et al. 2021, VizieR Online Data Catalog: The VISTA Hemisphere Survey (VHS) catalog DR5 (McMahon+, 2020),, VizieR On-line Data Catalog: II/367. Originally published in: 2013Mysngr.154...35M
  52. [55]McMahon, R. G., Banerji, M., Gonzalez, E., et al. 2013, The Messenger, 154, 35
  53. [56]Mishra, S., Fox, A. J., Smoker, J. V., Lucchini, S., & D’Onghia, E. 2025a, ApJ, 984, 104, doi: 10.3847/1538-4357/adc68aDOI
  54. [57]Mishra, S., Fox, A. J., Smoker, J. V., Lucchini, S., & D’Onghia, E. 2025b, ApJ, 984, 104, doi: 10.3847/1538-4357/adc68aDOI
  55. [58]Miyamoto, M., & Nagai, R. 1975, PASJ, 27, 533
  56. [59]Muñoz, C., Monachesi, A., Nidever, D. L., Majewski, S. R., et al. 2023, A&A, 680, A79, doi: 10.1051/0004-6361/202347046DOI
  57. [60]Muñoz, R. P., Puzia, T. H., Lancon, A., et al. 2014, ApJS, 210, 4, doi: 10.1088/0067-0049/210/1/4DOI
  58. [61]Navarro, J. F., Frenk, C. S., & White, S. D. M. 1996, ApJ, 462, 563, doi: 10.1086/177173DOI
  59. [62]Nidever, D. L., Majewski, S. R., & Burton, W. B. 2008, ApJ, 679, 432, doi: 10.1086/587042DOI
  60. [63]Nidever, D. L., Majewski, S. R., Burton, W. B., & Nigra, L. 2010, ApJ, 723, 1618, doi: 10.1088/0004-637X/723/2/1618DOI
  61. [64]Ordenes-Briceno, Y., Puzia, T. H., Eigenthaler, P., et al. 2018, ApJ, 860, 4, doi: 10.3847/1538-4357/aac1b8
  62. [65]Patel, E., Kallivayalil, N., Garavito-Camargo, N., et al. 2020, The Astrophysical Journal, 893, 121, doi: 10.3847/1538-4357/ab7b75
  63. [66]Paxton, B., Bildsten, L., Dotter, A., et al. 2011, ApJS, 192, 3, doi: 10.1088/0067-0049/192/1/3DOI
  64. [67]Paxton, B., Cantiello, M., Arras, P., et al. 2013, ApJS, 208, 4, doi: 10.1088/0067-0049/208/1/4DOI
  65. [68]Paxton, B., Marchant, P., Schwab, J., et al. 2015, ApJS, 220, 15, doi: 10.1088/0067-0049/220/1/15DOI
  66. [69]Petersen, M. S., & Peñarrubia, J. 2020, MNRAS, 494, L11, doi: 10.1093/mnrasl/slaa029DOI
  67. [71]Price-Whelan, A. M. 2017, The Journal of Open Source Software, 2, 388, doi: 10.21105/joss.00388
  68. [72]Prochaska, Westfall, Cooke, et al. 2025, PyPelt, 1.17.3, Zenodo, v1.17.3 Zenodo, doi: 10.5281/zenodo.14983432
  69. [73]Prochaska, J., Hennawi, J., Westfall, K., et al. 2020, The Journal of Open Source Software, 5, 2308, doi: 10.21105/joss.02308
  70. [74]Putman, M. E., Staveley-Smith, L., Freeman, K. C., Gibson, B. K., & Barnes, D. G. 2003, ApJ, 586, 170, doi: 10.1086/344477DOI
  71. [75]Rathore, H., Besla, G., van der Marel, R. P., & Kallivayalil, N. 2026, The Astrophysical Journal, 1000, 50, doi: 10.3847/1538-4357/ae4507
  72. [76]Růžička, A., Theis, C., & Palouš, J. 2010, ApJ, 725, 369, doi: 10.1088/0004-637X/725/1/369DOI
  73. [77]Salem, M., Besla, G., Bryan, G., Putman, M., et al. 2015, ApJ, 815, 77, doi: 10.1088/0004-637X/815/1/77DOI
  74. [78]Salpeter, E. E. 1955, The Astrophysical Journal, 121, 161, doi: 10.1086/145927
  75. [79]Schlafly, E. F., & Finkbeiner, D. P. 2011, ApJ, 737, 103, doi: 10.1088/0004-637X/737/2/103DOI
  76. [80]Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525, doi: 10.1086/305772DOI
  77. [81]Speagle, J. S. 2020, MNRAS, 493, 3132, doi: 10.1093/mnras/staa278DOI
  78. [82]The Via Collaboration. 2026, arXiv e-prints, arXiv:2606.18332, doi: 10.48550/arXiv.2606.18332
  79. [83]Toomre, A., & Toomre, J. 1972, ApJ, 178, 623, doi: 10.1086/151823DOI
  80. [84]Vasiliev, E. 2019, MNRAS, 482, 1525, doi: 10.1093/mnras/sty2672DOI
  81. [85]Vernet, J., Dekker, H., D’Odorico, S., et al. 2011, A&A, 536, A105, doi: 10.1051/0004-6361/201117752DOI
  82. [86]Virtanen, P., Gommers, R., Oliphant, T. E., et al. 2020, Nature Methods, 17, 261, doi: 10.1038/s41592-019-0686-2
  83. [87]Westmeier, T. 2018, MNRAS, 474, 289, doi: 10.1093/mnras/stx2757DOI
  84. [88]Wright, E. L., Eisenhardt, P. R. M., Mainzer, A. K., Ressler, M. E., et al. 2010, AJ, 140, 1868, doi: 10.1088/0004-6256/140/6/1868DOI
  85. [89]Zaritsky, D., Conroy, C., Naidu, R. P., et al. 2020, ApJL, 905, L3, doi: 10.3847/2041-8213/abcb83DOI
  86. [90]Zhang, L., Xue, X.-X., Yang, C., et al. 2023, AJ, 165, 224, doi: 10.3847/1538-3881/acc9bbDOI
  87. [91]Zivick, P., Kallivayalil, N., van der Marel, R. P., et al. 2018, ApJ, 864, 55, doi: 10.3847/1538-4357/aad4b0DOI
  88. [92]Zivick, P., Kallivayalil, N., Besla, G., et al. 2019, The Astrophysical Journal, 874, doi: 10.3847/1538-4357/ab0554

Paper details

vlnc:W_7jsrwkkr
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 10

downloads · 1

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents