betaFeedback

WorksPhysics and Astronomy

Asteroseismology of AI Velorum: a prototype of high-amplitude delta Scuti stars

Abstract

AI Velorum is exceptional among high-amplitude δScuti (HADS) stars because of its multimode pulsations. Despite being an excellent target for asteroseismology, this prototype star has not been studied in detail. We determine the stellar parameters, age, and evolutionary stage of AI Vel, constrain envelope convection and convective core overshooting, and test three opacity data sets: OPAL, OP, and OPLIB. We analysed TESS photometry, ground-based Strömgren photometry, and HARPS spectroscopy. Modes were identified using photometric and spectroscopic methods, and empirical values of the non-adiabatic f-parameter were derived for both radial modes. Seismic modelling was performed within a Bayesian framework supported by Monte Carlo simulations. We confirm the radial nature of the two dominant modes and identify one non-radial mode as most likely an axisymmetric dipole or quadrupole mode. Within the explored parameter space, only OPAL models reproduce the observed position of AI Vel in the HR diagram, whereas OP and OPLIB models are too cool and underluminous. All seismic models favour inefficient envelope convection (mixing-length parameter alpha_{mlt}\lesssim 0.5). Including the non-radial mode substantially reduces the formal posterior uncertainties, but the ambiguity in its identification introduces additional model-dependent uncertainties. AI Vel is an evolved, slowly rotating star in the hydrogen shell-burning phase. Our results demonstrate the diagnostic power of combining pulsation frequencies with empirical f-parameters. The preference for OPAL opacities is consistent with results for several other HADS and SX Phoenicis stars but does not imply a general superiority of OPAL opacities.

Introduction

Pulsating stars of the δ\delta Scuti type have been known for decades and, in some cases, for more than a century. Notably, the variability of the prototype star, δ\delta Scuti itself, was first reported in 1900 by Campbell & Wright [27].

In modern terminology, δ\delta Scuti variables are defined as A0–F5 spectral-type stars [6, 83] with luminosity classes ranging from V to III. They exhibit short-period brightness variations caused by both radial and non-radial pulsations in pressure (pp), mixed, and gravity (gg) modes. Their typical masses lie between 1.5 and 2.5 M⊙M_\odot [1, 15], and their pulsation periods range from approximately 0.5 to 6 hours (Breger 2000b). The pulsations are driven by the κ\kappa-mechanism operating primarily in the second helium ionization zone [7, 8, 99, 29].

Classical δ\delta Scuti stars belong to Population I and are located predominantly on the main sequence, although they can also occur in more evolved evolutionary stages [26, 17, 69, 3] or during the pre-main-sequence phase [103, 105, 104]. A distinct subgroup, known as High-Amplitude δ\delta Scuti (HADS) stars or dwarf Cepheids, is characterized by the photometric range in the Johnson VV band exceeding 0.3 mag. These stars are usually more evolved, often during the hydrogen shell-burning phase, and typically exhibit slow rotation and pulsations dominated by one or two radial modes (Breger 2000b). Apart from their large amplitudes, however, HADS stars appear observationally similar to other δ\delta Scuti variables and do not show clear differences in their fundamental stellar parameters [19].

δ\delta Scuti stars follow a well-defined period–luminosity (P–L) relation [76, 77, 64, 100], which makes them useful distance indicators in stellar populations. For HADS stars the relation is even tighter, as their pulsation properties resemble those of classical Cepheids [78].

In recent years, asteroseismology has become a powerful tool for probing the internal structure of δ\delta Scuti stars. By analysing their pulsation spectra, it is possible to place strong constraints on stellar parameters such as mass, age, chemical composition, internal rotation, and the efficiency of convective energy transport (e.g. Breger [20, 21]; Bowman & Kurtz [16]; Antoci et al. [3]; Murphy et al. [69]; Barac et al. [9]; Panda et al. [74]; Guo et al. [55]; Lv et al. 2024; Daszynska-Daszkiewicz et al. [38]).

However, modelling δ\delta Scuti stars remains challenging due to several physical uncertainties, including the treatment of convection in the outer envelope, rotational effects, and the choice of opacity data.

One of the most promising diagnostics for constraining some processes is the complex non-adiabatic ff-parameter, defined as the ratio of the bolometric flux variation to the radial displacement at the stellar surface. The empirical determination of this parameter from multicolour photometry and radial velocity data provides a powerful test of stellar models, as the theoretical value of ff is highly sensitive to the efficiency of envelope convection and to the adopted opacity tables (Daszynska-Daszkiewicz et al. [34]). Previous studies have demonstrated that comparisons between empirical and theoretical ff values can provide valuable constraints on convective energy transport in the stellar envelope as well as on the opacity data used in calculations.

Space-based photometry has further transformed the study of δ\delta Scuti and HADS stars. In particular, Bedding et al. [11] revealed remarkably regular high-frequency pulsation patterns in young δ\delta Scuti stars, opening new possibilities for mode identification and seismic modelling. Space photometry has also shown that HADS stars, traditionally regarded as relatively simple radial pulsators, can exhibit rich spectra of low-amplitude non-radial modes (e.g., Poretti et al. [79]). Detailed seismic modelling of double-mode HADS stars has demonstrated that the combination of radial-mode frequencies with additional constraints, such as the non-adiabatic ff-parameter and low-amplitude non-radial modes, can provide tight constraints on stellar parameters, convection, and opacity data (e.g., Daszynska-Daszkiewicz et al. [40, 41]). More recently, Yang & Zuo [98] used TESS observations to investigate the pulsation properties and mode identification of a larger sample of HADS stars, further demonstrating the potential of high-precision space photometry for studying this class of pulsators.

In this context, AI Velorum is an especially valuable target. As the prototype of high-amplitude δ\delta Scuti stars, it exhibits large-amplitude pulsations dominated by two radial modes, making mode identification relatively secure. At the same time, the presence of additional non-radial modes provides further seismic constraints. These properties make AI Velorum an ideal target for testing stellar structure and pulsation models.

In this work, we present a comprehensive asteroseismic analysis of AI Velorum using the most recent observational data available for this star. We combine TESS photometry, ground-based multicolour observations, and high-resolution spectroscopy with modern stellar evolution and pulsation models. The modelling is carried out within a Bayesian framework supported by Monte Carlo simulations, enabling us to derive precise stellar parameters and investigate the effects of opacity data and convective efficiency on models of δ\delta Scuti stars.

The paper is organized as follows. In Section 2, we summarize the basic properties of AI Velorum and derive its fundamental parameters from spectroscopic analysis. Section 3 describes the Fourier analysis of the available TESS data for AI Velorum. In Section 4, we present the results of mode identification. The asteroseismic modelling is described in Section 5. A summary and discussion are given in the final section.

AI Velorum

Previous studies and basic properties

AI Velorum (AI Vel, HD 69213) is the prototype of a class of HADS stars (Walraven et al. [97]). Its visual magnitude in the Johnson VV passband is 6.98 mag (Ducati [43]). The spectral classification of AI Vel has changed over the years. It was classified as A4 at maximum brightness in the VV filter and F2 at minimum brightness (Herbig [57]). Later, it was classified as A1p at maximum luminosity and F1p at minimum luminosity (Gratton & Lavnagino [52]). In the General Catalogue of Variable Stars (GCVS) the range of spectral types is given as A2p-F2pV/V (Samus et al. [87]).

The first reports of AI Vel’s pulsational variability were made by Walraven [95, 96], who described the light variations by two periods: P0=0.111574P_{0} = 0.111574 and P1=0.0862073P_{1} = 0.0862073 d. Walraven [95] also suspected the presence of additional periodicities in the photometric data, specifically at P2=0.046460P_{2} = 0.046460 d and P3=0.0317584P_{3} = 0.0317584 d, although he was unable to confirm their significance. Forty years later, Walraven et al. [97] confirmed the two dominant pulsation frequencies, 8.962608.96260 d−1^{-1} (P0P_{0}) and 11.59977911.599779 d−1^{-1} (P1P_{1}) and reported additional modes based on data collected over a 38-year timespan. These included 22.52182622.521826 d−1^{-1}, 15.97247915.972479 d−1^{-1}, 9.137619.13761 d−1^{-1} and 10.9300010.93000 d−1^{-1}.

The parallax of AI Vel, π=9.970±0.019\pi= 9.970 \pm0.019 mas, derived from Gaia DR3 data, corresponds to a distance of 100.40±0.19100.40 \pm0.19 pc. This value is in excellent agreement with the Hipparcos measurement of π=9.99±0.53\pi= 9.99 \pm0.53 mas (ESA 1997). The proper motion measured by Gaia is 26.222±0.01926.222 \pm0.019 mas yr−1^{-1} in right ascension and 31.094±0.02531.094 \pm0.025 mas yr−1^{-1} in declination. The brightness of the star in the Gaia GG band is 6.518 mag (Gaia Collaboration [48]).

The effective temperature derived from uvbyuvby photometry (Breger [18]) ranges between 7110 and 8520 K. Based on Gaia BP/RP spectra, the effective temperature is Teff=7188−12+14T_{\mathrm{eff}} = 7188^{+14}_{-12} K, the surface gravity is log⁡g=3.662−0.03+0.05\log g = 3.662^{+0.05}_{-0.03}, and the metallicity is [Fe/H]=−1.330−0.002+0.003[\mathrm{Fe}/\mathrm{H}] = -1.330^{+0.003}_{-0.002}. The extinction is estimated as AV=0.059−0.006+0.007A_{V} = 0.059^{+0.007}_{-0.006} mag in the Johnson VV band and AG=0.053−0.005+0.006A_{G} = 0.053^{+0.006}_{-0.005} mag in the Gaia GG band. The radial velocity is Vrad=17±3V_{\mathrm{rad}} = 17 \pm3 km s−1^{-1}. The Renormalized Unit Weight Error (RUWE) from Gaia DR3 is 0.940, indicating high-quality astrometric data (Gaia Collaboration [49]). However, the Gaia metallicity estimate is likely biased due to pulsation-induced spectral variability and low-resolution BP/RP spectra.

The reddening estimated from Hipparcos data is E(B−V)=0.10E(B-V) = 0.10, which, assuming a total-to-selective extinction ratio of RV=3.15R_{V} = 3.15, corresponds to AV=0.32A_{V} = 0.32 mag in the Johnson VV band (Gontcharov & Mosenkov [51]).

In the StarHorse catalogue (Anders et al. [2]), the distance to the star is estimated to be 100.92−0.40+0.38100.92^{+0.38}_{-0.40} pc. The extinction is reported as AV=0.24A_{V} = 0.24 mag in the Johnson VV band and AG=0.22A_{G} = 0.22 mag in the Gaia GG band. The effective temperature is given as Teff=7747.54−1276+999T_{\mathrm{eff}} = 7747.54^{+999}_{-1276} K, and the surface gravity as log⁡g=3.862−0.11+0.081\log g = 3.862^{+0.081}_{-0.11}. The metallicity is listed as [Fe/H]=−0.0486−0.17+0.35[\mathrm{Fe}/\mathrm{H}] = -0.0486^{+0.35}_{-0.17}, although the large uncertainty renders this estimate less reliable.

Several additional metallicity determinations for AI Vel can be found in the literature, although they often show considerable discrepancies. From Ca II triplet measurements, Matsuno et al. [68] derived [Fe/H]=+0.187±0.314[\mathrm{Fe}/\mathrm{H}] = +0.187 \pm0.314. Using Geneva photometric calibrations, Netopil [70] obtained [Fe/H]=−0.15[\mathrm{Fe}/\mathrm{H}] = -0.15. An independent spectroscopic analysis based on High Accuracy Radial Velocity Planet Searcher (HARPS) observations (Guiglion et al. [54]) was performed by Rainer et al. [80], who reported a metallicity of [Fe/H]=+0.11[\mathrm{Fe}/\mathrm{H}] = +0.11. Similar inconsistencies are seen in estimates of interstellar extinction. For example, Marshall et al. [67] reported E(B−V)=0.05E(B - V) = 0.05, which corresponds to AV=0.15A_V = 0.15 mag.

Spectroscopic analysis and fundamental parameters

As discussed in the previous subsection, the effective temperature of AI Vel reported in the literature shows significant discrepancies. To obtain a more reliable estimate, we analysed a set of 104 high-resolution spectra acquired with the HARPS spectrograph (Pepe et al. [75]). We adopted the HARPS-based spectroscopic parameters because the time-series spectra allow us to account explicitly for pulsation-induced variability rather than relying on phase-averaged measurements.

The observations were carried out over four consecutive nights, from 9 to 12 January 2011, with 26 spectra obtained each night (Guiglion et al. [54]). Each spectrum was recorded with an exposure time of 180 s and a typical signal-to-noise ratio of approximately 160. The wavelength coverage spans 3780–6900 Å, with a spectral resolution of 0.01 Å. We normalised all spectra using the neural-network-based Suppnet software (Różański et al. [86]).

The radial velocities and effective temperatures for all spectra were determined using the iSpec software package (Blanco-Cuaresma et al. [13]; Blanco-Cuaresma [12]). For radial velocity measurements, we employed the HARPS_SOPHIE.F0.360_698nm atomic line list as a reference template.

The derived radial velocities range from 12.5 to 50.8 km s−1^{-1}, with a mean value of 28.1(8) km s−1^{-1}. The typical uncertainties range from 0.06 to 0.2 km s−1^{-1}. The large variations in radial velocity are attributed to the high-amplitude pulsations characteristic of AI Vel. The discrepancy between our mean value and the radial velocity reported in the Gaia catalogue (17 ±\pm 3 km s−1^{-1}) likely results from pulsation-induced variability combined with differences in temporal sampling between the Gaia and HARPS observations.

The effective temperature of AI Vel, determined from the wings of the hydrogen lines Hα\alpha, Hβ\beta, and Hγ\gamma, was found to vary between 6747 and 7391 K, which gives Teff=7069±322T_{\rm eff} = 7069 \pm322 K. The mean surface gravity was derived as log⁡g=4.1(1)\log g = 4.1(1), and the microturbulent velocity as ξ=3.0(1)\xi= 3.0(1) km s−1^{-1}. These values of gravity and microturbulent velocity were obtained using the equivalent widths of selected Fe i and Fe ii lines located in the wavelength range 4620–4635 Å and having well-determined oscillator strengths (gfgf) (Landstreet [61]).

From well-isolated lines of Mg, Ca, Ti, and Fe, we derived a projected rotational velocity of Vrotsin⁡i=12.5(1)V_{\rm rot} \sin i = 12.5(1) km s−1^{-1} and an α\alpha-element abundance ratio of [α/Fe]=0.33(1)[\alpha/{\rm Fe}] = 0.33(1). The derived rotational velocity should be treated with caution, as line broadening may be significantly affected by stellar pulsations.

We determined the individual chemical abundances using spectral lines identified in the median-combined spectrum. Lines affected by telluric contamination were excluded from the analysis. The abundances of all analysed elements are listed in Table 1 and shown in Fig. A.1, where they are plotted as a function of atomic number. We found that the atmospheric abundances of most elements are slightly sub-solar. Notably low abundances are observed for iron-peak elements such as Mn, Cr, Fe, and Ni. In contrast, some elements—including O, Zr, and La—exhibit solar or even slightly super-solar abundances. Despite the underabundance of iron-group elements, the global metallicity remains relatively high because the total metal mass fraction is dominated by light elements, particularly C and O. Taking the individual elemental abundances into account, we obtain a global metallicity of Z=0.011Z = 0.011.

ZZElement[X/H]σ[X/H]\sigma_{\rm[X/H]}[X/Fe]
6C-0.240.050.31
8O0.130.160.68
11Na-0.270.060.28
12Mg-0.550.030.00
14Si-0.220.040.33
16S-0.310.090.24
20Ca-0.370.020.18
21Sc-0.160.030.39
22Ti-0.220.010.33
23V-0.300.100.25
24Cr-0.360.020.19
25Mn-0.580.03-0.03
26Fe-0.550.010.00
28Ni-0.530.020.02
30Zn-0.660.11-0.11
38Sr-0.110.110.44
39Y-0.160.050.39
40Zr0.060.080.63
56Ba-0.400.060.15
57La-0.010.140.54

Table 1. Chemical abundances of AI Vel derived from the median HARPS spectrum. The columns give the atomic number, ZZ, element, [X/H], its uncertainty σ[X/H]\sigma_{\rm[X/H]}, and [X/Fe].

For the derived effective temperature log⁡(Teff/K)=3.850±0.020\log(T_{\rm eff}/{\rm K}) = 3.850 \pm0.020, we adopt a bolometric correction of BC=−0.096±0.033BC = -0.096 \pm0.033 from Kurucz model atmospheres, using log⁡g=4.1(1)\log g = 4.1(1) and an assumed [m/H] = 0.0(1). To account for uncertainties in interstellar reddening, we considered the range of extinction values reported in the literature, AV=0.05A_V = 0.05–0.330.33 mag. Using the distance 100.92−0.40+0.38100.92^{+0.38}_{-0.40} pc from the StarHorse catalogue, we derive a stellar luminosity of log⁡(L/L⊙)=1.285±0.063\log(L/L_\odot) = 1.285 \pm0.063. This value is consistent with the period–luminosity relation for HADS stars (Poretti et al. [78]), given by

MV=−3.65log⁡P−1.83,M_V = -3.65 \log P - 1.83,

which yields MV=1.646M_V = 1.646 for P=0.111574P = 0.111574 d. Adopting Mbol=1.55M_{\rm bol} = 1.55, we obtain log⁡(L/L⊙)≈1.276\log(L/L_\odot) \approx1.276. The close agreement between both estimates supports the adopted distance and extinction. A summary of the available determinations of the parameters of AI Vel is provided in Table 2.

SourceTeffT_{\mathrm{eff}} [K]log⁡g\log g[Fe/H]AVA_V [mag]dd [pc]
Gaia DR37188−12+147188^{+14}_{-12}3.6623−0.026+0.0513.6623^{+0.051}_{-0.026}−1.3295−0.0023+0.0033-1.3295^{+0.0033}_{-0.0023}0.0589−0.0056+0.00660.0589^{+0.0066}_{-0.0056}100.40±0.19100.40 \pm0.19
StarHorse7748−1276+9997748^{+999}_{-1276}3.862−0.110+0.0813.862^{+0.081}_{-0.110}−0.049−0.17+0.35-0.049^{+0.35}_{-0.17}0.240.24100.92−0.40+0.38100.92^{+0.38}_{-0.40}
Hipparcos–––0.330.33–
uvby photometry [18]71107110–85208520––––
Netopil [70]––−0.15-0.15––
Rainer et al. [80]––+0.11+0.11––
Matsuno et al. [68]––+0.187±0.314+0.187 \pm0.314––
This work7069±3227069 \pm3224.1(1)4.1(1)−0.55(1)-0.55(1)0.050.05–0.330.33100.92−0.40+0.38100.92^{+0.38}_{-0.40}

Table 2. Selected determinations of the fundamental parameters of AI Vel.

Frequencies from TESS data

High-quality, nearly continuous space-based observations are available for AI Vel. The star was observed by the TESS satellite (Ricker et al. [81]) in six Sectors: 34, 35, 61, 62, 88 and 89. The detailed information on the observations is given in Table 3.

The TESS target pixel files for Sectors 34, 35, 61, and 62 were downloaded from the Mikulski Archive for Space Telescopes2 (MAST). We used target pixel files processed by the SPOC pipeline and extracted the light curves with the lightkurve package [see 63 https://docs.lightkurve.org/], using the corresponding SPOC aperture masks. The apertures adopted in our analysis are shown in Figure 1. We also made use of the astropy [5] and astroquery

The star images captured by the TESS satellite in Sectors 34, 35, 61 and 62

Figure 1. The star images captured by the TESS satellite in Sector 34 (upper left), 35 (upper right), 61 (bottom left) and 62 (bottom right). The red squares mark the photometric aperture. The images were generated using the Python Lightkurve package.

SectorDates (UTC)CCDCadence [s]N
3414 Jan. - 8 Feb. 2021112017376
359 Feb. - 6 Mar. 2021212014146
6118 Jan. - 12 Feb. 2023120010636
6212 Feb. - 10 Mar. 2023220010804
8814 Jan. - 11 Feb. 2025120011718
8911 Feb. - 12 Mar. 2025220011411

Table 3. Summary of the TESS observations of AI Vel. The columns list, from left to right, the sector number, observation dates, CCD detector number, cadence (in seconds), and the number of data points. All data were obtained with Camera 3.

[50] libraries. For Sectors 88 and 89, only Quick-Look Pipeline (QLP) light curves were available, and these were used in our analysis.

Figure 2 presents an image of AI Vel and its surrounding field (from Aladin sky atlas, v11.024, Bonnarel et al. 2000). The brightest star in its vicinity is CD–44 4195, with a visual magnitude of V=10.47V = 10.47 mag. It is classified as an F0-type star [90], with an effective temperature of Teff=7224.3±8.9T_{\mathrm{eff}} = 7224.3 \pm8.9 K [49]. Its surface gravity, log⁡g=3.9779−0.0037+0.0042\log g = 3.9779^{+0.0042}_{-0.0037}, is consistent with a main-sequence classification, and its subsolar metallicity is measured as [Fe/H] =−0.3218−0.0075+0.0071= -0.3218^{+0.0071}_{-0.0075}. Based on Gaia DR3 data, the star has a radial velocity of 18.1±1.118.1 \pm1.1 km s−1^{-1} and a parallax of π=2.677±0.014\pi= 2.677 \pm0.014 mas, corresponding to a distance of 377.7±1.8377.7 \pm1.8 pc.

Figure 2. Image of the field surrounding AI Vel obtained from the Aladin Sky Atlas.

CD–44 4195 is located at an angular separation of 56″ from AI Vel. Although its flux may fall partially within the TESS photometric aperture, the level of contamination is expected to be minimal due to its significantly fainter brightness. Nonetheless, some of the low-amplitude frequencies detected in the TESS light curve may originate from this nearby star.

The light curves from all sectors were subjected to detailed analysis. We computed Lomb-Scargle periodograms [65, 88] using an algorithm based on non-equally spaced fast Fourier transforms [e.g. 62]. Subsequently, we applied a non-linear least-squares fitting to the following function:

m(t)=∑i=1nAisin⁡(2π(t−T0)νi+φi).(1)m(t) = \sum_{i=1}^{n} A_i \sin\left(2\pi(t-T_0)\nu_i+\varphi_i\right). \tag*{(1)}

Here, m(t)m(t) represents the star’s brightness, tt denotes time, and νi\nu_i, AiA_i and φi\varphi_i are the adjustable parameters corresponding to frequency, amplitude, and phase, respectively. The reference time T0T_0 was set to the time of the first observation, determined separately for each analysed dataset.

Sectors 34 and 35 were observed consecutively, and thus were combined and analysed as a single dataset. The same approach was applied to sector pairs 61 and 62 as well as to 88 and 89. However, due to a gap of approximately two years between the observing periods, we chose not to merge all six sectors into one dataset, as the resulting spectral window would introduce aliases that complicate the frequency analysis. Additionally, possible variations in amplitudes and phases between the epochs could introduce additional uncertainties.

The combined data from Sectors 34+35 yielded about 31 500 data points, while Sectors 61+62 and 88+89 provided roughly 21 400 and 22 300 points, respectively. The Rayleigh frequency resolution for all combined datasets is approximately 0.019 d−10.019\ \mathrm{d}^{-1}. To assess the significance of detected peaks, we adopted a signal-to-noise ratio (S/N) threshold of 5.0, following the recommendation by Baran & Koen [10]. The noise level was calculated as the mean amplitude within a 50 d−150\ \mathrm{d}^{-1} window centred on the frequency of interest.

The results of the Fourier analysis are presented in Table 4, where only independent frequencies are listed. Periodograms of the original data and those obtained after prewhitening all significant signals are shown in Figure 3. Red vertical lines indicate the independent frequencies. In the second, fourth, and sixth panels, the black solid lines mark the threshold corresponding to five times the signal-to-noise ratio.

No.Frequency [d⁻¹]Amplitude [mag]φ [rad]S/N
Sector 34+35 (T0T_0 = BJD − 2457000.0 = 2228.75)
18.9626240(20)0.093437(19)4.00332(21)105.7
211.599782(20)†0.065320(16)4.34463(34)113.9
315.97823(11)†0.004898(34)1.686(17)36.7
415.95443(12)0.003637(57)0.914(15)34.7
522.45322(26)†0.002273(68)2.447(31)23.0
69.13580(13)†0.001594(14)3.288(22)16.8
717.9400(27)0.001774(55)5.351(33)19.5
831.4524(46)0.000510(24)1.131(68)11.1
910.8360(62)†0.000437(14)3.04(10)8.1
106.30180(43)0.000350(20)2.605(63)6.7
1120.7136(68)0.000214(16)0.34(11)5.5
Sector 61+62 (T0T_0 = BJD − 2457000.0 = 2962.80)
18.9626980(30)†0.088688(18)4.09012(20)106.3
211.599775(40)†0.067380(18)3.11499(26)116.5
315.97660(99)†0.004005(20)1.5884(50)33.6
415.9468(17)0.002878(20)1.6998(68)25.7
522.5162(85)†0.001212(20)2.977(17)18.9
69.13600(17)†0.001483(18)1.846(12)14.5
717.9424(69)0.001081(21)2.841(20)13.2
810.8536(16)†0.001401(18)1.757(13)17.5
90.45960(71)0.000948(52)5.02(10)10.7
106.31107(34)0.000781(22)2.870(57)10.0
1131.41607(29)0.000667(25)4.222(45)10.9
Sector 88+89 (T0T_0 = BJD − 2457000.0 = 3690.16)
18.962580(30)†0.096808(37)4.13728(43)107.9
211.599650(40)†0.074370(27)4.84872(72)120.4
315.973726(61)†0.004688(27)4.841(12)36.9
417.90009(12)0.002668(36)2.814(21)28.1
517.93526(19)0.003400(82)2.216(30)26.1
622.52186(10)†0.002657(26)4.668(20)25.8
710.85161(15)†0.001842(38)5.760(24)20.2
815.95356(38)0.001607(28)1.191(17)14.5
931.45314(30)0.000873(26)1.465(59)14.2
106.30044(12)0.000600(26)4.616(44)10.3
1120.6118(77)0.000333(25)0.589(75)7.0

Table 4. Independent frequencies of AI Vel derived from TESS data. The columns list the frequency identifier, frequency, amplitude, phase relative to the reference epoch T0T_0 adopted for each sector pair, and the signal-to-noise ratio. † Frequencies previously reported in the literature.

Figure 3. Fourier amplitude periodograms of AI Vel derived from the TESS light curves in Sectors 34+35 (upper two panels), 61+62 (middle two panels), and 88+89 (bottom two panels). For each sector pair, periodograms computed from the original data and from the residuals after prewhitening all significant frequencies are shown.

In the combined data from Sectors 34+35, we detected four harmonics of ν1\nu_1 and five harmonics of ν2\nu_2. In Sectors 61+62, four harmonics were identified for both ν1\nu_1 and ν2\nu_2, while in Sectors 88+89 we found three harmonics of ν1\nu_1 and five of ν2\nu_2. In all datasets, numerous combination frequencies involving the independent modes were also present.

Combining the datasets improved the frequency resolution, allowing us to resolve the close pair of frequencies at ∼15.98 d−1\sim15.98\ \mathrm{d}^{-1} and ∼15.95 d−1\sim15.95\ \mathrm{d}^{-1}, which were not always clearly distinguishable in the individual sectors. The combined data also yielded a significantly higher signal-to-noise ratio, leading to reduced uncertainties in the derived frequencies, amplitudes and phases.

The two highest-amplitude frequencies, ν1=8.963 d−1\nu_1 = 8.963\ \mathrm{d}^{-1} and ν2=11.600 d−1\nu_2 = 11.600\ \mathrm{d}^{-1}, were consistently detected across all analysed datasets. In the combined data from Sectors 34+35, the amplitude ratio is approximately A2/A1≈0.70A_2/A_1 \approx0.70, while the third independent frequency shows a much lower amplitude, with A3/A1≈0.05A_3/A_1 \approx0.05. All other detected frequencies have even smaller amplitudes. In the Sectors 61+62 dataset, the amplitude ratio increases slightly to A2/A1≈0.76A_2/A_1 \approx0.76, due to a decrease in the amplitude of the dominant mode accompanied by a simultaneous increase in the amplitude of the secondary mode. For Sectors 88+89 the amplitude ratio is on a similar level, A2/A1≈0.77A_2/A_1 \approx0.77.

The dominant frequency, ν1\nu_1, changed by approximately 7.4×10−5 d−17.4 \times10^{-5}\ \mathrm{d}^{-1} between Sectors 34+35 and 61+62. A slightly larger difference is observed between Sectors 61+62 and 88+89, amounting to −1.11×10−4 d−1-1.11 \times10^{-4}\ \mathrm{d}^{-1}. For the secondary frequency, ν2\nu_2, the shift is −7.0×10−6 d−1-7.0 \times10^{-6}\ \mathrm{d}^{-1} between Sectors 34+35 and 61+62, and −1.25×10−4 d−1-1.25 \times10^{-4}\ \mathrm{d}^{-1} between Sectors 61+62 and 88+89. These variations are larger than the expected evolutionary frequency changes. For hydrogen shell-burning δ\delta Scuti stars, the predicted secular variation of low-order radial-mode frequencies over a two-year interval (comparable to the gap between successive TESS observations) is typically of the order of 2×10−72 \times10^{-7} to 2×10−5 d−12 \times10^{-5}\ \mathrm{d}^{-1}. However, since AI Vel is a multiperiodic pulsator, such frequency shifts may instead arise from non-linear mode interactions or amplitude and phase modulation due to mode beating.

A portion of the TESS light curve from Sector 61 is shown in Fig. B.1, with the mean brightness value subtracted. The yellow line represents the model fitted to the combined 61+62 datasets using Eq. (1). The model reproduces the observations well, capturing both the amplitude and phase of the variability.

Our analysis confirmed all previously reported frequencies and revealed several new ones. In particular, we detected signals at 6.3 d−16.3\ \mathrm{d}^{-1} and 17.94 d−117.94\ \mathrm{d}^{-1}, both clearly present in the combined datasets from each pair of sectors. The ∼31.4 d−1\sim31.4\ \mathrm{d}^{-1} frequency also appears in all datasets. However, it is close to the combination frequency of the dominant mode at 8.962 d−18.962\ \mathrm{d}^{-1} and the 22.5 d−122.5\ \mathrm{d}^{-1} mode. The ∼20.7 d−1\sim20.7\ \mathrm{d}^{-1} frequency is detected in the combined Sectors 34+35 and 88+89, while a low-frequency signal at 0.46 d−10.46\ \mathrm{d}^{-1}, likely of instrumental origin, is present only in the 61+62 dataset.

Mode identification

Photometric method

The frequency ratio of the two dominant modes, ν1=8.963 d−1\nu_1 = 8.963\ \mathrm{d}^{-1} and ν2=11.600 d−1\nu_2 = 11.600\ \mathrm{d}^{-1}, suggests that they correspond to the radial fundamental (p1) and first overtone (p2) modes, respectively. Nevertheless, to confirm this identification, we performed an independent mode analysis using available observations in the Strömgren uvbyuvby photometric system [18].

We employed the method developed by Daszyńska-Daszkiewicz et al. [34], which allows for the simultaneous determination of the spherical degree ℓ\ell and two pulsational parameters, ff and ε\varepsilon, using photometric amplitudes and phases in at least three passbands.

The parameter ff is defined as the ratio of the local bolometric flux perturbation to the radial displacement at the stellar surface and enters the flux variation as follows:

δFbolFbol=Re⁡[εfYℓm(θ,ϕ)e−iωt],(2)\frac{\delta F_{\mathrm{bol}}}{F_{\mathrm{bol}}} = \operatorname{Re}\left[\varepsilon f Y_{\ell}^{m}(\theta,\phi)e^{-i\omega t}\right], \tag*{(2)}

where YℓmY_{\ell}^{m} is the spherical harmonic, and ω\omega is the angular pulsation frequency. Theoretical values of the ff-parameter are computed using linear non-adiabatic pulsation models, whereas empirical values are derived from multi-colour photometry.

The parameter ε\varepsilon represents the intrinsic amplitude of the radial displacement caused by stellar pulsation and is defined as

δrr=Re⁡[εYℓm(θ,ϕ)e−iωt].(3)\frac{\delta r}{r} = \operatorname{Re}\left[\varepsilon Y_{\ell}^{m}(\theta,\phi)e^{-i\omega t}\right]. \tag*{(3)}

The complex amplitude of the relative flux variation in a given photometric passband λ\lambda can be written as (e.g., Daszyńska-Daszkiewicz et al. [35]):

Aℓλ=Dℓλ(ε~f)+Eℓλε~,(4)\mathcal{A}_{\ell}^{\lambda} = \mathcal{D}_{\ell}^{\lambda}(\tilde{\varepsilon} f) + \mathcal{E}_{\ell}^{\lambda}\tilde{\varepsilon}, \tag*{(4)}

where ε~=εYℓm(i,0)\tilde{\varepsilon} = \varepsilon Y_{\ell}^{m}(i,0), and ii denotes the inclination angle of the stellar rotation axis.

The coefficients Dℓλ\mathcal{D}_{\ell}^{\lambda} and Eℓλ\mathcal{E}_{\ell}^{\lambda} describe the sensitivity of the monochromatic flux to variations in effective temperature and surface gravity (Daszyńska-Daszkiewicz et al. [34, 35]). They were computed using Vienna model atmospheres (Heiter et al. [56]). We adopted the non-linear limb-darkening law of Claret [30].

Next, using the four Strömgren passbands (uvbyuvby), we constructed a system of four complex equations which, for a given spherical degree ℓ\ell, can be solved for the complex parameters ε~\tilde{\varepsilon} and ff. To identify the most probable mode degree, we evaluate the quality of the fit using the following goodness-of-fit statistic:

χ2=12N−Np∑i=1N∣Aℓiobs−Aℓical∣2∣σAℓi∣2,(5)\chi^2 = \frac{1}{2N-N_p}\sum_{i=1}^{N}\frac{\left|\mathcal{A}_{\ell i}^{\mathrm{obs}}-\mathcal{A}_{\ell i}^{\mathrm{cal}}\right|^2}{\left|\sigma_{\mathcal{A}_{\ell i}}\right|^2}, \tag*{(5)}

where Aℓiobs\mathcal{A}_{\ell i}^{\mathrm{obs}} and Aℓical\mathcal{A}_{\ell i}^{\mathrm{cal}} are the observed and calculated complex amplitudes in the passband λi\lambda_i, σAℓi\sigma_{\mathcal{A}_{\ell i}} is the associated observational uncertainty, NN is the number of passbands, and NpN_p is the number of free parameters. Since both ε~\tilde{\varepsilon} and ff are complex, Np=4N_p = 4. Each passband provides two observational constraints (amplitude and phase), giving a total of 2N2N constraints.

If a clear minimum is found in the χ2\chi^2 values, i.e. in the difference between the observed and calculated amplitudes and phases, the corresponding spherical degree ℓ\ell, along with the associated values of ε~\tilde{\varepsilon} and ff, can be considered as the most probable.

Reliable amplitudes and phases in the Strömgren uvbyuvby passbands could be determined only for the two highest-amplitude modes, ν1\nu_1 and ν2\nu_2. Figure 4 presents the resulting χ2\chi^2 values as a function of mode degree ℓ\ell. The left panel corresponds to the frequency ν1\nu_1, while the right panel shows results for ν2\nu_2. Photometric mode identification could not be reliably performed for the remaining frequencies due to their low amplitudes.

Figure 4. The χ2\chi^2 values as a function of the mode degree, ℓ\ell. The left panel shows results for the dominant frequency, ν1\nu_1, while the right panel corresponds to the secondary frequency, ν2\nu_2. The effects of metallicity, [m/H], microturbulent velocity, ξt\xi_t, and position in the HR diagram are illustrated. Unless otherwise indicated in the legend, the models assume [m/H]=0.0, ξt=4\xi_t = 4 km s−1^{-1}, log⁡(Teff/K)=3.849\log(T_{\mathrm{eff}}/\mathrm{K}) = 3.849, and log⁡(L/L⊙)=1.230\log(L/L_\odot) = 1.230.

Since the discriminant χ2\chi^2 depends to some extent on the adopted model atmospheres, we investigated the effects of varying the microturbulent velocity, ξt\xi_t, and metallicity, [m/H][\mathrm{m}/\mathrm{H}], while keeping the stellar parameters (TeffT_{\mathrm{eff}}, log⁡g\log g) fixed to values corresponding to an evolutionary model located at the centre of the observational error box.

In a separate test, we also explored the sensitivity of the results to the adopted stellar parameters by considering two locations within the observational error box. We compared the χ2\chi^2 values for two different pairs of (log⁡Teff\log T_{\mathrm{eff}}, log⁡L/L⊙\log L/L_{\odot}), assuming fixed atmospheric parameters of [m/H]=0.0[\mathrm{m}/\mathrm{H}] = 0.0 and ξt=4 km s−1\xi_t = 4\,\mathrm{km}\,\mathrm{s}^{-1}.

In all tested configurations of the parameters, the χ2\chi^2 minimum consistently occurs at ℓ=0\ell= 0 for both dominant frequencies, indicating that they are most likely radial modes. According to the frequency ratio, ν1/ν2∼0.77\nu_1/\nu_2 \sim0.77, the most probable radial orders are n=1n = 1 (p1p_1) and n=2n = 2 (p2p_2) for ν1\nu_1 and ν2\nu_2, respectively.

An additional advantage of the photometric method of mode identification is that it provides empirical estimates of the ff-parameter for the two highest-amplitude modes of AI Vel. These empirical values can be compared with theoretical predictions, which are highly sensitive to several stellar and pulsational properties, including:

  • pulsation frequency, ω\omega,

  • the shape of the eigenfunctions, ℓ(ℓ+1)/ω2\ell(\ell+ 1)/\omega^2,

  • stellar parameters, such as mass and effective temperature,

  • chemical composition (e.g., XX, ZZ, and element mixture),

  • opacity data, κ\kappa,

  • efficiency of subphotospheric convection, described by the mixing-length parameter αMLT\alpha_{\mathrm{MLT}}.

The ff-parameter therefore serves as an additional seismic observable, complementary to the pulsation frequencies, and provides strong constraints on various stellar properties. In particular, for δ\delta Scuti stars, ff is highly sensitive to the efficiency of convective energy transport in the outer envelope, commonly parameterized by the mixing-length parameter αMLT\alpha_{\mathrm{MLT}} [34]. This quantity is included as a constraint in our asteroseismic analysis, described in detail in § 5.

Spectroscopic method

As described in Sect. 2.2, we analysed 104 HARPS spectra of AI Vel. Several spectral lines show clear time-dependent variations in their profiles, which we attribute to stellar pulsations.

Figure 5 illustrates the variability of the Ca i line. The upper-left panel shows line profiles from all spectra obtained over four nights. The middle-left panel presents profiles from the second night only, shifted vertically for clarity, while the bottom-left panel displays the mean line profile derived from all spectra.

The right panels show the phase-dependent behaviour of the line intensity. The upper-right panel presents a colour-coded map of intensity variations, whereas the bottom-right panel shows the residuals after subtracting the mean profile, emphasizing the variability pattern.

Figure 5. Variability of the Ca i line in the HARPS spectra. Left panels: line profiles from all four nights (upper; successive nights shown in red, black, green, and blue), profiles from the second night with vertical offsets for clarity (middle), and the mean line profile (bottom). Right panels: phase-dependent variations in line intensity (upper) and residuals obtained after subtracting the mean profile (bottom).

For each wavelength point across the profiles of five lines of Ca and Fe, we fitted sinusoidal functions to determine the amplitudes and phases corresponding to the six independent frequencies with the highest photometric amplitudes: 8.963 d−1^{-1}, 11.600 d−1^{-1}, 15.976 d−1^{-1}, 22.497 d−1^{-1}, 9.136 d−1^{-1}, and 10.847 d−1^{-1}. As an example, Figure 6 presents the results for the Ca i line at 6122.217 Å.

Figure 6. Amplitude and phase variations across the Ca I 6122.217 Å line profile in the HARPS spectra of AI Vel for the selected pulsation modes. The first and third rows show the amplitude distributions across the line profile, while the second and fourth rows present the corresponding phase variations. The blue line indicates the mean spectral line profile of Ca I.

Photometric mode identification presented in the previous subsection showed that the frequencies 8.963 d−1^{-1} and 11.600 d−1^{-1} correspond to radial modes. As shown in Figure 6, the pronounced asymmetry in the amplitude profile of the dominant frequency closely resembles the asymmetry observed in the amplitude profile of the 9.136 d−1^{-1} frequency. The latter is unlikely to be a radial mode, as its frequency lies between two securely identified radial modes and does not match the expected sequence of radial overtones; it must therefore correspond to a non-radial mode. This clear asymmetry can be naturally explained by the mutual interaction of the two closely spaced frequencies (9.136 - 8.963 = 0.173 d−1^{-1}). Given the short time span of the spectroscopic observations, any contribution of hypothetical orbital motion to the measured radial velocities can be considered negligible.

The fitted amplitudes, phases, and zero points for each frequency were then used in our attempts to identify the corresponding pulsation modes. We employed the FAMIAS software [101, 102], which performs mode identification by determining the degree, ℓ\ell, and azimuthal order, mm, of the observed frequencies.

We computed synthetic line profiles across the visible stellar surface, accounting for temperature variations and geometric deformation. The total synthetic spectra were obtained by integrating the contributions from all surface elements at the same observational phases as the real data. For a given combination of (ℓ,m)(\ell, m), theoretical amplitude and phase profiles were then calculated and compared with the observed ones.

Due to the close spacing and mutual interference between the 8.963 d−1^{-1} and 9.136 d−1^{-1} frequencies, we were unable to reproduce the amplitude and phase profiles of the dominant mode. However, we confirmed the photometric mode identification for the 11.600 d−1^{-1} frequency.

In addition, we successfully modelled the amplitude, A(λ)A(\lambda), and phase, ϕ(λ)\phi(\lambda), profiles of the 15.976 d−1^{-1} frequency. The spectroscopic analysis favours an axisymmetric mode, m=0m = 0, but does not provide a unique constraint on the spherical degree, ℓ\ell. The observed amplitude and phase profiles are compatible with several values of ℓ\ell. Because geometrical cancellation becomes increasingly strong for higher-degree modes, low-degree solutions are more likely. We therefore consider dipole (ℓ=1\ell= 1) and quadrupole (ℓ=2\ell= 2) identifications as the most plausible possibilities. For the remaining frequencies, the mode identification remained inconclusive.

Thus, the spectroscopic analysis provides a partial mode identification for the 15.976 d−1^{-1} frequency, constraining it to an axisymmetric, low-degree non-radial mode. In the following section, we investigate both the dipole and quadrupole hypotheses and show how the inclusion of this additional frequency affects the seismic constraints.

Asteroseismic modelling

AI Vel is an excellent candidate for asteroseismic analysis, offering a rare combination of high-quality observational constraints. These include two securely identified radial pulsation modes, ν1\nu_1 and ν2\nu_2, the corresponding empirical values of the non-adiabatic ff-parameter, and a well-constrained position in the HR diagram. In addition, the spectroscopic analysis indicates that the frequency at 15.976 d−1^{-1} corresponds to an axisymmetric, m=0m = 0, most likely low-degree non-radial mode. In the following analysis, we therefore consider both dipole, ℓ=1\ell= 1, and quadrupole, ℓ=2\ell= 2, hypotheses for this frequency.

To perform the seismic modelling, we applied a Bayesian analysis based on Monte Carlo simulations. The likelihood was assumed to be Gaussian and is given by [e.g. 60, 33, 82, 59]:

L(O∣θ)=∏i12πσi2exp⁡[−(Oi−Mi)22σi2],(6)\mathcal{L}(\mathcal{O}|\boldsymbol{\theta}) = \prod_i \frac{1}{\sqrt{2\pi\sigma_i^2}} \exp\left[-\frac{(O_i-\mathcal{M}_i)^2}{2\sigma_i^2}\right], \tag*{(6)}

where OiO_i denotes the observed quantities, Mi\mathcal{M}_i the corresponding model predictions, and σi\sigma_i are the uncertainties adopted for the individual constraints. The vector θ\boldsymbol{\theta} represents the set of adjustable model parameters. In our analysis, these comprise the stellar mass MM, initial hydrogen abundance X0X_0, metallicity ZZ, initial rotational velocity Vrot,0V_{\mathrm{rot},0}, convective core overshooting parameter αov\alpha_{\mathrm{ov}}, and mixing-length parameter αmlt\alpha_{\mathrm{mlt}}. For all adjustable parameters, we adopted uniform priors over the ranges explored by the model grid. This methodology has been successfully applied in our previous seismic analyses of β\beta Pic-Pegasi [40], AE Ursae Majoris and RV Arietis [41], V2367 Cygni [37], five SX Phoenicis stars in ω\omega Centauri [38], and BL Camelopardalis (Sewczuk et al. 2026).

The evolutionary tracks were computed using the Warsaw–New Jersey stellar evolution code [72, 73]. Unless stated otherwise, the models adopt the solar chemical mixture of Asplund et al. (2009) (AGSS09) and the OPAL2005 equation of state [84, 85]. Convective core overshooting is included following the prescription of Dzięmbowski & Pamyatnykh [45].

Our previous studies have shown that, for δ\delta Scuti stars, the choice of opacity tables can significantly affect the results [e.g. 39]. Therefore, we performed simulations using three different opacity datasets: OPAL [58], OP [89] and OPLIB [32, 31]. At lower temperatures, log T<3.95T < 3.95, all three datasets were supplemented with the low-temperature opacity tables of Ferguson et al. [47].

Stellar pulsations were computed using the linear non-adiabatic pulsation code developed by Dziembowski [44]. The code includes both first- and second-order rotational corrections to the pulsation frequencies within the framework of perturbation theory. The calculations were performed using the frozen-convection approximation, which assumes that the convective flux remains unchanged throughout the pulsation cycle. This approximation is expected to be appropriate when convective energy transport in the stellar envelope is relatively inefficient, as is likely the case for a moderately massive star such as AI Vel. Nevertheless, even weak convection may influence the theoretical values of the ff-parameter.

Two radial modes

In the first approach to the seismic modelling of AI Vel, we considered only two radial modes. The model predictions were compared with eight observational constraints: the effective temperature TeffT_{\mathrm{eff}}, luminosity L/L⊙L/L_{\odot}, the frequencies of the two radial modes ν1\nu_1 and ν2\nu_2, and the corresponding non-adiabatic ff-parameters, namely Re⁡f(ν1)\operatorname{Re} f(\nu_1), Im⁡f(ν1)\operatorname{Im} f(\nu_1), Re⁡f(ν2)\operatorname{Re} f(\nu_2), and Im⁡f(ν2)\operatorname{Im} f(\nu_2).

We computed a large grid of models (approximately 200 000 in total for each opacity source), from which posterior probability distributions were derived. The resulting histograms are presented in Fig. C.3, and the median values of model parameters are listed in Table 5.

Table 5: Parameters derived from simulations including two radial modes. The table columns contain: mass, MM, metallicity ZZ, initial hydrogen abundance, X0X_0, initial surface rotational velocity, Vrot,0V_{\mathrm{rot},0}, current surface rotational velocity, VrotV_{\mathrm{rot}}, convective mixing length parameter, αmlt\alpha_{\mathrm{mlt}}, convective core overshooting parameter, αov\alpha_{\mathrm{ov}}, effective temperature, log⁡Teff\log T_{\mathrm{eff}}, luminosity, log⁡L/L⊙\log L/L_{\odot}, age since the zero-age main sequence, Age, gravity, log⁡g\log g, radius, RR, and amplitudes of relative radius variations due to pulsation in the radial fundamental, ε(ν1)\varepsilon(\nu_1), and first overtone, ε(ν2)\varepsilon(\nu_2), modes as well as the instability parameters for the radial fundamental, η(ν1)\eta(\nu_1), and first overtone, η(ν2)\eta(\nu_2), modes. The atmospheric microturbulent velocity, ξt\xi_t, was fixed at 4 km s−1^{-1}.

OpacityM[M⊙]M [M_{\odot}]ZZX0X_0Vrot,0V_{\mathrm{rot},0}VrotV_{\mathrm{rot}}αmlt\alpha_{\mathrm{mlt}}αov\alpha_{\mathrm{ov}}log⁡Teff\log T_{\mathrm{eff}}log⁡L/L⊙\log L/L_{\odot}
OPAL1.737−0.032+0.0421.737^{+0.042}_{-0.032}0.01392−0.00088+0.001200.01392^{+0.00120}_{-0.00088}0.7105−0.0073+0.00790.7105^{+0.0079}_{-0.0073}12.3−8.7+7.512.3^{+7.5}_{-8.7}11.6−8.1+7.011.6^{+7.0}_{-8.1}0.467−0.088+0.0940.467^{+0.094}_{-0.088}0.094−0.056+0.0550.094^{+0.055}_{-0.056}3.8591−0.0015+0.00173.8591^{+0.0017}_{-0.0015}1.2595−0.0088+0.0111.2595^{+0.011}_{-0.0088}
OP1.566−0.025+0.0241.566^{+0.024}_{-0.025}0.01419−0.00084+0.000640.01419^{+0.00064}_{-0.00084}0.7207−0.011+0.00530.7207^{+0.0053}_{-0.011}5.7−5.7+9.65.7^{+9.6}_{-5.7}5.4−5.4+9.25.4^{+9.2}_{-5.4}0.331−0.028+0.0260.331^{+0.026}_{-0.028}0.146−0.038+0.0660.146^{+0.066}_{-0.038}3.8179−0.0014+0.00173.8179^{+0.0017}_{-0.0014}1.0634−0.0069+0.0101.0634^{+0.010}_{-0.0069}
OPLIB1.696−0.040+0.0341.696^{+0.034}_{-0.040}0.0178−0.0015+0.00130.0178^{+0.0013}_{-0.0015}0.7160−0.0094+0.00840.7160^{+0.0084}_{-0.0094}6.6−6.6+9.86.6^{+9.8}_{-6.6}6.3−6.3+9.26.3^{+9.2}_{-6.3}0.413−0.033+0.0420.413^{+0.042}_{-0.033}0.098−0.029+0.0480.098^{+0.048}_{-0.029}3.8276−0.0019+0.00153.8276^{+0.0015}_{-0.0019}1.1250−0.0084+0.00851.1250^{+0.0085}_{-0.0084}
OpacityAge [Gyr]log⁡g\log gR[R⊙]R[R_{\odot}]ε(ν1)\varepsilon(\nu_1)ε(ν2)\varepsilon(\nu_2)η(ν1)\eta(\nu_1)η(ν2)\eta(\nu_2)
OPAL1.398−0.068+0.0781.398^{+0.078}_{-0.068}3.8055−0.0029+0.00383.8055^{+0.0038}_{-0.0029}2.728−0.016+0.0212.728^{+0.021}_{-0.016}0.01029−0.00013+0.000170.01029^{+0.00017}_{-0.00013}0.00480−0.00002+0.000020.00480^{+0.00002}_{-0.00002}0.051−0.006+0.0060.051^{+0.006}_{-0.006}0.073−0.005+0.0040.073^{+0.004}_{-0.005}
OP2.099−0.067+0.132.099^{+0.13}_{-0.067}3.7913−0.0022+0.00243.7913^{+0.0024}_{-0.0022}2.633−0.015+0.0132.633^{+0.013}_{-0.015}0.0221−0.0011+0.00110.0221^{+0.0011}_{-0.0011}0.0089−0.0005+0.00050.0089^{+0.0005}_{-0.0005}0.109−0.001+0.0010.109^{+0.001}_{-0.001}0.085−0.003+0.0030.085^{+0.003}_{-0.003}
OPLIB1.710−0.057+0.0761.710^{+0.076}_{-0.057}3.8029−0.0038+0.00323.8029^{+0.0032}_{-0.0038}2.704−0.021+0.0172.704^{+0.017}_{-0.021}0.0194−0.0012+0.00130.0194^{+0.0013}_{-0.0012}0.0077−0.0005+0.00050.0077^{+0.0005}_{-0.0005}0.099−0.001+0.0010.099^{+0.001}_{-0.001}0.092−0.001+0.0010.092^{+0.001}_{-0.001}

Table 5.

We adopted the median rather than the mean as the representative value for each parameter, since many of the posterior distributions are significantly non-Gaussian (for example, initial hydrogen abundance or rotational velocity in Fig. C.3). In such cases, the median provides a more reliable and stable estimate of the central value.

Uncertainties were estimated as the difference between the median and the 16th and 84th percentiles of the posterior distributions. For a Gaussian distribution, the interval between the 16th and 84th percentiles corresponds approximately to the conventional 1σ1\sigma confidence interval.

All models listed in Table 5 reproduce the observed frequencies of the radial modes within the adopted uncertainties. Instead of using the formal uncertainties derived from the TESS data analysis, we adopted a frequency uncertainty of 0.001 d−10.001\,\mathrm{d}^{-1}. This value reflects both the numerical precision of the computed pulsation frequencies and the frequency variations observed between different TESS sectors.

In our calculations, we used three opacity data sources: OPAL, OP, and OPLIB. A significant result is that only the models computed with OPAL opacities simultaneously reproduce both the effective temperature and luminosity of AI Vel. In contrast, our seismic models based on OP and OPLIB opacities lie well below the observational error box. Figure 7 presents the Hertzsprung–Russell (HR) diagram, where the position of AI Vel is marked with the seismic models obtained from our simulations. As demonstrated, only the OPAL-based models are consistent with the observed location of AI Vel. Similar behaviour has previously been reported for other A- and F-type pulsating stars (e.g., Dąbrowska-Dzierżanowska et al. 2023a, 2025).

Figure 7. The HR diagram showing the position of AI Vel. The orange rectangle indicates the observational error box. Seismic models computed using OPAL, OP, and OPLIB opacity tables are also plotted for comparison (see text for details). The evolutionary tracks were calculated with X0=0.7X_0 = 0.7, Z=0.010Z = 0.010, Vrot,0=10 km s−1V_{\mathrm{rot},0} = 10\,\mathrm{km\,s}^{-1}, αmlt=1.0\alpha_{\mathrm{mlt}} = 1.0 and αov=0.0\alpha_{\mathrm{ov}} = 0.0.

For the OPAL models, the derived stellar mass is ∼1.737 M⊙\sim1.737\,M_{\odot}, with a metallicity close to the solar value (Z∼0.0139Z \sim0.0139). The initial hydrogen content aligns with X0∼0.71X_0 \sim0.71, and the inferred rotational velocity is low, Vrot∈(4,19) km s−1V_{\mathrm{rot}} \in(4,19)\,\mathrm{km\,s}^{-1}, indicating that AI Vel is a slow rotator.

The efficiency of convective energy transport in the outer layers is found to be very low, with a mixing-length parameter αmlt∼0.5\alpha_{\mathrm{mlt}} \sim0.5. Similarly, the extent of convective core overshooting during the main-sequence phase is small, with αov∼0.1\alpha_{\mathrm{ov}} \sim0.1, suggesting limited mixing beyond the core boundary. The estimated stellar age is 1.4 Gyr and the models indicate that AI Vel is in the hydrogen shell-burning phase of post-main-sequence evolution. The central hydrogen abundance is fully depleted, with Xc=0X_{\mathrm{c}} = 0.

Both considered radial modes are predicted to be excited. To assess mode stability, we used the normalized instability parameter η\eta (Stellingwerf [91]), which ranges from −1-1 (strongly damped) to +1+1 (strongly excited). For the dominant frequency ν1\nu_1, we obtained η∼0.05\eta\sim0.05, and for the secondary frequency ν2\nu_2, we got η∼0.07\eta\sim0.07, confirming that both modes are unstable and self-excited in the models.

The OP and OPLIB models differ significantly from the OPAL models. They are characterised by higher metallicities, Z∈(0.014,0.019)Z \in(0.014,0.019), and indicate larger initial hydrogen abundances, X0∈(0.716,0.72)X_0 \in(0.716,0.72).

Models computed with all opacity tables are slow rotators, with Vrot≲19 km s−1V_{\mathrm{rot}} \lesssim19\,\mathrm{km\,s}^{-1}. In all cases, the models require low efficiency of convective energy transport, as well as a small extent of convective core overshooting.

Figure 8 presents the empirical and theoretical values of the ff-parameter in the ∣f∣|f|–arg⁡(f)\arg(f) plane. The quantity ∣f∣|f| represents the amplitude of the bolometric flux variations, while the phase arg⁡(f)\arg(f) corresponds to the phase lag between variations in stellar flux and radius. The agreement between the observed and predicted values for ν1\nu_1 (upper panel) is generally good, with most models falling within the 1σ1\sigma uncertainties of ff. In contrast, for the radial first overtone mode (bottom panel), a significant discrepancy is evident for ∣f∣|f|. All models used in this simulation were computed assuming a microturbulent velocity of ξt=4 km s−1\xi_{\mathrm{t}} = 4\,\mathrm{km\,s}^{-1}. The instability parameter η\eta is colour-coded. As noted earlier, all considered modes are unstable; however, a correlation between η\eta and the ff-parameter is apparent. Interestingly, better agreement between the empirical and theoretical ff values tends to occur for modes with higher η\eta.

Figure 8. Empirical and theoretical values of the non-adiabatic ff-parameter for the radial fundamental (upper panel) and first overtone (bottom panel) modes. Models from the OPAL simulation are plotted. Different colours correspond to different values of the instability parameter, η\eta.

Because of the small frequency difference, the theoretical ff values for the two radial modes are very similar. In contrast, the empirical amplitude, ∣f∣|f|, derived for ν2\nu_2 is significantly larger.

This discrepancy may be related to uncertainties in the determination of the empirical ff-parameter for this mode. Alternatively, it may indicate the presence of an additional physical effect not accounted for in our analysis.

Adding the non-radial mode

Including a non-radial mode in the seismic modelling provides an additional constraint that can substantially narrow the posterior distributions of the stellar parameters, provided that the mode identification is correct. In this analysis, we incorporated the frequency ν3=15.976\nu_3 = 15.976 d−1^{-1}. The spectroscopic analysis indicates that this frequency corresponds to an axisymmetric mode (m=0m = 0), whereas its spherical degree ℓ\ell could not be uniquely constrained. We therefore considered two plausible low-degree identifications: a dipole mode, (ℓ,m)=(1,0)(\ell,m) = (1,0), and a quadrupole mode, (ℓ,m)=(2,0)(\ell,m) = (2,0).

For each of these hypotheses, we performed an additional seismic simulation including ν3\nu_3 as an observational constraint. Only OPAL opacities were considered at this stage, because, within the explored parameter space, our models computed with OP and OPLIB opacities lie well outside the observational error box in the HR diagram.

The inclusion of this mode leads to a substantial tightening of the posterior distributions for each assumed mode identification, as summarized in Table 6. Most of the derived stellar parameters remain consistent with those obtained in the previous simulations, but their formal statistical uncertainties are considerably reduced. However, although the formal uncertainties decrease for a given mode identification, the true mode degree remains uncertain. Therefore, this apparent improvement in precision does not reflect the total uncertainty of the stellar parameters, as it does not account for the systematic differences between the alternative mode-identification hypotheses.

OpacityM[M⊙]M[M_\odot]ZZX0X_0Vrot,0V_{\mathrm{rot},0}VrotV_{\mathrm{rot}}αmlt\alpha_{\mathrm{mlt}}αov\alpha_{\mathrm{ov}}log⁡Teff\log T_{\mathrm{eff}}log⁡LL⊙\log\frac{L}{L_\odot}
OPAL1.7258−0.020+0.00471.7258^{+0.0047}_{-0.020}0.01335−0.00024+0.000310.01335^{+0.00031}_{-0.00024}0.7132−0.0077+0.00350.7132^{+0.0035}_{-0.0077}14.4−5.2+7.714.4^{+7.7}_{-5.2}13.5−4.9+7.213.5^{+7.2}_{-4.9}0.158−0.095+0.120.158^{+0.12}_{-0.095}0.0690−0.0029+0.00400.0690^{+0.0040}_{-0.0029}3.8578−0.0013+0.00133.8578^{+0.0013}_{-0.0013}1.2515−0.0058+0.00591.2515^{+0.0059}_{-0.0058}
(ℓ=1,m=0)(\ell=1,m=0)
OPAL1.7114−0.0079+0.00521.7114^{+0.0052}_{-0.0079}0.01375−0.00013+0.000170.01375^{+0.00017}_{-0.00013}0.7052−0.0055+0.00260.7052^{+0.0026}_{-0.0055}8.8−5.9+14.98.8^{+14.9}_{-5.9}8.3−5.5+14.08.3^{+14.0}_{-5.5}0.17−0.11+0.140.17^{+0.14}_{-0.11}0.0785−0.0024+0.00200.0785^{+0.0020}_{-0.0024}3.8581−0.0013+0.00123.8581^{+0.0012}_{-0.0013}1.2509−0.0057+0.00501.2509^{+0.0050}_{-0.0057}
(ℓ=2,m=0)(\ell=2,m=0)
OPAL1.7102−0.0023+0.00881.7102^{+0.0088}_{-0.0023}0.01391−0.00003+0.000290.01391^{+0.00029}_{-0.00003}0.6999−0.0084+0.00070.6999^{+0.0007}_{-0.0084}18.4−7.1+18.918.4^{+18.9}_{-7.1}17.2−7.6+17.817.2^{+17.8}_{-7.6}0.28−0.15+0.130.28^{+0.13}_{-0.15}0.0813−0.0038+0.00290.0813^{+0.0029}_{-0.0038}3.869−0.028+0.0053.869^{+0.005}_{-0.028}1.297−0.11+0.0191.297^{+0.019}_{-0.11}
GN93 (ℓ=1,m=0)(\ell=1,m=0)
OPAL1.7367−0.0081+0.00221.7367^{+0.0022}_{-0.0081}0.01438−0.00031+0.000100.01438^{+0.00010}_{-0.00031}0.7135−0.0022+0.00120.7135^{+0.0012}_{-0.0022}10.7−5.7+2.210.7^{+2.2}_{-5.7}10.0−5.6+2.110.0^{+2.1}_{-5.6}0.082−0.033+0.0540.082^{+0.054}_{-0.033}0.087−0.002+0.0050.087^{+0.005}_{-0.002}3.854−0.016+0.0013.854^{+0.001}_{-0.016}1.240−0.065+0.0051.240^{+0.005}_{-0.065}
Kurucz (ℓ=1,m=0)(\ell=1,m=0)
OpacityAge [Gyr]log⁡g\log gR[R⊙]R[R_\odot]ε(ν1)\varepsilon(\nu_1)ε(ν2)\varepsilon(\nu_2)η(ν1)\eta(\nu_1)η(ν2)\eta(\nu_2)η(ν3)\eta(\nu_3)ξt\xi_t
OPAL1.395−0.016+0.0141.395^{+0.014}_{-0.016}3.8041−0.0012+0.00063.8041^{+0.0006}_{-0.0012}2.7221−0.011+0.00292.7221^{+0.0029}_{-0.011}0.0103−0.0001+0.00020.0103^{+0.0002}_{-0.0001}0.0050−0.0001+0.00010.0050^{+0.0001}_{-0.0001}0.0365−0.0057+0.00640.0365^{+0.0064}_{-0.0057}0.0577−0.0033+0.00470.0577^{+0.0047}_{-0.0033}0.0637−0.0016+0.00670.0637^{+0.0067}_{-0.0016}3.19−0.19+0.283.19^{+0.28}_{-0.19}
(ℓ=1,m=0)(\ell=1,m=0)
OPAL1.391−0.017+0.0201.391^{+0.020}_{-0.017}3.8031−0.0004+0.00083.8031^{+0.0008}_{-0.0004}2.7140−0.0041+0.00322.7140^{+0.0032}_{-0.0041}0.0102−0.0001+0.00020.0102^{+0.0002}_{-0.0001}0.0050−0.0001+0.00010.0050^{+0.0001}_{-0.0001}0.0382−0.0045+0.00700.0382^{+0.0070}_{-0.0045}0.0588−0.0027+0.00610.0588^{+0.0061}_{-0.0027}0.0645−0.0014+0.00920.0645^{+0.0092}_{-0.0014}3.18−0.21+0.343.18^{+0.34}_{-0.21}
(ℓ=2,m=0)(\ell=2,m=0)
OPAL1.218−0.038+0.0391.218^{+0.039}_{-0.038}3.8024−0.0022+0.00043.8024^{+0.0004}_{-0.0022}2.7151−0.0052+0.00382.7151^{+0.0038}_{-0.0052}0.0103−0.0002+0.00020.0103^{+0.0002}_{-0.0002}0.0050−0.0001+0.00010.0050^{+0.0001}_{-0.0001}−0.003−0.011+0.007-0.003^{+0.007}_{-0.011}0.0389−0.0074+0.00540.0389^{+0.0054}_{-0.0074}0.0714−0.0022+0.00990.0714^{+0.0099}_{-0.0022}2.69−0.36+0.272.69^{+0.27}_{-0.36}
GN93 (ℓ=1,m=0)(\ell=1,m=0)
OPAL1.442−0.012+0.0081.442^{+0.008}_{-0.012}3.8059−0.0002+0.00053.8059^{+0.0005}_{-0.0002}2.7267−0.0062+0.00152.7267^{+0.0015}_{-0.0062}0.0113−0.0008+0.00020.0113^{+0.0002}_{-0.0008}0.0054−0.0004+0.00010.0054^{+0.0001}_{-0.0004}0.043−0.003+0.0380.043^{+0.038}_{-0.003}0.060−0.002+0.0120.060^{+0.012}_{-0.002}0.060−0.018+0.0010.060^{+0.001}_{-0.018}2.002.00
Kurucz (ℓ=1,m=0)(\ell=1,m=0)

Table 6. Same as Table 5, but for simulations including the non-radial mode ν3\nu_3 and restricted to OPAL opacities. The microturbulent velocity, ξt\xi_t, inferred from the modelling is also listed. Different modelling assumptions are considered, including dipole- and quadrupole-mode identifications of ν3\nu_3, the GN93 chemical mixture, and Kurucz model atmospheres.

The posterior distributions of selected parameters are presented in Fig. C.4 (for the dipole mode hypothesis) and in Fig. C.5 (for the quadrupole mode hypothesis). We emphasize that the quoted uncertainties reflect the statistical spread within the adopted model grid and do not account for systematic uncertainties associated with the input physics or the ambiguity in the identification of the non-radial mode.

Including this mode increases the total number of observational constraints to nine: log⁡Teff\log T_{\mathrm{eff}}, log⁡L/L⊙\log L/L_{\odot}, ν1\nu_1, ν2\nu_2, Ref(ν1)Ref(\nu_1), Imf(ν1)Imf(\nu_1), Ref(ν2)Ref(\nu_2), Imf(ν2)Imf(\nu_2), and ν3\nu_3. This additional constraint allows us to treat the microturbulent velocity, ξt\xi_t, as a free parameter in the derivation of the empirical ff-parameter. Although

ξt\xi_t is not a stellar-structure parameter, it influences the empirical determination of ff through the adopted model atmospheres. However, since the value derived from the HARPS spectroscopic analysis is ξt=3.0(1) km s−1\xi_t = 3.0(1)\ \mathrm{km\ s^{-1}}, we restricted our analysis to models with ξt≲3.5 km s−1\xi_t \lesssim3.5\ \mathrm{km\ s^{-1}}.

Figure 9 compares the empirical and theoretical values of the ff-parameter for the dipole-mode hypothesis. The colour coding indicates the value of ξt\xi_t adopted in deriving the empirical ff values. The corresponding results for the quadrupole-mode hypothesis are shown in Fig. C.6.

Figure 9. Empirical and theoretical values of the non-adiabatic ff-parameter for the radial fundamental (upper panel) and first overtone (bottom panel) modes for the seismic models obtained under the dipole-mode hypothesis for ν3\nu_3. The models were computed with OPAL opacities. Different colours correspond to different values of the microturbulent velocity, ξt\xi_{\mathrm{t}}.

For models fitting two radial modes and additionally the ν3\nu_3 frequency as a dipole mode, the main difference, apart from the microturbulent velocity, is found in the mixing-length parameter, which is significantly reduced to αmlt≈0.16\alpha_{\mathrm{mlt}}\approx0.16. For the remaining parameters, we primarily observe a reduction in their spreads, leading to smaller uncertainties.

The radial order that best reproduces the observed dipole mode falls in the range n=9n=9–1111 in our simulations. In hydrogen shell-burning models, these frequencies correspond to gravity modes, since most of their kinetic energy is concentrated in the deep interior of the star. The ratio of the kinetic energy contained within the gravity-mode propagation cavity to the total kinetic energy is Ekin,grav/Ekin=0.94−0.06+0.04E_{\mathrm{kin,grav}}/E_{\mathrm{kin}}=0.94^{+0.04}_{-0.06}. Consequently, the frequency at 15.976 d−115.976\ \mathrm{d}^{-1} can be identified with the g9g_9, g10g_{10}, or g11g_{11} mode.

In Fig. C.1 we plotted the mode inertia, highlighting the different nature of the radial modes and the mixed dipole mode. For the radial modes, the inertia amplitude is distributed over a large region between log⁡(T/K)∼6.5\log(T/\mathrm{K})\sim6.5 and log⁡(T/K)∼4.5\log(T/\mathrm{K})\sim4.5. The dipole mode g10g_{10}, identified with the frequency ν3\nu_3, is basically insensitive to that part of the star. The highest inertia of the dipole mode is concentrated in the stellar core, where the eigenfunction exhibits ten closely spaced radial nodes. As a result, the g10g_{10} mode is particularly sensitive to the structure near the convective core boundary established during the main-sequence phase of stellar evolution. The model parameters correspond to those listed in Table 6 labelled as (ℓ=1,m=0)(\ell=1,m=0).

The eigenfunction ff is shown in detail in Figure 10, where both its amplitude, ∣f∣|f|, and phase, arg⁡(f)\arg(f), are plotted as functions of temperature within a seismic model of Al Vel. The quantity referred to as the ff-parameter corresponds to the value of this eigenfunction evaluated at the photosphere.

Figure 10. Profiles of the eigenfunction ff for the seismic model of AI Vel, whose parameters are listed in Table 6. The left-hand axis shows the amplitude ∣f∣\lvert f\rvert, while the right-hand axis indicates the argument, arg⁡(f)\arg(f). The upper panel corresponds to the radial fundamental mode, the middle panel to the first overtone radial mode, and the bottom panel to the dipole mode g10g_{10}. The empirical values for the two radial modes are also marked for comparison.

From top to bottom, the panels in Figure 10 show the ff-eigenfunction for the radial fundamental mode, the radial first overtone, and the dipole g10g_{10} mode, respectively. For the two radial modes, the empirical values of the ff-parameter, corresponding to the value of the eigenfunction at the photosphere, are also marked for comparison. The empirical ff-parameter for the radial fundamental mode is reproduced well by the theoretical model. In contrast, for the first overtone mode, the amplitude of the theoretical ff-parameter is slightly underestimated.

The instability parameter as a function of frequency for mode degrees ℓ=0\ell=0, 1, and 2 is shown in Fig. 11. Vertical lines indicate the independent frequencies detected in Al Vel. Most of the observed frequencies fall within the unstable regime. An intriguing feature is the presence of the low-frequency mode ν10=6.3 d−1\nu_{10}=6.3\ \mathrm{d}^{-1}. If this frequency corresponds to an independent pulsation mode, it remains strongly damped in our models. A similar situation occurs for the highest detected frequency, ∼31.45 d−1\sim31.45\ \mathrm{d}^{-1}, whose instability parameter is approximately −0.6-0.6. It is therefore possible that these two frequencies are not independent modes but instead represent unresolved combination frequencies. On the other hand, the 6.3 d−16.3\ \mathrm{d}^{-1} signal can be reproduced by an axisymmetric dipole g28g_{28} mode, while the ∼31.45 d−1\sim31.45\ \mathrm{d}^{-1} frequency may correspond to an axisymmetric quadrupole p2p_2 mode. The difficulty in exciting these modes could therefore indicate that the opacity profile adopted in the models is not fully adequate and may require modifications [e.g. 93, 94]. A detailed investigation of this issue, however, is beyond the scope of the present paper.

Details of the considered seismic model structure are presented in Appendix § 10, Fig. C.2. The upper panel shows the mean opacity coefficient, log⁡κ\log\kappa, along with its derivatives with respect to temperature, ∂log⁡κ/∂log⁡T\partial\log\kappa/\partial\log T, and density, ∂log⁡κ/∂log⁡ρ\partial\log\kappa/\partial\log\rho. In the bottom panel, the left-hand axis presents the energy generation rates from the pp chain (ϵpp\epsilon_{\mathrm{pp}}) and the CNO cycle (ϵCNO\epsilon_{\mathrm{CNO}}), while the right-hand axis shows the hydrogen (XX) and helium (YY) abundances. The model is clearly in the hydrogen shell-burning phase of evolution, with the CNO cycle as the dominant source of nuclear energy.

Finally, we consider the frequency ν4=15.952\nu_4 = 15.952 d−1^{-1}, which lies close to the dipole axisymmetric mode ν3=15.976\nu_3 = 15.976 d−1^{-1}. Still, the separation between the two frequencies is larger than the Rayleigh frequency resolution. This frequency could represent a rotationally split component of a dipole mode, with ν3\nu_3 and ν4\nu_4 forming two members of the same triplet.

Finally, we consider the frequency ν4=15.952\nu_4 = 15.952 d−1^{-1}, which lies close to ν3=15.976\nu_3 = 15.976 d−1^{-1}. Still, the separation between the two frequencies is larger than the Rayleigh frequency resolution. This frequency could represent a rotationally split component of a dipole mode, with ν3\nu_3 and ν4\nu_4 forming two members of the same triplet.

The observed separation between ν3\nu_3 and ν4\nu_4 is approximately 0.0240.024 d−1^{-1}. For our seismic models, the Ledoux constant of the relevant dipole modes is Cn=9−11,ℓ=1=0.49−0.10+0.07C_{n=9-11,\ell=1} = 0.49^{+0.07}_{-0.10}, while the inferred rotational frequency is νrot=0.098−0.035+0.052\nu_{\mathrm{rot}} = 0.098^{+0.052}_{-0.035} d−1^{-1}. The corresponding first-order rotational splitting is approximately 0.025−0.0870.025-0.087 d−1^{-1}. Thus, the observed separation lies close to the lower limit of the range predicted by our models. Although this does not exclude the possibility that ν3\nu_3 and ν4\nu_4 are components of the same rotationally split multiplet, the identification remains tentative. Moreover, no convincing third component of the expected triplet is detected in the periodogram. We therefore did not include ν4\nu_4 as an additional seismic constraint.

Figure 11. Instability parameter, η\eta, as a function of frequency for the seismic model of AI Vel, whose parameters are listed in Table 6. Vertical lines indicate the independent frequencies detected in AI Vel.

For the quadrupole-mode hypothesis, we obtained seismic parameters broadly similar to those derived under the dipole-mode hypothesis (Table 6). However, for several parameters, the shift between the two solutions is comparable to, or larger than, the formal uncertainties inferred from the individual posterior distributions. This comparison demonstrates that uncertainties associated with mode identification introduce a systematic component that is not captured by the statistical posterior uncertainties.

The quadrupole mode that best reproduces the observed frequency has a radial order of n=15n = 15 and corresponds to the gravity mode g15g_{15}. The distribution of its mode inertia is similar to that of the dipole g10g_{10} mode shown in Fig. C.1. Most of its inertia is concentrated in the deep stellar interior, with only a small contribution from the outer layers. The ff-eigenfunction of the quadrupole g15g_{15} mode is also very similar to that of the dipole g10g_{10} mode presented in the bottom panel of Figure 10. Thus, despite their different spherical degrees and radial orders, the dipole and quadrupole modes reproducing the same observed frequency have broadly similar properties in the seismic models of AI Vel.

At present, we find no statistically significant preference for either the dipole or quadrupole identification of ν3\nu_3. Both hypotheses reproduce the available observational constraints comparably well, and the spectroscopic mode identification does not uniquely constrain the spherical degree ℓ\ell. Therefore, the (ℓ,m)=(1,0)(\ell, m) = (1, 0) solution should not be regarded as uniquely preferred. We use the dipole solution below only as a representative seismic model for illustrating the internal structure and pulsational properties of AI Vel.

To investigate the effect of the adopted chemical mixture, we also computed models using the elemental abundance mixture GN93 [53]. The resulting model parameters are listed in Table 6. The overall stellar properties remain similar, although several parameters shift by amounts exceeding their formal posterior uncertainties. The most pronounced difference is found for the stellar age, which decreases from about 1.40 to 1.22 Gyr.

We also investigated the influence of the adopted model atmospheres. Instead of the Vienna model atmospheres, we used the Kurucz models [28]. The results are given in Table 6. This calculation differs from the others because of limitations of the available atmosphere grid. In particular, we were restricted to a single value of the microturbulent velocity, ξt=2\xi_t = 2 km s−1^{-1}, since this was the only value available over the required metallicity range. The resulting seismic models have somewhat higher masses and slightly higher metallicities than those obtained with the Vienna atmospheres. More importantly, however, these models provide a poorer match to the central empirical ff values for both radial modes (see Fig. C.7).

Nevertheless, our tests with different model atmospheres show that the adopted flux derivatives can significantly affect the results of the seismic modelling. The derived parameters should therefore be interpreted with some caution, particularly because our pulsation calculations rely on the frozen-convection approximation and neglect nonlinear pulsation effects. The inferred efficiency of envelope convection in AI Vel is low, suggesting that the limitations of the frozen-convection approximation may have a relatively small impact. On the other hand, AI Vel pulsates in two high-amplitude radial modes, and nonlinear effects may therefore be non-negligible. Such effects could potentially contribute to the discrepancy between the empirical and theoretical ff-parameters. A detailed nonlinear pulsation analysis, however, is beyond the scope of the present study.

Another important aspect of the empirical determination of the ff-parameter is the reliability of its uncertainties. In our standard approach, the uncertainties of ff are obtained as formal errors from the least-squares solution of Eq. (4). The observational uncertainties of the amplitudes and phases enter this procedure primarily as weights and therefore have only a limited effect on the resulting formal uncertainties of the empirical ff-parameters.

To obtain a more realistic estimate of these uncertainties, we additionally performed a Bayesian analysis in which posterior distributions of the empirical ff-parameter were determined. Because these calculations are computationally demanding and require a large number of iterations to achieve convergence, we restricted this test to a representative sample of 10 000 models. The result is nevertheless clear: the formal uncertainties obtained from the least-squares method are systematically about two to three times smaller than those inferred from the Bayesian posterior distributions.

When these larger uncertainties are taken into account, the empirical and theoretical ff-parameters for both radial modes become statistically consistent within the estimated uncertainties. However, the central values of the empirical ff-parameters remain essentially unchanged. Consequently, a systematic offset between the empirical and theoretical values of ∣f∣|f| for the first overtone is still present, although its statistical significance is reduced.

An important conclusion from the comparison of different modelling assumptions is that the resulting systematic variations in the inferred stellar parameters can exceed the formal uncertainties derived from the Bayesian posterior distributions. The latter should therefore be interpreted as statistical uncertainties within a given set of modelling assumptions rather than as estimates of the total uncertainty of the inferred parameters.

Discussion and Conclusions

The case of AI Vel demonstrates that even a limited set of pulsation constraints, consisting of two radial modes and one additional low-degree non-radial mode, can provide detailed insight into the internal structure of a star when combined with Bayesian asteroseismic modelling. Our analysis highlights the importance of pulsation amplitudes and phases, which offer unique constraints on the efficiency of convective energy transport through the empirical ff-parameter.

By fitting both the frequencies of the radial modes and the corresponding non-adiabatic ff-parameters, we were able to tightly constrain key stellar parameters, including mass, initial hydrogen abundance, metallicity, rotational velocity, age, the mixing-length parameter, and the convective core overshooting parameter. Additionally, the presence of a non-radial mode allowed us to further refine the derived stellar parameters and place constraints on the atmospheric microturbulent velocity. The inclusion of this additional frequency substantially reduced the formal posterior uncertainties. However, this improvement depends on the adopted identification of the non-radial mode. In particular, the dipole and quadrupole hypotheses yield slightly different stellar parameters, with differences that can exceed the formal uncertainties obtained for either hypothesis. The quoted posterior uncertainties should therefore not be interpreted as the total uncertainties of the inferred stellar parameters. Nevertheless, the overall seismic picture remains consistent across the two mode-identification hypotheses, supporting the conclusion that AI Vel is an evolved, slowly rotating star.

One unresolved issue concerns the systematic offset between the empirical and theoretical ff-parameter for the first overtone mode. Although the two values become statistically consistent when the larger uncertainties inferred from our Bayesian analysis are taken into account, the central empirical value of ∣f∣|f| remains systematically higher than the theoretical prediction. This residual discrepancy may point to limitations in the atmospheric models, the frozen-convection approximation, or neglected nonlinear pulsation effects, and deserves further investigation.

Another noteworthy result is that the metallicity inferred from seismic modelling, Z∼0.014Z \sim0.014, is close to the solar value and slightly higher than the atmospheric metallicity derived from spectroscopic analysis (Z∼0.011Z \sim0.011). Although the seismic parameter space explored in this work included low-metallicity models, no satisfactory solutions were found for metallicities below the solar value. The significant underabundance of iron, as well as other elements such as Ni, Zn, and Mg, may indicate that diffusion processes are active in AI Vel. In particular, gravitational settling of heavy elements could lead to reduced metal abundances in the stellar atmosphere (e.g., Deal et al. [42]). The relatively slow rotation rate inferred from the seismic modelling is consistent with this interpretation, as weak rotational mixing may allow diffusion processes to operate more efficiently.

AI Vel is another δ\delta Scuti star for which we find relatively low efficiency of convective energy transport in the outer layers. For δ\delta Scuti itself, we derived that the mixing-length parameter αmlt\alpha_{\mathrm{mlt}} is generally smaller than 1.0 (Daszyńska-Daszkiewicz et al. [36]). In our earlier studies, we obtained αmlt≈0.6\alpha_{\mathrm{mlt}} \approx0.6 for BP Peg, 0.4 for AE UMa, and 0.5 for RV Ari. For the Population II star SX Phe, we found αmlt≈1.0\alpha_{\mathrm{mlt}} \approx1.0 (Daszyńska-Daszkiewicz et al. [39]), while for RY Lep the value is αmlt≈0.3\alpha_{\mathrm{mlt}} \approx0.3 (Niewiadomski et al. [71]) as opposed to the solar value ∼1.8\sim1.8.

The modelling of AI Vel also highlights the critical role of stellar opacity data in asteroseismic studies. Within the explored parameter space and adopted modelling assumptions, only models computed with OPAL opacities reproduce the observed properties of AI Vel consistently. Similar results were obtained in our previous studies of other δ\delta Scuti pulsators, including BP Peg, AE UMa, and RV Ari. This behaviour, however, is not restricted to δ\delta Scuti stars. A number of Population II SX Phoenicis variables exhibit the same preference for OPAL opacities. In particular, SX Phe itself, as well as the ω\omega Centauri stars V194, V220, and V225, can only be modelled successfully using OPAL opacity tables (Daszyńska-Daszkiewicz et al. [38]). In contrast, some stars, including δ\delta Scuti itself and the ω\omega Centauri variables NV326 and V237, can be reproduced equally well with OPAL, OP, or OPLIB opacity datasets.

This diversity suggests that the opacity dependence of seismic models is a complex problem without a simple or universal explanation. Nevertheless, we identified a possible correlation between the opacity effect and a star’s position in the HR diagram. Figure 12 presents HR diagrams in which the colour scale represents the difference in the frequency ratio of the radial fundamental and first overtone modes between models computed with different opacity datasets: νp1/νp2(OPAL)−νp1/νp2(OP)\nu_{\mathrm{p1}}/\nu_{\mathrm{p2}}(\mathrm{OPAL})-\nu_{\mathrm{p1}}/\nu_{\mathrm{p2}}(\mathrm{OP}) (upper panel), νp1/νp2(OPAL)−νp1/νp2(OPLIB)\nu_{\mathrm{p1}}/\nu_{\mathrm{p2}}(\mathrm{OPAL})-\nu_{\mathrm{p1}}/\nu_{\mathrm{p2}}(\mathrm{OPLIB}) (middle panel), and νp1/νp2(OPLIB)−νp1/νp2(OP)\nu_{\mathrm{p1}}/\nu_{\mathrm{p2}}(\mathrm{OPLIB})-\nu_{\mathrm{p1}}/\nu_{\mathrm{p2}}(\mathrm{OP}) (bottom panel). Here, νp1\nu_{\mathrm{p1}} and νp2\nu_{\mathrm{p2}} denote the frequencies of the radial fundamental and first overtone modes, respectively. The models shown in the figure were computed assuming X0=0.70X_0 = 0.70, Z=0.010Z = 0.010, αmlt=0.5\alpha_{\mathrm{mlt}} = 0.5, αov=0.1\alpha_{\mathrm{ov}} = 0.1, and an initial rotational velocity of Vrot,0=10 km s−1V_{\mathrm{rot},0} = 10\ \mathrm{km\ s^{-1}}.

Figure 12. HR diagram showing the colour-coded differences in the frequency ratio νp1/νp2\nu_{\mathrm{p1}}/\nu_{\mathrm{p2}} between models computed with different opacity tables. Upper panel: OPAL minus OP, νp1/νp2(OPAL)−νp1/νp2(OP)\nu_{\mathrm{p1}}/\nu_{\mathrm{p2}}(\mathrm{OPAL})-\nu_{\mathrm{p1}}/\nu_{\mathrm{p2}}(\mathrm{OP}). Middle panel: OPAL minus OPLIB, νp1/νp2(OPAL)−νp1/νp2(OPLIB)\nu_{\mathrm{p1}}/\nu_{\mathrm{p2}}(\mathrm{OPAL})-\nu_{\mathrm{p1}}/\nu_{\mathrm{p2}}(\mathrm{OPLIB}). Bottom panel: OPLIB minus OP, νp1/νp2(OPLIB)−νp1/νp2(OP)\nu_{\mathrm{p1}}/\nu_{\mathrm{p2}}(\mathrm{OPLIB})-\nu_{\mathrm{p1}}/\nu_{\mathrm{p2}}(\mathrm{OP}). Here, νp1\nu_{\mathrm{p1}} is the frequency of the radial fundamental mode and νp2\nu_{\mathrm{p2}} the frequency of the first overtone.

Near the zero-age main sequence, the OPAL frequency ratio is smaller than both its OP and OPLIB counterparts. The difference grows during evolution until the star reaches the end of the main sequence phase. From this point, the differences decrease, reaching zero. In more advanced evolutionary phases, the OPAL frequency ratio becomes larger than those from the OP and OPLIB opacity data. The smooth variation of these differences across the HR diagram clearly indicates a systematic offset between the opacity datasets and underscores that the opacity issue in stellar modelling remains unresolved.

Acknowledgements. The work was financially supported by the Polish National Science Centre grant 2023/50/A/ST9/00144. Calculations were carried out using resources provided by Warsaw Centre for Networking and Supercomputing (http://wcss.pl), grant no. 265. 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. This paper includes data collected by the TESS mission. We acknowledge the use of TESS High Level Science Products (HLSP) produced by the Quick-Look Pipeline (QLP) at the TESS Science Office at MIT, which are publicly available from the Mikulski Archive for Space Telescopes (MAST). Funding for the TESS mission is provided by NASA’s Science Mission Directorate. This research made use of Lightkurve, a Python package for Kepler and TESS data analysis. Some of the data presented in this paper were obtained from the Mikulski Archive for Space Telescopes (MAST) at the Space Telescope Science Institute. The specific observations analysed can be accessed via https://doi.org/10.17909/t9-nmc8-f686 and via https://doi.org/10.17909/0cp4-2j79. STScI is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS5–26555. Support to MAST for these data is provided by the NASA Office of Space Science via grant NAG5–7584 and by other grants and contracts.

References

Spectral analysis

In Fig. A.1, we present the relative abundances of 20 chemical elements derived from the HARPS spectra. Compared to solar values, several iron-group elements—most notably Mn, Fe, Ni, and Zn—show a clear underabundance. Most of the remaining elements are also slightly sub-solar. In contrast, a few elements, including O, Zr, and La, exhibit solar or mildly enhanced abundances.

Figure A.1. Chemical abundances of AI Vel derived from the median HARPS spectrum, plotted as a function of atomic number. The horizontal red line indicates the iron abundance, [Fe/H][\mathrm{Fe}/\mathrm{H}].

TESS data

In Fig. B.1, we present the TESS light curve from the first few days of observations in Sector 61, where the variability is clearly visible. The solid yellow line shows the model described by Eq. (1), constructed using frequencies, amplitudes, and phases derived from the Fourier analysis of the combined Sectors 61+62.

Segment of the TESS light curve obtained during the first three days of observations in Sector 61

Figure B.1. Segment of the TESS light curve obtained during the first three days of observations in Sector 61. The solid line represents the Fourier model derived from the combined Sectors 61+62 dataset. The brightness values, mm, were shifted by subtracting the mean brightness, ⟨m⟩\langle m\rangle. TBJD denotes the TESS Barycentric Julian Date, defined as TBJD = BJD − 2457000.0.

Seismic models

In Fig. C.1, we present the mode inertia of a representative seismic model obtained under the dipole-mode hypothesis, labelled (ℓ=1\ell= 1, m=0m = 0) in Table 6. Figure C.2 illustrates the internal structure of the same seismic model. The upper panel shows the Rosseland mean opacity, log⁡κ\log\kappa, together with its derivatives with respect to temperature, ∂log⁡κ/∂log⁡T\partial\log\kappa/ \partial\log T, and density, ∂log⁡κ/∂log⁡ρ\partial\log\kappa/ \partial\log\rho. In the bottom panel, the left-hand axis presents the energy generation rates associated with the pp chain (εpp\varepsilon_{\mathrm{pp}}) and the CNO cycle (εCNO\varepsilon_{\mathrm{CNO}}), while the right-hand axis presents the hydrogen (XX) and helium (YY) abundance profiles.

Mode inertia plots for radial and dipole modes

Figure C.1. Mode inertia for the radial modes (ℓ=0\ell= 0, p1p_1) and (ℓ=0\ell= 0, p2p_2), as well as the dipole mode (ℓ=1\ell= 1, g10g_{10}). Parameters of the seismic model are listed in Table 6.

Opacity, derivatives, energy generation rates, and abundance profiles

Figure C.2. Upper panel: opacity coefficient, log⁡κ\log\kappa, and its derivatives with respect to temperature and density. Bottom panel: left-hand axis – energy generation rates from the pp chain and CNO cycle; right-hand axis – hydrogen (XX) and helium (YY) abundances. Parameters of the seismic model are listed in Table 6.

In Figs. C.3, C.4 and C.5 we present the posterior distributions of selected parameters describing the seismic models of AI Vel. The simulation shown in Fig. C.3 uses as constraints the position of the star in the HR diagram, the empirical ff-parameters, and the frequencies of the two radial modes. In Fig. C.4, an additional constraint from the non-radial mode is included, assuming a dipole-mode identification. Figure C.5 presents the corresponding results obtained under the quadrupole-mode hypothesis.

Figure C.3. Histograms of the adjustable stellar parameters derived from our simulations including two radial modes and OPAL opacity tables. Shown are the distributions of mass, MM, metallicity, ZZ, initial hydrogen abundance, X0X_0, current surface rotational velocity, VrotV_{\mathrm{rot}}, convective mixing length parameter, αmlt\alpha_{\mathrm{mlt}}, and convective core overshooting parameter, αov\alpha_{\mathrm{ov}}. Also included are the amplitudes of relative radius variations due to pulsation in the radial fundamental mode, ε(ν1)\varepsilon(\nu_1), and first overtone, ε(ν2)\varepsilon(\nu_2), as well as the instability parameters for the radial fundamental, η(ν1)\eta(\nu_1), and first overtone, η(ν2)\eta(\nu_2) modes.

Figure C.4. Same as in Fig. C.3, but for simulations including the dipole-mode identification of ν3\nu_3. In addition, histograms for the microturbulent velocity, ξt\xi_t, and stellar age, Age [Gyr], are also shown.

Figure C.5. Same as in Fig. C.3, but for simulations including the quadrupole-mode identification of ν3\nu_3. In addition, histograms for the microturbulent velocity, ξt\xi_t, and stellar age, Age [Gyr], are also shown.

Figure C.6. The same as in Figure 9 but for the quadrupole mode hypothesis.

Figure C.7. The same as in Figure 9 but for Kurucz atmosphere models. Microturbulent velocity was set to ξt=2 km s−1\xi_{\mathrm{t}} = 2\ \mathrm{km}\ \mathrm{s}^{-1}.

References

  1. [1]Aerts, C., Christensen-Dalsgaard, J., & Kurtz, D. W. 2010, AsteroseismologyDOI
  2. [2]Anders, F., Khalatyan, A., Chiappini, C., et al. 2019, A&A, 628, A94
  3. [3]Antoci, V., Cunha, M. S., Bowman, D. M., et al. 2019, MNRAS, 490, 4040
  4. [5]Astropy Collaboration, Price-Whelan, A. M., Lim, P. L., et al. 2022, ApJ, 935, 167
  5. [6]Baglin, A., Breger, M., Chevalier, C., et al. 1973, A&A, 23, 221
  6. [7]Baker, N. & Kippenhahn, R. 1962, ZAp, 54, 114
  7. [8]Baker, N. & Kippenhahn, R. 1965, ApJ, 142, 868
  8. [9]Barac, N., Bedding, T. R., Murphy, S. J., & Hey, D. R. 2022, MNRAS, 516, 2080
  9. [10]Baran, A. S. & Koen, C. 2021, Acta Astron., 71, 113
  10. [11]Bedding, T. R., Murphy, S. J., Hey, D. R., et al. 2020, Nature, 581, 147
  11. [12]Blanco-Cuaresma, S. 2019, MNRAS, 486, 2075
  12. [13]Blanco-Cuaresma, S., Soubiran, C., Heiter, U., & Jofré, P. 2014, A&A, 569, A111
  13. [15]Bowman, D. M. 2017, Amplitude Modulation of Pulsation Modes in Delta Scuti StarsDOI
  14. [16]Bowman, D. M. & Kurtz, D. W. 2018, MNRAS, 476, 3169
  15. [17]Bowman, D. M., Kurtz, D. W., Breger, M., Murphy, S. J., & Holdsworth, D. L. 2016, MNRAS, 460, 1970
  16. [18]Breger, M. 1977, PASP, 89, 339
  17. [19]Breger, M. 1980, ApJ, 235, 153
  18. [20]Breger, M. 1995, in Astronomical Society of the Pacific Conference Series, Vol. 76, GONG 1994. Helio- and Astro-Seismology from the Earth and Space, ed. K. G. Ulrich, E. J. Rhodes, Jr., & W. Dappen, 596
  19. [21]Breger, M. 2000a, Baltic Astronomy, 9, 149
  20. [26]Breger, M. & Pamyatnykh, A. A. 1998, A&A, 332, 958
  21. [27]Campbell, W. W. & Wright, W. H. 1900, ApJ, 12, 254
  22. [28]Castelli, F. & Kurucz, R. L. 2003, in IAU Symposium, Vol. 210, Modelling of Stellar Atmospheres, ed. N. Piskunov, W. W. Weiss, & D. F. Gray, A20
  23. [29]Chevalier, C. 1971, A&A, 14, 24
  24. [30]Claret, A. 2000, A&A, 363, 1081
  25. [31]Colgan, J., Kilcrease, D. P., Magee, N. H., et al. 2015, High Energy Density Physics, 14, 33
  26. [32]Colgan, J., Kilcrease, D. P., Magee, N. H., et al. 2016, ApJ, 817, 116
  27. [33]da Silva, L., Girardi, L., Pasquini, L., et al. 2006, A&A, 458, 609
  28. [34]Daszyńska-Daszkiewicz, J., Dziembowski, W. A., & Pamyatnykh, A. A. 2003, A&A, 407, 999
  29. [35]Daszyńska-Daszkiewicz, J., Dziembowski, W. A., & Pamyatnykh, A. A. 2005, A&A, 441, 641
  30. [36]Daszyńska-Daszkiewicz, J., Pamyatnykh, A. A., Walczak, P., et al. 2021, MNRAS, 505, 88
  31. [37]Daszyńska-Daszkiewicz, J., Szewczuk, W., & Walczak, P. 2024, MNRAS, 532, 1140
  32. [38]Daszyńska-Daszkiewicz, J., Walczak, P., Góra, P., & Szewczuk, W. 2025, MNRAS, 539, 3381
  33. [39]Daszyńska-Daszkiewicz, J., Walczak, P., Pamyatnykh, A., Szewczuk, W., & Niewiadomski, W. 2023a, ApJL, 942, L38
  34. [40]Daszyńska-Daszkiewicz, J., Walczak, P., Pamyatnykh, A. A., & Szewczuk, W. 2022, MNRAS, 512, 3551
  35. [41]Daszyńska-Daszkiewicz, J., Walczak, P., Szewczuk, W., & Niewiadomski, W. 2023b, MNRAS, 526, 1951
  36. [42]Deal, M., Richard, O., & Vauclair, S. 2016, A&A, 589, A140
  37. [43]Ducati, J. R. 2002, VizieR OnLine Data Catalog
  38. [44]Dziembowski, W. A. 1977, Acta Astron., 27, 203
  39. [45]Dziembowski, W. A. & Pamyatnykh, A. A. 2008, MNRAS, 385, 2061
  40. [47]Ferguson, J. W., Alexander, D. R., Allard, F., et al. 2005, ApJ, 623, 585
  41. [48]Gaia Collaboration. 2020, VizieR Online Data Catalog: Gaia EDR3 (Gaia Collaboration, 2020), VizieR On-line Data Catalog: I/350. Originally published in: 2021A&A...649A...1G
  42. [49]Gaia Collaboration. 2022, VizieR Online Data Catalog: Gaia DR3 Part 1. Main source (Gaia Collaboration, 2022), VizieR On-line Data Catalog: I/355. Originally published in: doi:10.1051/0004-6361/2022DOI
  43. [50]Ginsburg, A., Sipőcz, B. M., Brasseur, C. E., et al. 2019, AJ, 157, 98
  44. [51]Gontcharov+, 2018, VizieR On-line Data Catalog: HIP and TGAS stars reddening and extinction (Gontcharov+ 2018), VizieR On-line Data Catalog: II/354. Originally published in: 2017MNRAS.472.3805G
  45. [52]Gratton, L. & Lavanigno, C. J. 1953, ZAp, 32, 69
  46. [53]Grevesse, N. & Noels, A. 1993, in Origin and Evolution of the Elements, ed. N. Prantzos, E. Vangioni-Flam, & M. Casse, 15–25
  47. [54]Guiglion, G., Nardetto, N., Mathias, P., et al. 2013, A&A, 550, L10
  48. [55]Guo, Z., Bedding, T. R., Pamyatnykh, A. A., et al. 2024, MNRAS, 535, 2927
  49. [56]Heiter, U., Kupka, F., van’t Veer-Menneret, C., et al. 2002, A&A, 392, 619
  50. [57]Herbig, G. H. 1949, ApJ, 110, 156
  51. [58]Iglesias, C. A. & Rogers, F. J. 1996, ApJ, 464, 943
  52. [59]Jiang, C. & Gizon, L. 2021, Research in Astronomy and Astrophysics, 21, 226
  53. [60]Jørgensen, B. R. & Lindergren, L. 2005, A&A, 436, 127
  54. [61]Landstreet, J. D. 1998, A&A, 338, 1041
  55. [62]Leroy, B. 2012, A&A, 545, A50
  56. [63]Lightkurve Collaboration, Cardoso, J. V. d. M., Hedges, C., et al. 2018, Lightkurve: Kepler and TESS time series analysis in Python, Astrophysics Source Code Library
  57. [64]Liu, Y.-Q., Chen, X.-D., Wang, S., et al. 2025, Research in Astronomy and Astrophysics, 25, 055019
  58. [65]Lomb, N. R. 1976, Ap&SS, 39, 447
  59. [67]Marshall, D. J., Robin, A. C., Reylé, C., Schultheis, M., & Picaud, S. 2006, A&A, 453, 635
  60. [68]Matsuno, T., Starkenburg, E., Balbinot, E., & Helmi, A. 2024, A&A, 685, A59
  61. [69]Murphy, S. J., Hey, D., Van Reeth, T., & Bedding, T. R. 2019, MNRAS, 485, 2380
  62. [70]Netopil, M. 2017, MNRAS, 469, 3042
  63. [71]Niewiadomski, W., Daszyńska-Daszkiewicz, J., Walczak, P., et al. 2026, ApJ, 1002, 169
  64. [72]Paczyński, B. 1969, Acta Astron., 19, 1
  65. [73]Pamyatnykh, A. A. 1999, Acta Astron., 49, 119
  66. [74]Panda, S. K., Dhanpal, S., Murphy, S. J., Hanosage, S., & Bedding, T. R. 2024, ApJ, 960, 94
  67. [75]Pepe, F., Mayor, M., Delabre, B., et al. 2000, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 4008, Optical and IR Telescope Instrumentation and Detectors, ed. M. Iye & A. F. Moorwood, 582–592
  68. [76]Petersen, J. O. & Hog, E. 1998, A&A, 331, 989
  69. [77]Poleski, R., Soszyński, I., Udalski, A., et al. 2010, Acta Astron., 60, 1
  70. [78]Poretti, E., Clementini, G., Di Fabrizio, L., Bragaglia, A., et al. 2008, The Astrophysical Journal, 685, 947
  71. [79]Poretti, E., Rainer, M., Weiss, W. W., et al. 2011, A&A, 528, A147
  72. [80]Rainer, M., Poretti, E., Misto, A., et al. 2016, A&A, 527, 107
  73. [81]Ricker, G. R., Winn, J. N., Vanderspek, R., et al. 2015, Journal of Astronomical Telescopes, Instruments, and Systems, 1, 014003
  74. [82]Rodrigues, T. S., Bossini, D., Miglio, A., et al. 2017, MNRAS, 467, 1433
  75. [83]Rodríguez, E., López-González, M. J., & López de Coca, P. 2000, A&A, 144, 469
  76. [84]Rogers, F. J. & Nayfonov, A. 2002, ApJ, 576, 1064
  77. [85]Rogers, F. J., Swenson, F. J., & Iglesias, C. A. 1996, ApJ, 456, 902
  78. [86]Różański, T., Niemczura, E., Lemesz, J., Pośłek, N., & Różański, P. 2022, A&A, 659, A199
  79. [87]Samus, N. N., Kazarovets, E. V., Durlevich, O. V., Kireeva, N. N., & Pastukhova, E. N. 2017, Astronomy Reports, 61, 80
  80. [88]Scargle, J. D. 1982, ApJ, 263, 835
  81. [89]Seaton, M. J. 2005, MNRAS, 362, L1
  82. [90]Skiff, B. A. 2014, VizieR Online Data Catalog: Catalogue of Stellar Spectral Classifications (Skiff 2009-), VizieR On-line Data Catalog: B/mk. Originally published in: Lowell Observatory (October 2014)
  83. [91]Stellingwerf, R. F. 1978, AJ, 83, 1184
  84. [93]Walczak, P., Daszyńska-Daszkiewicz, J., Pigulski, A., et al. 2019, MNRAS, 485, 3544
  85. [94]Walczak, P. & Kopacz, A. 2024, MNRAS, 529, 4176
  86. [95]Walraven, T. 1952, Bull. Astron. Inst. Netherlands, 11, 421
  87. [96]Walraven, T. 1955, Bull. Astron. Inst. Netherlands, 12, 223
  88. [97]Walraven, T., Walraven, J., & Balona, L. A. 1992, MNRAS, 254, 59
  89. [98]Yang, T. & Zuo, Z. 2025, ApJ, 995, 188
  90. [99]Zhevakin, S. A. 1963, AR&A, 1, 367
  91. [100]Ziaali, E., Bedding, T. R., Murphy, S. J., Van Reeth, T., & Hey, D. R. 2019, MNRAS, 486, 4348
  92. [101]Zima, W. 2006, A&A, 455, 227
  93. [102]Zima, W. 2008, Communications in Astroseismology, 155, 17
  94. [103]Zwintz, K. 2008, The Astrophysical Journal, 673, 1088
  95. [104]Zwintz, K. 2015, in Proceedings of the International Astronomical Union, Vol. 301, Precision Asteroseismology (Cambridge University Press), 3–14
  96. [105]Zwintz, K., Fossati, L., Ryabchikova, T., Guenther, D. B., & Aerts, C. 2014, Science, 345, 550

Paper details

vlnc:W_55ccq6zj
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

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