Thomas Lebzelter

Random forest automated supervised classification of Hipparcos periodic variable stars

We present an evaluation of the performance of an automated classification of the Hipparcos periodic variable stars into 26 types. The sub-sample with the most reliable variability types available in the literature is used to train supervised algorithms to characterize the type dependencies on a number of attributes. The most useful attributes evaluated with the random forest methodology include, in decreasing order of importance, the period, the amplitude, the V − I colour index, the absolute magnitude, the residual around the folded light-curve model, the magnitude distribution skewness and the amplitude of the second harmonic of the Fourier series model relative to that of the fundamental frequency. Random forests and a multistage scheme involving Bayesian network and Gaussian mixture methods lead to statistically equivalent results. In standard 10-fold cross-validation (CV) experiments, the rate of correct classification is between 90 and 100 per cent, depending on the variability type. The main mis-classification cases, up to a rate of about 10 per cent, arise due to confusion between SPB and ACV blue variables and between eclipsing binaries, ellipsoidal variables and other variability types. Our training set and the predicted types for the other Hipparcos periodic stars are available online.

Long period variables in 47 Tuc: direct evidence for lost mass

Received ; accepted Abstract. We have identified 22 new variable red giants in 47 Tuc and dete rmined periods for another 8 previously known variables. All red giants redder than V-Ic = 1.8 are variable at the limits of our detection threshold, wh ich corresponds to δV≈ 0.1 mag. This colour limit corresponds to a luminosity log L/L⊙=3.15 and it is considerably below the tip of the RGB at log L/L⊙=3.35. Linear non-adiabatic models without mass loss on the giant branch can not reproduce the observed PL laws for the low amplitude pulsators. Models that have undergone mass loss do reproduce the observed PL relations and they show that mass loss of the order of 0.3 M⊙ occurs along the RGB and AGB. The linear pulsation periods do not agree well with the observed periods of the large amplitude Mira variables, which pulsate in the fundamental mode. The solution to this problem appears to be that the nonlinear pulsation periods in these l ow mass stars are considerably shorter than the linear pulsation periods due to a rearrangement of stellar structure caused by the pulsation. Both observations and theory show that stars evolve up the RGB and first part of the AGB pulsating in low orde r overtone modes, then switch to fundamental mode at high luminosities.

Tracing the Development of Dust around Evolved Stars: The Case of 47 Tuc

We observed mid-infrared (7.5-22 μm) spectra of asymptotic giant branch (AGB) stars in the globular cluster 47 Tuc with the Spitzer Space Telescope, and we find significant dust features of various types. A comparison of the characteristics of the dust spectra with the location of the stars in a (log P, K)-diagram shows that the dust mineralogy and the position on the AGB are related. A 13 μm feature is seen in spectra of low-luminosity AGB stars. More luminous AGB stars show a broad feature at 11.5 μm. The spectra of the most luminous stars are dominated by the amorphous silicate bending vibration centered at 9.7 μm. For 47 Tuc AGB stars, we conclude that early on the AGB dust (consisting primarily of Mg, Al, and Fe oxides) is formed. With further AGB evolution, amorphous silicates become the dominant species.

AGB stars of the intermediate-age LMC cluster NGC 1846 - II. Dredge up along the AGB

Aims. We investigate the change in the surface abundance of 12 C during the evolution along the AGB, aiming to constrain third dredge-up models. Methods. High-resolution, near-infrared spectra of a sample of AGB stars in the LMC cluster NGC 1846 were obtained. A cluster sample ensures a high level of homogeneity with respect to age, metallicity, and distance. The C/O ratio and the ratio of 12 C/ 13 Cw ere measured and compared with our evolutionary models. Results. For the first time, we show the evolution of the C/ Oa nd 12 C/ 13 C ratios along a cluster AGB. Our findings allow us to check the reliability of the evolutionary models and, in particular, the efficiency of the third dredge up. The increase in both C/ Oa nd 12 C/ 13 C in the observed O-rich stars is reproduced by the models well. However, the low carbon isotopic ratios of the two C-stars in our sample indicate the late occurrence of moderate extra mixing. The extra mixing affects the most luminous AGB stars and is capable of increasing the abundance of 13 C, while leaving unchanged the C/O ratio, which has been fixed by the cumulative action of several third dredge-up episodes. We find indications that the F abundance also increases along the AGB, supporting an in situ production of this element.

Lithium abundances along the red giant branch: FLAMES-GIRAFFE spectra of a large sample of low-mass bulge stars

Context: A small number of K-type giants on the red giant branch (RGB) is known to be very rich in lithium (Li). This fact is not accounted for by standard stellar evolution theory. The exact phase and mechanism of Li enrichment is still a matter of debate. Aims: Our goal is to probe the abundance of Li along the RGB, from its base to the tip, to confine Li-rich phases that are supposed to occur on the RGB. Methods: For this end, we obtained medium-resolution spectra with the FLAMES spectrograph at the VLT in GIRAFFE mode for a large sample of 401 low-mass RGB stars located in the Galactic bulge. The Li abundance was measured in the stars with a detectable Li 670.8 nm line by means of spectral synthesis with COMARCS model atmospheres. A new 2MASS (J-K) - Teff calibration from COMARCS models is presented in the Appendix. Results: Thirty-one stars with a detectable Li line were identified, three of which are Li-rich according to the usual criterion (

The pulsation of AGB stars in the Magellanic Cloud clusters NGC 1978 and 419

\log\epsilon({\rm Li})>1.5

Multi-colour light variation of AGB stars observed with ISO

). The stars are distributed all along the RGB, not concentrated in any particular phase of the red giant evolution (e.g. the luminosity bump or the red clump). The three Li-rich stars are clearly brighter than the luminosity bump and red clump, and do not show any signs of enhanced mass loss. Conclusions: We conclude that the Li enrichment mechanism cannot be restricted to a clearly defined phase of the RGB evolution of low-mass stars (

Velocity Observations of Multiple-Mode Asymptotic Giant Branch Variable Stars

M\sim1M_{\sun}

Technetium and the third dredge up in AGB stars. I. Field stars

), contrary to earlier suggestions from disk field stars.

Low-mass lithium-rich AGB stars in the Galactic bulge: evidence for Cool Bottom Processing? ?

The intermediate-age Magellanic Cloud clusters NGC 1978 and 419 are each found to contain substantial numbers of pulsating asymptotic giant branch (AGB) stars, both oxygen rich and carbon rich. Each cluster also contains two pulsating AGB stars which are infrared sources with a large mass-loss rate. Pulsation masses have been derived for the AGB variables, from the lowest luminosity O-rich variables to the most evolved infrared sources. It is found that the stars in NGC 1978 have a mass of 1.55 M ⊙ early on the AGB while the NGC 419 stars have a mass of 1.87 M ⊙ early on the AGB. These masses are in good agreement with those expected from the cluster ages determined by main-sequence turn-off fitting. Non-linear pulsation models fitted to the highly evolved AGB stars show that a substantial amount of mass-loss has occurred during the AGB evolution of these stars. An examination of the observed mass-loss on the AGB, and the AGB tip luminosities, shows that in both clusters the mass-loss rates computed from the formula of Vassiliadis & Wood reproduce the observations reasonably well. The mass-loss rates computed from the formula of Blocker terminate the AGB in both clusters at a luminosity which is much too low.