RR Lyrae stars with variable mean magnitudes

TL;DR

This study shows that a subset of RR Lyrae stars in the Galactic bulge exhibit genuine long-term mean-magnitude changes, not attributable to photometric errors. By combining extended multi-survey light curves with a novel mean-magnitude-changing Fourier fit and infrared-based SEDs, the authors identify $72$ stars likely driven by astrophysical processes, most consistent with variable extinction from circumstellar or circumbinary dust, while finding no pervasive infrared excess. An incidence rate of about $0.9\%$ after correcting for selection effects implies the effect is rare and has limited impact on RR Lyrae utility as standard candles. The work also uncovers a possible link to binarity, supported by an eclipse-like event and LTTE-associated systems, and outlines several plausible scenarios with clear paths for follow-up observations to pin down the dominant mechanism. Overall, the findings illuminate a new environmental aspect of RR Lyrae stars and highlight the need for multi-wavelength, time-resolved studies to understand dust-related phenomena in old stellar populations.

Abstract

Context. A number of RR Lyrae stars show variable mean magnitudes in the OGLE survey light curves of the Galactic bulge. Hitherto this phenomenon was not studied, as it was generally assumed to be related to problems with the photometry. Aims. We investigate whether the mean magnitude variability of RR Lyrae variables is due to genuine astrophysical phenomena. Methods. We make use of the extended, and in many cases overlapping, light curves from multiple microlensing surveys, to study RR Lyrae stars with apparent mean-magnitude variations. A modified Fourier-series based fitting method is introduced to analyze the light curves showing mean-magnitude variations. Data from infrared surveys are also used to construct spectral energy distributions (SEDs). Results. 72 stars are presented where the mean-magnitude variations are most probably of genuine astrophysical origin, and not the result of problems with the photometry. The ratio of variation between the V and I bands is compatible with variable extinction by dust in most cases, but no infrared excess is detected in the SEDs. The occurrence rate of the phenomenon, after correcting for selection effects, is $\sim0.9\%$ among RR Lyrae variables in the OGLE bulge fields.

Figures (8)

• Figure 1: Summary plot for one of the analyzed stars. Panel a: the original OGLE $I$-band light curve folded with the period of pulsation, with the corresponding Fourier fit overlaid. Panel b: same as ( a), but with the long-term changes removed. Panel c: same as ( a), but for the $V$-band data. Panel d: same as ( b), but for the $V$-band data. Panel e: the $I$-band data after the subtraction of the Fourier light-curve model. The orange dots and lines show the fit of the long-term brightness changes. Panel f: same as ( e), but for the $V$-band light curve. All light-curve points are colored according to the estimated mean-magnitude values at the times of the observations, with lighter and darker points corresponding to brighter and fainter epochs, respectively. Above the panels, the OGLE ID, the pulsation period, the amplitude ($\Delta_\mathrm{IDR}(I)$) and the amplitude ratio ($r_{I,V}=A(I)/A(V)$) of the mean-magnitude changes are given. The complete set of summary plots is available as online material.
• Figure 2: Comparison of RRL mean-magnitude changes between different surveys. Each panel shows the OGLE-IV, OGLE-III, MACHO, EROS-2, MOA, and KMTNet light curves (black, blue, brown, purple, green, and red dots, respectively; see their descriptions in Appendix \ref{['sec:data2']}) after the subtraction of (separate) Fourier light-curve models for each data set. Light curves are offset by arbitrary amounts for clarity. The overlaid orange dots and lines show the mean-magnitude change modeling according to the light-curve fits, similarly to panel ( e) of Fig. \ref{['fig1']}. Above the panels, the OGLE ID, the pulsation period, the amplitude of the mean-magnitude changes, and the measured amplitude ratio $r_{I,V}=A(I)/A(V)$ are given for the corresponding RRL, based on the OGLE-IV data.
• Figure 3: Top panel: period-amplitude (Bailey) diagram of the OGLE bulge RRab sample (grey 2D histogram, shown on a logarithmic scale) and the mean-magnitude-changing RRL variables circles colored according to the measured $I$-band mean-magnitude changes. Middle panel: the distribution of the epoch-independent phase differences ($\phi_{31}$) of the RRab stars. Bottom panel: the dependency of the amplitude ratio $r_{I,V}$ on the amplitude of the $I$-band mean-magnitude changes. RRab stars are marked with circles, and are colored according to their pulsation periods, while the only RRc star is marked with a black square. The horizontal dashed lines show the expected extinction ratios for different $R_{V}$ values of the standard interstellar extinction law 1989ApJ...345..245C.
• Figure 4: The eclipse-like event of OGLE-BLG-RRLYR-09197. Top: The residual light curve of the 2008 observational season (both pulsation and long-term mean-magnitude changes are subtracted). OGLE and MOA observations are shown by blue and green dots, respectively. For clarity, daily average magnitudes for the MOA data are also shown with orange points. Bottom: Same as the top panel, but showing the event between the limits marked by dashed vertical grey lines.
• Figure 5: The distribution of the available $I$-band data points and mean magnitudes in the OGLE survey for RRab stars (black points). Stars of our final sample are marked with red circles. The dashed blue lines denote the region used to calculate the incidence rate of the mean-magnitude changing effect in RRab stars.