Secular Evolution in Mira Variable

Stellar evolution theory predicts that asymptotic giant branch stars undergo a series of short thermal pulses that significantly change their luminosity and mass on timescales of hundreds to thousands of years. These pulses are confirmed observationally by the existence of the short-lived radioisotope technetium in the spectra of some of these stars, but other observational consequences of thermal pulses are subtle, and may only be detected over many years of observations. Secular changes in these stars resulting from thermal pulses can be detected as measurable changes in period if the star is undergoing Mira pulsations. It is known that a small fraction of Mira variables exhibit large secular period changes, and the detection of these changes among a larger sample of stars could therefore be useful in evolutionary studies of these stars. The American Association of Variable Star Observers (AAVSO) International Database currently contains visual data for over 1500 Mira variables. Light curves for these stars span nearly a century in some cases, making it possible to study the secular evolution of the pulsation behavior on these timescales. In this paper, we present the results of our study of period change in 547 Mira variables using data from the AAVSO. We use wavelet analysis to measure the period changes in individual Mira stars over the span of available data. By making linear fits to the period versus time measurements, we determined the average rates of period change, d ln P/dt, for each of these stars. We find non-zero d ln P/dt at the 2-σ significance level in 57 of the 547 stars, at the 3-σ level in 21 stars, and at the level of 6-σ or greater in eight of the 547. The latter eight stars have been previously noted in the literature , and our derived rates of period changes largely agree with published values. The largest and most statistically significant d ln P/dt are consistent with the rates of period change expected during thermal pulses on the AGB. A number of other stars exhibit non-monotonic period change on decades-long timescales, – 2 – the cause of which is not yet known. In the majority of stars, the period variations are smaller than our detection threshold, meaning the available data are not sufficient to unambiguously measure slow evolutionary changes in the pulsation period. It is unlikely that more stars with large period changes will be found among heretofore well-observed …