TT Arietis: New approach to the analysis of quasi-periodic oscillations

TL;DR

TT Arietis exhibits prominent quasi-periodic oscillations (QPOs) superposed on a dominant superhump signal during its high state. The authors analyze 361.2 hours of uninterrupted MOST data with a Lomb-Scargle-based frequency-group approach, segmenting the light curve into 0.1-day blocks and tracking persistently evolving frequencies. They identify 160 frequency groups (P ≥ 0.10) spanning 14–53 minutes, with two main peaks at 18.5 and 33 minutes, and show that QPOs persist for hours and can occur concurrently at multiple frequencies; a subset of groups is validated as real against randomized data. While flickering can generate false positives, especially at lower power thresholds, the majority of detected groups cannot be explained by stochastic processes, indicating genuine QPO activity with systematic frequency evolution and no clear tie to the negative superhump phase.

Abstract

Context. TT Arietis (TT Ari) is a nova-like cataclysmic variable of the VY Scl subtype with light-curve variations on multiple timescales. In addition to the superhump modulation, quasi-periodic oscillations (QPOs) have been found. Aims. Our aim is to determine the occurrence, strength, and variability of QPOs in TT Ari based on more complete data than in previous works. Methods. The data were obtained during the high state of TT Ari in October 2012 by the MOST space telescope, covering a total of 361.2 hours of continuous observation. We searched for frequencies over subsets of time using a Fourier-like power spectrum and then added the frequencies together, forming groups. Results. Our method has revealed QPOs that occur in "frequency groups", which are events with a continuous oscillation of similar, constant or slowly variable frequency. We report a total of 160 frequency groups in the period range between 14 and 53 minutes (27 and 98 days-1), with two peaks in the power spectrum at 18.5 and 33.8 minutes (42.5 and 77.5 days-1). The duration of these frequency groups varies between 0.72 and 7.5 hours (average 2.8 hours) revealing between 3 and 18 complete cycles in the light curve. Most of them show significant frequency variations over the course of their duration. Sometimes two frequency groups occur simultaneously. An analysis with randomised data confirms that stochastic processes can only explain a fraction of the QPOs found. The occurrence of QPOs appears not to be related to the superhump phase.

Figures (11)

• Figure 1: Light-curve segment of TT Ari, showing a typical time interval in its high state. The top panel shows a three-day span from October 2012, while the middle panel shows segment of a 0.4 days (9.6 hours) within the same time interval. The bottom panel shows the residual data from the spline interpolation. Removing the superhump modulation reveals other brightness variations, including QPOs.
• Figure 2: Power spectrum of the frequency range, calculated for certain time periods. The dashed line corresponds to the power threshold of 0.1. Top panel: Period with one QPO. This corresponds to a specific frequency group visible in Fig. \ref{['fig:figure3']} at starting date JD 206.7. Bottom panel: Time period without QPOs, appearing as a gap in Fig. \ref{['fig:figure3']} around JD 217.7.
• Figure 3: Frequency groups throughout the entire 15 days of observations. Each panel corresponds to a three-day segment. Each line in a group represents a frequency peak in the power spectrum, as explained in sect. \ref{['sec:frequency groups']}. The light grey region at JD 206.7 corresponds to the power spectrum analysed in Fig. \ref{['fig:figure2']} (top panel), while the filled region at JD 217.7 represents the bottom panel of Fig. 2.
• Figure 4: Multiple time intervals from TT Ari’s light curve, and a sinusoidal curve overlay. Each panel shows a different frequency group. The curve is made using a combination of five significant frequencies extracted from the power-spectrum for that segment. Here, F1 corresponds to the most relevant frequency in the power spectrum, with each following frequency having diminished impact on the curve.
• Figure 5: Group duration histograms for $P \ge 0.10$. Light grey: Duration histogram for the 160 groups of the residual data. The average duration is 2.8 hours, with few groups lasting more than 7.5 hours. Dark grey: Duration histogram for randomised data. Random data create fewer and shorter groups.