We present a probabilistic generative approach for constructing topographic maps of light curves from eclipsing binary stars. The model defines a low-dimensional manifold of local noise models induced by a smooth non-linear mapping from a low-dimensional latent space into the space of probabilistic models of the observed light curves. The local noise models are physical models that describe how such light curves are generated. Due to the principled probabilistic nature of the model, a cost function arises naturally and the model parameters are fitted via MAP estimation using the Expectation-Maximisation algorithm. Once the model has been trained, each light curve may be projected to the latent space as the the mean posterior probability over the local noise models. We demonstrate our approach on a dataset of artificially generated light curves and on a dataset comprised of light curves from real observations.

Before the age of modern photographic and CCD observations alpha Cassiopeiae was labelled as a variable star, though this variability has not been seen with modern instrumentation. We present an analysis of 3 years of high precision space-based photometric measurements of the suspected variable star alpha Cassiopeiae, obtained by the broad band Solar Mass Ejection Imager (SMEI) instrument on board the Coriolis satellite. Over the 3 years of observations the star appears to not show any significant variability. Also, data from the Hipparcos epoch photometry annex shows no significant variability.

Aims: As g-mode pulsators, gamma-Doradus-class stars may naively be expected to show a large number of modes. Taking advantage of the long photometric time-series generated by the Solar Mass Ejection Imager (SMEI) instrument, we have studied the star gamma Doradus to determine whether any other modes than the three already known are present at observable amplitude. Methods: High-precision photometric data from SMEI taken between April 2003 and March 2006 were subjected to periodogram analysis with the PERIOD04 package. Results: We confidently determine three additional frequencies at 1.39, 1.87, and 2.743 cycles per day. These are above and beyond the known frequencies of 1.320, 1.364, and 1.47 cycles per day. Conclusions: Two of the new frequencies, at 1.39 and 1.87 cycles per day, are speculated to be additional modes of oscillation, with the third frequency at 2.743 cycles per day a possible combination frequency.

Sun-like oscillations, that is p-modes excited stochastically by convective noise, have now been observed in a number of Red Giant stars. Compared to those seen in the Sun, these modes are of large amplitude and long period, making the oscillations attractive prospects for observation. However, the low Q-factor of these modes, and issues relating to the rising background at low frequencies, present some interesting challenges for identifying modes and determining the related asteroseismic parameters. We report on the analysis procedure adopted for peak-bagging by our group at Birmingham, and the techniques used to robustly ensure these are not a product of noise. I also show results from a number of giants extracted from multi-year observations with the SMEI instrument.

We present an analysis of 4.5 years of high precision (0.1%) space-based photometric measurements of the Cepheid variable Polaris, obtained by the broad band Solar Mass Ejection Imager (SMEI) instrument on board the Coriolis satellite. The data span from April 2003 to October 2007, with a cadence of 101 minutes and a fill factor of 70%. We have measured the mean peak to peak amplitude across the whole set of observations to be 25 mmag. There is, however, a clear trend that the size of the oscillations has been increasing during the observations, with peak to peak variations less than 22 mmag in early 2003, increasing to around 28 mmag by October 2007, suggesting that the peak to peak amplitude is increasing at a rate of 1.39 \pm 0.12 mmag yr^{-1}. Additionally, we have combined our new measurements with archival measurements to measure a rate of period change of 4.90 \pm 0.26 s yr^{-1} over the last 50 years. However, there is some suggestion that the period of Polaris has undergone a recent decline, and combined with the increased amplitude, this could imply evolution away from an overtone pulsation mode into the fundamental or a double pulsation mode depending on the precise mass of Polaris.

Aims: From observations of the K4III star Beta UMi we attempt to determine whether oscillations or any other form of variability is present. Methods: A high-quality photometric time series of approximately 1000 days in length obtained from the SMEI instrument on the Coriolis satellite is analysed. Various statistical tests were performed to determine the significance of features seen in the power density spectrum of the light curve. Results: Two oscillations with frequencies 2.44 and 2.92 microhertz have been identified. We interpret these oscillations as consecutive overtones of an acoustic spectrum, implying a large frequency spacing of 0.48 microhertz. Using derived asteroseismic parameters in combination with known astrophysical parameters, we estimate the mass of Beta UMi to be 1.3 +/- 0.3 solar masses. Peaks of the oscillations in the power density spectrum show width, implying that modes are stochastically excited and damped by convection. The mode lifetime is estimated at 18 +/- 9 days.

We present new results on oscillations of the K1.5 III giant Arcturus (alpha Boo), from analysis of just over 2.5 yr of precise photometric observations made by the Solar Mass Ejection Imager (SMEI) on board the Coriolis satellite. A strong mode of oscillation is uncovered by the analysis, having frequency 3.51+/-0.03 micro-Hertz. By fitting its mode peak, we are able offer a highly constrained direct estimate of the damping time (tau = 24+/-1 days). The data also hint at the possible presence of several radial-mode overtones, and maybe some non-radial modes. We are also able to measure the properties of the granulation on the star, with the characteristic timescale for the granulation estimated to be 0.50+/-0.05 days.