A method for the detection of planetary transits in large time series data sets

We present a fast, efficient, and easy-to-apply computational method for the detection of planetary transits in large photometric data sets. The code was specifically produced to analyze an ensemble of 21,950 stars in the globular cluster 47 Tue for the photometric signature indicative of a transiting hot Jupiter planet, the results of which are the subject of a separate paper. Using cross-correlation techniques and Monte Carlo-tested detection criteria, each photometric time series is compared with a database of transit models of appropriate depth and duration. The algorithm recovers transit signatures with high efficiency while maintaining a low false-detection probability, even in rather noisy data. The code has been optimized, and with a 3 GHz machine is capable of analyzing and producing candidate transits for 10,000 stars in 24 hr. We illustrate our algorithm by describing its application to our large 47 Tue data set, for which the algorithm produced a weighted-mean transit recoverabilty spanning 85%-25% for orbital periods of 1 -16 days (half the temporal span of the data set), despite gaps in the time series caused by the weather and observing the duty cycle. The code is easily adaptable and is currently designed to accept time series data produced using difference imaging analysis.