The International Deep Planet Survey: II. The frequency of directly imaged giant exoplanets with stellar mass

Context. Radial velocity and transit methods are effective for the study of short orbital period exoplanets but they hardly probe objects at large separations for which direct imaging can be used. Aims. We carried out the international deep planet survey of 292 young nearby stars to search for giant exoplanets and determine their frequency. Methods. We developed a pipeline for a uniform processing of all the data that we have recorded with NIRC2/Keck II, NIRI/Gemini North, NICI/Gemini South, and NACO/VLT for 14 yr. The pipeline first applies cosmetic corrections and then reduces the speckle intensity to enhance the contrast in the images. Results. The main result of the international deep planet survey is the discovery of the HR 8799 exoplanets. We also detected 59 visual multiple systems including 16 new binary stars and 2 new triple stellar systems, as well as 2279 point-like sources. We used Monte Carlo simulations and the Bayesian theorem to determine that 1.05^+2.80_-0.70 % of stars harbor at least one giant planet between 0.5 and 14 M_J and between 20 and 300 AU. This result is obtained assuming uniform distributions of planet masses and semi-major axes. If we consider power law distributions as measured for close-in planets instead, the derived frequency is 2.30^+5.95_-1.55%, recalling the strong impact of assumptions on Monte Carlo output distributions. We also find no evidence that the derived frequency depends on the mass of the hosting star, whereas it does for close-in planets. Conclusions. The international deep planet survey provides a database of confirmed background sources that may be useful for other exoplanet direct imaging surveys. It also puts new constraints on the number of stars with at least one giant planet reducing by a factor of two the frequencies derived by almost all previous works.