Comparing HARPS and Kepler surveys: The alignment of multiple-planet systems

Aims. We study a subset of the planetary population characterized both by HARPS and Kepler surveys. We compare the statistical properties of planets in systems with m.sin i >5-10 M_Earth and R>2 R_Earth. If we assume that the underlying population has the same characteristics, the different detection sensitivity to the orbital inclination relative to the line of sight allows us to probe the planets' mutual inclination. Methods. We considered the frequency of systems with one, two and three planets as dictated by HARPS data. We used Kepler's planetary period and host mass and radii distributions (corrected from detection bias) to model planetary systems in a simple yet physically plausible way. We then varied the mutual inclination between planets in a system according to different prescriptions (completely aligned, Rayleigh distributions and isotropic) and compared the transit frequencies with one, two or three planets with those measured by Kepler. Results. The results show that the two datasets are compatible, a remarkable result especially because there are no tunable knobs other than the assumed inclination distribution. For m.sin i cutoffs of 7-10 M_Earth, which are those expected to correspond to the radius cutoff of 2 R_Earth, we conclude that the results are better described by a Rayleigh distribution with mode of 1 deg or smaller. We show that the best-fit scenario only becomes a Rayleigh distribution with mode of 5 deg if we assume a rather extreme mass-radius relationship for the planetary population. Conclusions. These results have important consequences for our understanding of the role of several proposed formation and evolution mechanisms. They confirm that planets are likely to have been formed in a disk and show that most planetary systems evolve quietly without strong angular momentum exchanges (abridged).