The ongoing High Accuracy Radial velocity Planet Search ( HARPS ) has found that 30-50 % of GK dwarfs in the solar neighborhood host planets with M _ { pl } \lesssim M _ { Nep } in orbits of P \leq 50 days .
At first glance , this overall occurrence rate seems inconsistent with the planet frequency measured during Q0-Q2 of the Kepler Mission , whose 1,235 detected planetary candidates imply that \sim 15 % of main sequence dwarfs harbor short-period planets with R _ { pl } < 4 R _ { \oplus } .
A rigorous comparison between the two surveys is difficult , however , as they observe different stellar populations and measure different planetary properties .
Here we report the results of a Monte Carlo study that can account for this discrepancy via plausible distributions of planetary compositions .
We find that a population concurrently consisting of ( 1 ) dense silicate-iron planets and ( 2 ) low-density gas-dominated worlds provides a natural fit to the current data .
In this scenario , the fraction of dense planets decreases with increasing mass , from f _ { rocky } = 90 \% at M = 1 M _ { \oplus } to f _ { rocky } = 10 \% at M = M _ { Nep } .
Our best fit population has a total occurrence rate of 40 % for 2 \leq P \leq 50 days and 1 \leq M \leq 17 M _ { \oplus } , and is characterized by simple power-law indices of the form N ( M ) dM \propto M ^ { \alpha } dM and N ( P ) dP \propto P ^ { \beta } dP with \alpha = -1.0 and \beta = 0.0 .
Our model population therefore contains four free parameters and is readily testable with future observations .
Furthermore , our model ’ s insistence that at least two distinct types of planets must exist in the survey data indicates that multiple formation mechanisms are at work to produce the population of planets commonly referred to as “ super-Earths ” .