Young Planet Population Predictions from Protoplanetary Disk Gaps

Detecting young planets in protoplanetary disks is essential to infer where and when planets form and how massive they are, putting stringent constraints on planet formation theory. Despite thousands of exoplanets having been discovered, only a few of them are around young stars within 10 million years old. Substructures induced by young planets in protoplanetary disks can be used to infer the potential young planets’ properties. Hydrodynamical simulations have been used to study the relationships between the planet’s properties and these disk features. One approach is to azimuthally average simulation results to derive some linear relationships between the gap width/depth and the planet mass, but this loses information on asymmetric features. To preserve 2D information, we develop convolutional neural networks (CNNs) PGNets to rapidly and directly infer the planet mass from radio dust continuum images. We use these tools to study the young planet population