Search for: All records

Abstract This study probes the lithosphere‐asthenosphere system beneath 155 Ma Pacific seafloor using teleseismic S‐to‐p receiver functions at the Pacific Lithosphere Anisotropy and Thickness Experiment project ocean‐bottom‐seismometers. Within the lithosphere, a significant velocity decrease at 33–50 km depth is observed. This mid‐lithospheric discontinuity is consistent with the velocity contrast between the background mantle and thin, trapped layers of crystallized partial melt, in the form of either dolomite or garnet granulite. These melts possibly originated from deeper asthenospheric melting beneath the flanks of spreading centers, and were transported within the cooling lithosphere. A positive velocity increase of 3%–6% is observed at 130–155 km depth and is consistent with the base of a layer with partial melt in the asthenosphere. A shear velocity decrease associated with the lithosphere‐asthenosphere boundary at 95–115 km depth is permitted by the data, but is not required. 

Generative models for 3D shapes represented by hierar- chies of parts can generate realistic and diverse sets of out- puts. However, existing models suffer from the key practi- cal limitation of modelling shapes holistically and thus can- not perform conditional sampling, i.e. they are not able to generate variants on individual parts of generated shapes without modifying the rest of the shape. This is limiting for applications such as 3D CAD design that involve adjust- ing created shapes at multiple levels of detail. To address this, we introduce LSD-StructureNet, an augmentation to the StructureNet architecture that enables re-generation of parts situated at arbitrary positions in the hierarchies of its outputs. We achieve this by learning individual, probabilis- tic conditional decoders for each hierarchy depth. We eval- uate LSD-StructureNet on the PartNet dataset, the largest dataset of 3D shapes represented by hierarchies of parts. Our results show that contrarily to existing methods, LSD- StructureNet can perform conditional sampling without im- pacting inference speed or the realism and diversity of its outputs.