Navamita Ray
LANL
CAI1
Biography
Navamita Ray is a staff scientist in the co-design team in CCS-7 with expertise in meshing and numerical methods. She has worked on developing various mesh-based algorithms such as surface reconstruction, optimization and adaptivity targeted towards high-order numerical methods as well as designing unstructured mesh data structures for parallel computing. At LANL, she has worked on developing software libraries for upcoming exascale machines such as the runtime library FleCSI and mesh-to-mesh data remapping library, Portage, to perform inline remapping and link LANL application codes. She is involved in developing meshing capabilities in the Crosslink library, which is LANL’s in-house mesh generation code as well as exploring machine learning for mesh generation problems.
SRP Project Title
Automatic Shape Matching for Complex 3D CAD Models
Topical Areas
Applied Computer Science; Applied Mathematics; Artificial Intelligence and Intelligent Systems
Abstract
Automatic hexahedral mesh generation for general 3D shapes is an open problem. This internship will explore machine learning methods for automating shape recognition within complex models that would aid in mesh generation via application of domain knowledge specific templates.
Desired Skills
Software skills such as familiarity with using unix commands, copying files to and fro from remote systems, etc. Programming skills such as a basic knowledge of compiled (C/C++) and interpreted languages (python). ML skills such as basic knowledge of supervised learning and other standard ML approaches.
Lightning Talk Title
Automatic Shape Matching for Complex 3D CAD Models
Keywords
Geometric modeling, shape matching, machine learning