Statistics – Computation
Scientific paper
May 2009
adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=2009aas...21443204l&link_type=abstract
American Astronomical Society, AAS Meeting #214, #432.04; Bulletin of the American Astronomical Society, Vol. 41, p.703
Statistics
Computation
Scientific paper
Radiation transport is critical not only for analysis of astrophysical objects but also for the dynamical transport of energy within. Increased fidelity and dimensionality of the other components of such models requires a similar improvement in the radiation transport.
Modern astrophysical simulations can be large enough that the values for a single variable for the entire computational domain cannot be stored on a single compute node. The natural solution is to decompose the physical domain into pieces with each node responsible for a single sub-domain. Using localized plus "ghost" zone data works well for problems like explicit hydrodynamics or nuclear reaction networks with modest impact from inter-process communication. Unfortunately, radiation transport is an inherently non-local process that couples the entire model domain together and efficient algorithms are needed to conquer this problem.
In this poster, I present the early development of a new parallel, 3-D transport code using ray tracing to formally solve the transport equation across numerically decomposed domains. The algorithm model takes advantage of one-sided communication to develop a scalable, parallel formal solver. Other aspects and future direction of the parallel code development such as scalability and the inclusion of scattering will also be discussed.
No associations
LandOfFree
New Algorithms for Large-scale 3D Radiation Transport does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with New Algorithms for Large-scale 3D Radiation Transport, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and New Algorithms for Large-scale 3D Radiation Transport will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-1107510