SyMBA GPU
From HP-SEE Wiki
Contents |
General Information
- Application's name: Study of the initial formation stages of the solar system
- Application's acronym: SyMBA_GPU
- Scientific domain: Space Science
- Contact person: Kleomenis Tsiganis, tsiganis<>auth.gr
- Main Developers: Kleomenis Tsiganis, Paschalis Korosoglou, Aristotle University of Thessaloniki/Physics Department
- Co-developers: Harold Levison
- Allocation period: 2013
- Web site:
Objectives of the computing project
To develop and evaluate the efficiency of GPU-based codel on N-body collisional dynamicsThe code should supports scales (or, time steps) and recursion, according to a symplectic schemethat ensures better preservation of fundamental quantities. The code will be applied to the problem of planet growth (from embryoto giant core) where, by including also non-gravitational forces, we will try to understand the effect of thenumber and size distribution of bodies on the formation time-scale
Application's description
It's a N-body gravitational code but with some special handling of close encounters between bodies. This is because it is dedicated to planetary-system like applications where, contrary to galaxies and other stellar systems, encounters dominate the resulting evolution. So, essentially it's a code which works as a symplectic N-body when no bodies are encountering each other but then switches to a different algorithm (a recursive symplectic scheme)for bodies that are in an encounter (typically a "big" and some "smaller" body). The effect of surrounding gas nebula can be added as "extra" forces, to evaluate the role of pressure effects on enhancing the local surface density of solids.
Results expected in the allocation period
On the technical side, we expect to develop a fully symplectic (i.e. energy preserving)code that will be able to handle large, collisional, N-body systems much more efficiently than before. This is of crucial importance to some astrophysical problems, such as the planet-formation problem, where large N's (~100,000) should be combined to high accuracy and huge number of time-steps (~10-100 millions).
On the scientific side, we hope to obtain realistic simulations of planet formation, starting from an "embryos" state (many small planetary embryos plus a huge number of even smaller "planetesimals") of fully interacting bodies, in the presence of additional pressure effects, caused by a surrounding gas nebula (epoch of planet formation). Our goal is to test the results of the code, comparing with previous analytical and numerical results before starting "production" simulations on testing how the growth rate of a giant planet cores (such as the ones in the interiors of the giant planets) is affected by the number density and size distribution of the initial population
