Objective: Our scientific goal is to understand the processes that formed our solar system and to explore the initial conditions that produced the diverse types of bodies in this system. We would than apply this knowledge to understand the possible formation of other solar systems.
Approach: We have developed a numerical simulation of planetesimal accretion to study the formation of planet-sized bodies from small planetesimals (typically km-sized bodies). This numerical model is the first to treat the simultaneous evolution of a number of adjacent zones at different heliocentric distances and is also the first code to employ both a statistical approach (for early stages of accretion when there are large numbers of small planetesimals) and a discrete body algorithm (to treat the case when there are only a few bodies at the large mass end of the size spectrum) where the statistical approach breaks down. The overall goal of our project is to apply this model to study: i) accretion of planetesimals in both gas-free and gas-rich scenarios; ii) effects of orbital resonances with gas drag present on the accretion process; iii) studies of the timescale for formation of the outer planets and the size distribution of cometesimals (the bodies that supply our present comet population).
Accomplishments: Our code was originally developed for a serial machine, so the first task was to port it to a parallel platform. At the same time, we have been restructuring the logic of the code to work more efficiently in a parallel environment. We tried two approaches to parallelization: one is to use a data parallel method for a Thinking Machines Corp. CM-5 platform, while the other is to apply a message-passing strategy operating under PVM software. The work of this past year has shown that the data parallel approach, while it works on the CM-5 platform, gives little performance improvement over a fast RISC workstation without a complete code rewrite adapting it to massively parallel machines. Given the uncertain gains likely for massively parallel machines in the immediate future following the collapse of Thinking Machines Corp. and the rapidly increasing performance of RISC workstations, we have chosen to focus our efforts on the message passing approach. We have developed prototype message passing software using PVM and plan to further refine this approach in the coming year.
Significance: Understanding how our solar system formed addresses one of the stated goals of NASA and provides insight into the conditions necessary for the formation of planets. Applying this knowledge to other stellar systems will enable us to better assess the frequency of other solar systems. This type of theoretical work complements the observational searches that are underway to detect other solar systems.
Status/Plans: We are refining the PVM software to optimally treat the message passing processes and are working to improve the efficiency of the computational processes relative to the message passing procedures. The goal is to have the time spent dealing with message passing be a small fraction of the total execution time. We are also planning to develop more effective graphics to visually demonstrate how planets form from a protostellar cloud.
Point of Contact:
Dr. Donald R. Davis
Planetary Science Institute
drd@psi.edu
520-622-6300