| Our latest work is featured on the cover of the January 2000 issue of Icarus! Click the image to the right for the high-resolution version. The paper is available here (your site needs to have a subscription to IDEAL). The revised preprint is still available below. |
|
Planet formation in the inner Solar System is thought to proceed in four stages:
To date most numerical studies have been restricted to the third stage and relied on statistical methods or small computational domains to make the problem tractable. We have designed a numerical code capable of modeling both the third and fourth stage, that is, the transition from runaway growth to the regime of the infrequent large impacts leading to the final formation of the planets.
In order to start with planetesimals of an "interesting" size (about 100 km) and to model the entire inner Solar System disk (from Mercury to Mars, pictured here using density contours), millions of planetesimals are required. This is many orders of magnitude more than any previous study has attempted. Furthermore, it takes as long as a million years to form protoplanets, and detecting collisions among the planetesimals requires timesteps of days.
We are approaching this goal of large computational domains, high spatial and temporal resolution needs, and long integration time. We have modified our stable cosmology code to search for particle collisions and to optimize the orbital integration for the central force field of the Sun. The results shown below are from a 1000-year simulation on the NASA Goddard SGI T3E using 128 nodes for about 200 wallclock hours. The run consisted of 1 million planetesimals in a thin cold disk around the Sun, and included the effects of the giant planets. In 100 years Jupiter has already begun to carve out resonance structure in the disk. Meanwhile, particles in the Earth-Mars region have started to agglomerate on the way to building planets.
We are currently developing and testing a perturbative technique to increase the speed of our integrations by another factor of 10 to 100. However, to perform simulations of 10 million planetesimals for 10 million years, we need roughly a further 100X improvement in the combined speed of computers and algorithms.
HINT: click on the images to see animations!
|
This material has been published in Icarus 143, 45-59 (2000), the only definitive repository of the content that has been certified and accepted after peer review. Copyright and all rights therein are retained by Academic Press. This material may not be copied or reposted without explicit permission.
Direct Large-Scale N-Body Simulations of Planetesimal Dynamics
Copyright © 2000 by Academic Press. Available through IDEAL.
D. C. Richardson,
T. Quinn,
J. Stadel, and
G. Lake
University of Washington
25 manuscript pages including 1 table
14 figures including 1 in color
ABSTRACT (plain text)
The following MPEG movies show the evolution of Model B in local density and particle orbital elements (snapshots from these movies are shown in Figure 3 of the text). The first 100 yr of evolution are sampled at a rate of one frame per year; the remaining frames are sampled at 10 yr intervals.
Downloading options:
Figures 3 & 10 are included only in greyscale format in the "full text with figures" versions. If you would like the (big!) color figures, grab the compressed files at the end of this list.
| Last modified: Oct 6, 2000 |
|