Gravity Simulation

[Deku-Oji]

1st off, this isn't about incorperating it into Celestia (though I'm sure that's a feature we're all waiting for). If this doesn't go here, then please move it, or delete it, or, whatever. In the interest of science fiction (the system I'm building) I went searching for gravity sims. By simply googling "Gravity Simulator" the first site I came to had a good-looking sim (this is the one at http://www.orbitsimulator.com/gravity ). My question is, does anyone know of this program? Is it good and accurate (relative to, say, other freeware)? And if it's not very good, can you suggest a better (also freeware) one? It's mostly because I want to test my orbits (can't have an empire's homeworld crash into their sun, now can I?). Thanks in advance.

[selden]

The author of that program is a member of the Celestia Forum, although he hasn't been very vocal lately.

The math being used in Tony's simulator is a topic of discussion on the BAUT forum. You might take a look at the thread at
http://www.bautforum.com/astronomy/5733 ... ogram.html
It's quite long.

He also has a Forum on the OrbitSimulator site where such things are discussed. http://www.orbitsimulator.com/cgi-bin/yabb/YaBB.pl

[tony873004]

...although he hasn't been very vocal lately...

I guess I should post more often

...My question is, does anyone know of this program? Is it good and accurate (relative to, say, other freeware)?...

It's actually quite good for testing the stability of orbits, in my biased opinion . You wouldn't expect a fellow Californian to write a sloppy program, would you?

But you probably want to know if your system will hold together for billions of years, and it would take decades of simulator time to accurately simulate billions of years of planetary orbits. An unstable system can appear stable for hundreds of thousands of years.

But there's a trick I learned about on the Systemic forum. You can read about it here: http://207.111.201.70/php/forummsg.php?post=140 . The people on this forum try to find evidence of exosolar planets hidden in blocks of radial velocity data. They too want to know if the system is stable, because if it is not, then their newfound planet probably doesn't really exist. The method they use is that they simulate the orbits for 100 years. Any planet whose semi-major changes by less than 1% over this 100 year period is deemed stable.

They have lots of users submitting lots of planetary systems which is probably why they only go to 100 years. Besides, most exosolar planets have very short orbital periods, so 100 years typically gives them thousands of orbits.

I've used their criteria with Gravity Simulator and got identical results. The systems they deem unstable are also shown to be unstable by Gravity Simulator. And since the simulation is running on your own computer, you're not limited to 100 years of simulation. You'd actually want a lot more if you were simulating planets whose orbital periods are decades or centuries, such as Jupiter - Neptune in our system. Just as a guess off the top of my head, I'd say you'd want your slowest (furthest) planet to complete at least a few hundred orbits.

In Gravity Simulator, its easy to do this. The program has a feature called "Output File" under the File menu. This creates a text file of the orbital elements of your planets, at an interval specified by you. Just open it in Excel or any other spreadsheet program and you can graph the semi-major axis, or use the formula (max-min)/mean on your semi-major axis column. Gravity Simulator's beta version has an even nicer version of "Output File". The beta is public, and is available on the web forum: http://www.orbitsimulator.com/cgi-bin/y ... 1176774875

If you use Gravity Simulator, keep the time step low or your results will not be accurate. A time step of 16 seconds is accurate enough to simulate the Newtonian value of the precession of Mercury. And using this timestep should allow you to simulate a few thousand years of your system if you let it run over night with the graphics turned off. And if you're not that patient, you can probably get usable results with the time step being as high as 1024 seconds as long as you don't have any planets closer to their star than Mercury is to our Sun.

Let me know if you need any help setting up your systems.

[ajtribick]

However bear in mind there are configurations which may appear to be stable for hundreds of millions of years and then they go haywire: testing for these requires an awful lot of patience (or extremely good hardware).