Suggestion: Rigid-Body Linkages

[thalamus]

It'd be nice to have the ability to connect a pair of bodies with a rigid-body linkage (something like an infinitely-stiff bar). This would make modeling things like nonspherical gravity sources and spin-orbit synchronization much easier, and (although I'm not sure) I don't imagine it would be too difficult to program.

[Greenleaf]

That is not a bad suggestion, and it is actually being tested right now for different purposes.
This video shows an orbital ring made up of 100 ( i think it was 100) bodies which are linked by constraints to two neighbors and set in rotation so the entire ring is under tension. Such a ring can be unstable, if it doesn't have thrusters to keep it in the exact right orbit, which is seen here.


http://youtu.be/nNRR69QqO6A


The tether system, as it is, is in a sense springs with specific material properties, which makes it computationally rather expensive, which is why I started testing with constraints.


But your suggestion was not a string, but a stick constraint so you could make more complicated gravity fields, and that is something which does make a lot of sense actually. Could clearly show how hard stable orbits can be around real world objects with more complicated mass distribution.


Another way of handling with that could be to extend the gravity calculation, so it would not only take a distance from a gravity source to find the gravitational strength, but could also consider the direction. This relates to multipole expansion http://en.wikipedia.org/wiki/Multipole_expansion and is also relevant to be able to calculate for detailed tree based gravity.

[thalamus]

A spherical harmonic/multipole expansion gravity code would be fantastic, but since I'm at best a poor programmer, I didn't dare suggest that, since I didn't know whether it would be practical to implement. Actually, though, the way you described it (gravity depending on distance and angle just clarified the whole multipole thing for me. I think I was making it overly complicated in my head. Maybe I'll take a shot at re-writing my previous non-spherical gravity simulator...

[Greenleaf]

A spherical harmonic/multipole expansion gravity code would be fantastic, but since I'm at best a poor programmer, I didn't dare suggest that, since I didn't know whether it would be practical to implement. Actually, though, the way you described it (gravity depending on distance and angle just clarified the whole multipole thing for me. I think I was making it overly complicated in my head. Maybe I'll take a shot at re-writing my previous non-spherical gravity simulator...

It sounds like you know what you are talking about? I feel its an added bonus when we can make something people enjoy, even if they know about physics and astronomy in advance. On one hand, you can probably appreciate the complexity of things but on the other hand, you might also really notice where we come up short :-)


I see the multipole description of gravity much like for example a strain tensor, which essentially lets you pick any direction and by a simple matrix vector multiplication get the strain in that direction... except not as easily.


If you make some interesting progress with your simulator, do share :-)

[thalamus]

I certainly will! But it sounds to me like a job like this is going to require me to learn how to implement a proper Runge-Kutta--I'll admit I'm a poor enough programmer that I still haven't learned it <_<

[Greenleaf]

I certainly will! But it sounds to me like a job like this is going to require me to learn how to implement a proper Runge-Kutta--I'll admit I'm a poor enough programmer that I still haven't learned it <_<

Well, if you ask specifics, I am sure you can get answers. After all, everyone here is interested in this topic :-)