3 ms·

Hi, I work on N-body orbit determination software for asteroid detection and planetary defense (for example, https://github.com/moeyensj/thor https://github.com

Hi, I work on N-body orbit determination software for asteroid detection and planetary defense (for example, https://github.com/moeyensj/thor https://github.com/moeyensj/thor). Ask me anything. You're right that orbit determination is unstable over long runs, and very sensitive to initial conditions. But the good news is that the solar system is big. Most trajectories do not intersect with the earth's location. So perturbing an orbit that we think has a small chance of hitting us tends to be safe. Also, of course, after deflection, any object is going to be monitored really closely for a long time.

pageandrew 4y ago
Since DART crashed into Dimorphos, changing its orbit around Didymos rather than its orbit around the sun, isn’t there almost no cause for concern because Didymos is still on the same non-intersecting orbit around the sun? Is the concern that Dimorphos’s slightly different orbit could pull Didymos slightly off course in its orbit around the sun, thereby causing it to potentially intersect with Earth at some point in the future, where it wouldn’t have prior to DART. Can we quantify how much Dimorphos’s new orbit has affected Didymos’s orbit? Is that calculable?
nighthawk454 4y ago
Hey, this is a bit tangential, but is Barnes-Hut still used in modern n-body simulations? Some of the math has huge crossover with machine learning and dimension reduction, but it’s rarely traced back to the source in literature. Thanks! (I’ll also check out the GitHub repo you linked but I haven’t read it fully yet)
Hendrikto 4y ago
Maybe a bit off topic, but: * What language is that software written in? * What was the hardest problem you had to solve, while working on it?
- spenczar5 4y ago
  Mostly Python and Fortran. See for example https://github.com/oorb/oorb https://github.com/oorb/oorb. The hardest problems are always the social ones. How do you get uptake of a new method, how do you get funding, how do you politely tell a collaboration they are doing the wrong thing, etc. But if you mean pure technical stuff - the hardest problem I had to solve was rethinking some of the inner loops of the THOR algorithm. The problem was essentially to speed up a Hough transform in 6D space. Lots of time spent profiling CPU cache timings to get that fast.
TechBro8615 4y ago
Shouldn't we have some concern for downstream cascading effects of disrupting one orbit? I don't think anyone is concerned we'll knock this particular asteroid into an orbit that eventually collides with earth. But what about the other orbits that it interacts with, and every degree of interaction beyond that? I don't see why we should even be confident in estimating these effects past one or two degrees of separation. Most of the objects interacting with each other have had millennia to settle into a balanced system - is it really so alarmist to think that a small perturbance to one rock could lead to unrecognizably subtle but strong effects on other rocks? And if two rocks smash into each other, then we've got a bunch of new smaller rocks to worry about. Btw, there is an article on this topic that I posted to HN a while back but didn't get much attention - it's a good read though: https://nautil.us/should-we-really-be-messing-with-asteroid-orbits-244252/ https://nautil.us/should-we-really-be-messing-with-asteroid-...
- spenczar5 4y ago
  Asteroid collisions actually happen all the time. Those collisions are usually the way objects leave the asteroid belt to go hit a planet. The system overall is not as stable as you might think. Gravitational forces of asteroids on each other are negligible, other than the real biggies like Ceres, or binary systems right next to each other. That article is a bit odd. For example, the claim that we only do asteroid discovery and not follow-up for orbital changes is absurd on its face. If an orbit changes then detection systems will perceive it as a new asteroid, and it will stop appearing in predicted locations. The broader point - that this is a technology with perhaps too much power, like what if nations shoot asteroids at each other, is a bit more valid but still far-fetched.
- zamadatix 4y ago
  When looking at an individual asteroid meaningful orbital interactions with another asteroid are extremely rare but it has more to do with the inverse square law and the vastness of space than a perfect set of orbital parameters. When looking at the nearly million we know about rare becomes common but still hasn't become runaway. After all if things not being perfectly aligned led to a quick chain reaction then we'd expect all of the asteroids on a near collision course with Earth should cause chain reactions after passing and leaving with very random orbits. We don't observe that though, or a resulting chain reaction from those orbits being changed, or any chain reaction from asteroid orbits chain reacting from other interactions. We just observe the occasional random interaction and then a long period of nothing for the individual asteroid. While it's not a perfect map to the exact perturbation that'd happen in reality you can get an idea how rare asteroid-asteroid interactions are with an n-body simulator. The other concerns of the article have more to do with different kinds of deflection (planetary, mining, intentional). While they're interesting topics with varying levels of risk (perceived or real) they don't really interact with the avoidance case.
themitigating 4y ago
I feel like he didn't expect someone who has such specific domain knowledge to reply

pageandrew 4y ago

Since DART crashed into Dimorphos, changing its orbit around Didymos rather than its orbit around the sun, isn’t there almost no cause for concern because Didymos is still on the same non-intersecting orbit around the sun? Is the concern that Dimorphos’s slightly different orbit could pull Didymos slightly off course in its orbit around the sun, thereby causing it to potentially intersect with Earth at some point in the future, where it wouldn’t have prior to DART. Can we quantify how much Dimorphos’s new orbit has affected Didymos’s orbit? Is that calculable?

nighthawk454 4y ago

Hey, this is a bit tangential, but is Barnes-Hut still used in modern n-body simulations? Some of the math has huge crossover with machine learning and dimension reduction, but it’s rarely traced back to the source in literature. Thanks! (I’ll also check out the GitHub repo you linked but I haven’t read it fully yet)

Hendrikto 4y ago

Maybe a bit off topic, but: * What language is that software written in? * What was the hardest problem you had to solve, while working on it?

spenczar5 4y ago
Mostly Python and Fortran. See for example https://github.com/oorb/oorb https://github.com/oorb/oorb. The hardest problems are always the social ones. How do you get uptake of a new method, how do you get funding, how do you politely tell a collaboration they are doing the wrong thing, etc. But if you mean pure technical stuff - the hardest problem I had to solve was rethinking some of the inner loops of the THOR algorithm. The problem was essentially to speed up a Hough transform in 6D space. Lots of time spent profiling CPU cache timings to get that fast.

TechBro8615 4y ago

Shouldn't we have some concern for downstream cascading effects of disrupting one orbit? I don't think anyone is concerned we'll knock this particular asteroid into an orbit that eventually collides with earth. But what about the other orbits that it interacts with, and every degree of interaction beyond that? I don't see why we should even be confident in estimating these effects past one or two degrees of separation. Most of the objects interacting with each other have had millennia to settle into a balanced system - is it really so alarmist to think that a small perturbance to one rock could lead to unrecognizably subtle but strong effects on other rocks? And if two rocks smash into each other, then we've got a bunch of new smaller rocks to worry about. Btw, there is an article on this topic that I posted to HN a while back but didn't get much attention - it's a good read though: https://nautil.us/should-we-really-be-messing-with-asteroid-orbits-244252/ https://nautil.us/should-we-really-be-messing-with-asteroid-...

spenczar5 4y ago
Asteroid collisions actually happen all the time. Those collisions are usually the way objects leave the asteroid belt to go hit a planet. The system overall is not as stable as you might think. Gravitational forces of asteroids on each other are negligible, other than the real biggies like Ceres, or binary systems right next to each other. That article is a bit odd. For example, the claim that we only do asteroid discovery and not follow-up for orbital changes is absurd on its face. If an orbit changes then detection systems will perceive it as a new asteroid, and it will stop appearing in predicted locations. The broader point - that this is a technology with perhaps too much power, like what if nations shoot asteroids at each other, is a bit more valid but still far-fetched.
zamadatix 4y ago
When looking at an individual asteroid meaningful orbital interactions with another asteroid are extremely rare but it has more to do with the inverse square law and the vastness of space than a perfect set of orbital parameters. When looking at the nearly million we know about rare becomes common but still hasn't become runaway. After all if things not being perfectly aligned led to a quick chain reaction then we'd expect all of the asteroids on a near collision course with Earth should cause chain reactions after passing and leaving with very random orbits. We don't observe that though, or a resulting chain reaction from those orbits being changed, or any chain reaction from asteroid orbits chain reacting from other interactions. We just observe the occasional random interaction and then a long period of nothing for the individual asteroid. While it's not a perfect map to the exact perturbation that'd happen in reality you can get an idea how rare asteroid-asteroid interactions are with an n-body simulator. The other concerns of the article have more to do with different kinds of deflection (planetary, mining, intentional). While they're interesting topics with varying levels of risk (perceived or real) they don't really interact with the avoidance case.

themitigating 4y ago

I feel like he didn't expect someone who has such specific domain knowledge to reply