galactic collision

[Thundercracker]

ive just started off with some of the new galaxies, and i am getting unexpected results.

when galaxies collide, i expect them to fall through each other. instead, they make contact, shred each other up a bit, and then recoil, as if two solid objects bouncing off of each other.

using the default obit selection when placing a sample of random galaxies in a tight cluster, one would expect everything to eventually fall into the large elliptical galaxy in the center. instead, everything ends up on hyperbolic trajectories with nothing bound by gravity.

[Tseudonym]

I've noticed this too. Though a small galaxy and a considerably larger galaxy (such as the Milky Way and Large Magellanic Cloud) will merge as expected.

[24 is the highest number]

Without a highly detailed model (using something like smooth-particle hydrodynamics[en.wikipedia.org]), it'll be more likely that the galaxies will be flung apart than merge together realistically. This is because the mass is concentrated in the black holes rather than distributed throughout, so it's easy for one black hole to slingshot the other away.

The nebulae themselves are not attracting, so everything moves according to the black holes. Due to how many nebulae there are, it's not feasible to have them all attract and still have good performance, at least not how we currently simulate things. You can see this in any simulation where there are only even a few dozen attracting objects.

We'll continue to think about how to improve this, but it may just come down to requiring the SPH (smooth particle hydrodynamics) model.

[Thundercracker]

Originally posted by burt reynolds wrap:

Without a highly detailed model (using something like smooth-particle hydrodynamics[en.wikipedia.org]), it'll be more likely that the galaxies will be flung apart than merge together realistically. This is because the mass is concentrated in the black holes rather than distributed throughout, so it's easy for one black hole to slingshot the other away.

The nebulae themselves are not attracting, so everything moves according to the black holes. Due to how many nebulae there are, it's not feasible to have them all attract and still have good performance, at least not how we currently simulate things. You can see this in any simulation where there are only even a few dozen attracting objects.

We'll continue to think about how to improve this, but it may just come down to requiring the SPH (smooth particle hydrodynamics) model.

i understand this, but this is not following even that expected model.

first, galaxies seem to be flung away at above escape velocity, meaning that they will never merge, the trajectories are hyperbolic.

when two galaxies of near size "collide" the black holes are staying centered. they are not getting slingshot. they seem to be rebounding off of each other as of the nebulae are solid mass.

the black holes controlling things get to a minimum distance, right as the fringes interact, then just bounce off.

[Thundercracker]

a question:

can nebula just not be sucked up by the galactic black holes? increasing the simulation's gravity resulted not in feeding black holes, but all nebula particles being flung out at or above escape velocity.

also, pretty much any edit of a galactic black hole's properties will result in the galaxy self-shredding. i've been playing with spin, and any edit sreds the galaxy.

[Jar]

Those are all good questions, unfortunately ones I can't answer of the top of my head. I'll defer to our astrophysicist & developer who worked on galaxies and see if we can come up with some answers next week.

[Jar]

Okay, here are some explanations:

Originally posted by Dr Gori:

can nebula just not be sucked up by the galactic black holes?

Nebulae do not collide, which allows them to pass through each other as you'd expect. But that does have the side-effect that they will not collide with the black hole either.

Originally posted by Dr Gori:

also, pretty much any edit of a galactic black hole's properties will result in the galaxy self-shredding. i've been playing with spin, and any edit sreds the galaxy.

This makes sense when adjusting the mass (or mass via radius and density), as the orbits will all change. But the spin is a little more curious, we'll look into that one.

I'm a bit unsure of what you're describing with the "bouncing" effect. Do you have a sim you can share of this?

[Thundercracker]

Originally posted by Jar:

I'm a bit unsure of what you're describing with the "bouncing" effect. Do you have a sim you can share of this?

sure thing, i'll set one up in just a moment.

https://steamcommunity.com/sharedfiles/filedetails/?id=1789308700

i've run this one a few times now to confirm. in this sim, pay attention to stasc.

the effect i've been seeing has probably not been a rebound, but instead seems to be something else. galaxies do not seem to be strongly bound to each other, and smaller, closer galaxies seem to have a larger effect than a much more massive but slightly farther away one. this is combining with orbital velocities to "pull" one galaxy falling into another just enough to prevent a merger and eventually send everything flying away from everything else.

were these galaxies representing galaxy clusters, it would be a good demonstration of the large scale structure of the universe. however, as a small cluster of galaxies, it is not working at all.

this simulation also highlights the slingshot effect, and general shredding.

[Jar]

Okay, yes, that does look like the merging we'd expect to see. This is pretty much rewording what has already been said, but...: The only things that have full gravity simulation for galaxies are the black holes, so the movement should be simulated in exactly the same way as though you placed a cluster of them, though their masses for the n-body calculations are taken from the whole mass of the galaxy.

From there, the nebulae are attracted to those points of mass, and their placement and structure do affect how they appear, so that is something to note as well. The calculations that keep the structure intact when there isn't another attracting body nearby go out the window with a merger.

So we'll continue to think about how we can improve this. I don't think I'm seeing anything beyond what we'd expect with the current state of mergers, but if I'm misunderstanding anything, please let me know.

[Thundercracker]

Originally posted by Jar:

So we'll continue to think about how we can improve this. I don't think I'm seeing anything beyond what we'd expect with the current state of mergers, but if I'm misunderstanding anything, please let me know.

i guess the root of my issue stems from two fundamentals:

1. nebula do not interact with black holes, so it is the fate of every nebula particle to eventually be flung into a hyperbolic trajectory.

2. not once have i seen supermassive black holes merge, instead they fling each other out on a hyperbolic course. often, one or both SMBH will be flung out at such speed ( sometimes, more than 1000 km/s more than the speed of light!) that its galactic nebula become unbound. curiously, these will just fly off hyperbolically rather than fall into orbit around any remaining nearby galaxy.

because of this, i know the outcome of every galactic simulation. every nebula particle and SMBH flying away from each other at high speed, never to interact with anything again.

dont get me wrong, the galaxy visuals are great, and everything simulates as expected during an interaction up to the point where the galactic nuclei start to interact.

[Chocice75]

Looking back, the older version of the Galaxy Collision simulation actually Andromeda and the Milky Way merging instead of having their black holes getting flung into outer space.

This is still the case if you load an older version of the simulation in Update 23, so there must be something different between the new and old galaxies which causes the galaxies to tear each other apart.

[Jar]

Originally posted by Dr Gori:

1. nebula do not interact with black holes, so it is the fate of every nebula particle to eventually be flung into a hyperbolic trajectory.

2. not once have i seen supermassive black holes merge, instead they fling each other out on a hyperbolic course.

We have an idea for nebula & black hole interaction that should help with this. And we are now looking into a specific bug with this black hole behavior. Hopefully we'll be able to improve this!

Originally posted by ChocIce75:

This is still the case if you load an older version of the simulation in Update 23, so there must be something different between the new and old galaxies which causes the galaxies to tear each other apart.

Hmm, I'm not sure what you're referring to here. We removed the galaxy sims that were included pre-Update 23, so that sim shouldn't be accessible anymore. And while you can load a previously saved simulation, it'll replace the old galaxies with the new ones, and this should make the behavior the same as any other sim with the new galaxies. Are you seeing something different, and if so, how?

[Thundercracker]

Originally posted by Jar:

We have an idea for nebula & black hole interaction that should help with this. And we are now looking into a specific bug with this black hole behavior. Hopefully we'll be able to improve this!

i'm looking forward to these fixes!

on a related side note, along with realistic constraints on galactic properties, can you guys check under the hood at the velocity component? when galaxies start falling into each other, they're generally moving at something like 1% lightspeed for me. would this be related to the bug you mentioned, or the inflated physical properties, or could it be something else?

[Chocice75]

Originally posted by Jar:

Hmm, I'm not sure what you're referring to here. We removed the galaxy sims that were included pre-Update 23, so that sim shouldn't be accessible anymore. And while you can load a previously saved simulation, it'll replace the old galaxies with the new ones, and this should make the behavior the same as any other sim with the new galaxies. Are you seeing something different, and if so, how?

Many different people uploaded them to the Workshop. Although they look more like the new galaxies, the black holes don't always fling each other away.

Normally, the black holes are flung away at a slow speed, taking a few nebula with them.
Once, they actually started orbiting each other and the galaxies merged.

This is the simulation I used:
https://steamcommunity.com/sharedfiles/filedetails/?id=1694196101

Since the galaxies in that sim weren't classed as actual galaxies and couldn't be modified at all, their behaviour seems to be the same as in Update 22.2.