Galaxy Collisions: Softening Parameter and Dark Matter Halo

[Eni]

Lots of stars/star clusters are being jettisoned during a galaxy collision.  The astronomers here at my university study galaxy collisions are curious to how the physics of the collisions were set up.  One said the jettison issue was common (I think?) and could be lessened with a softening parameter -- to which I couldn't find controls for.  (All the controls were for planetary bodies.)

He also wanted me to ask about dark matter halos around the galaxies.

I know the team *just* got someone with an astronomy background on board and I realize you may or may not have an answer for me.

[FiahOwl]

Right now Dark Matter is not simulated, sadly.

Also, welcome to the community . I'm Fiah and you probably will see me around often.

[vh]

Hello, the problem with Universe Sandbox and gravity collisions is that currently, every object is simulated as a single point in space with a mass. The 'stars' you see that make up the galaxy are only particles with no mass.  Some people have made galaxies with real stars instead of particles, but they run very slowly.

[Dan Dixon]

Great questions...

There is currently no softening parameter or dark matter halos in the current version. Yes, I think softening will fix the jettison issue as well.

Both of these features are planned for the next major update (v3) due out later this year (and a free update for everyone who has already purchased).

[vh]

I'm not familiar with the 'jettison issue' I googled it and got nothing. Is it some astronomical term/issue/thing, or is it a problem with just universe sandbox?

Something like 0:15 of this?
Milkomedia

[Dan Dixon]

'Jettison issue' isn't a technical term, so that's why you found nothing.

And yes that video demonstrates it... (and even the stray stars that shoot out before the main burst will not happen once some smoothing is added)

[Eni]

Thanks.  I'll pass the information about the current physics and general setup to the astronomers.  (Bossman is kind of bummed.  )

Any idea of when v3 will be released?  And, are you taking volunteers for beta testers?

We were hoping to use Universe Sandbox to produce clips for a full-dome movie about intergalactic stars, which one of the astronomers here study along with galaxy collisions.  Currently we have a bunch of different clips and Bossman and I were hoping to make them look more uniform.

... did I mention that I work in a planetarium?

[FiahOwl]

... did I mention that I work in a planetarium?

Pure awesomeness.

[smjjames]

... did I mention that I work in a planetarium?

Pure awesomeness.

Thats pretty cool

As for when v3 will be released, Dan doesn't know yet other than sometime this year. I'm hoping for a summer release myself.

[Seventh cavalry]

Hello, the problem with Universe Sandbox and gravity collisions is that currently, every object is simulated as a single point in space with a mass. The 'stars' you see that make up the galaxy are only particles with no mass.  Some people have made galaxies with real stars instead of particles, but they run very slowly.

Can i see that simulator ? And oh thats why so many starts get ejected when the explosion happens. If the stars' gravity in the galaxy is counted, i think would be different ? maybe some stars will go into black hole ? and most of the time planets get ejected by black hole instead of going inside is this natural ?

[Seventh cavalry]

Oh and i realized something else, i watched the milky way and andromeda collision again, the normal speed is much different with the 10 times normal speed ? what the speed define ? like speed of everything the planet move faster or the time moves faster ? if time moves faster then shall be no difference ? :/

[vh]

When you speed up the simulation too fast, something called time step messes with the orbits of the dust

Time-step is how often Universe Sandbox will calculate the positions of the bodies. For example, if you set the time step of a simulation of our solar system to 2 minutes, universe sandbox moves each of the objects 2 minutes forwards in time and position every calculation, or "step". Setting a simulations time step too high will cause inaccuracies in the simulation. This is because some of the objects might move too far forwards in time and position. If you run the "Shoemaker-Levy 9 crashes into Jupiter in 1994" simulation with a time step too high, the comets will fly through the comet. The program calculates the comet before and after the collision, but not in between.

[Seventh cavalry]

Does the new update give the stars in the galaxies a gravity ?

[vh]

I think it will not. Giving stars in a galaxy mass would cause massive (pun not intended) lag. Say the program must simulate 10,000 stars, that would require the calculations of 10,000 objects to each other which is about 99 million
You can make a 'galaxy' on your own though with universe sandbox with stars that have mass.

[Dan Dixon]

For 10,000 objects, it's not 10000!, it's 10000x9999. (Which is still quite a bit)

For everyone that's confused, 10! means 10 factorial = 10x9x8x7x6x5x4x3x2x1 = 3628800, but with 10 objects in Universe Sandbox the number of acceleration calculation is only 10x9 = 90. But each calculation does require a few steps.
http://en.wikipedia.org/wiki/Factorial

Eventually I suspect we'll have two modes, galaxies with stars without mass, and galaxies with stars with mass. Although no promises regarding the 2012 release.

[vh]

What makes large simulations with multiple bodies lag? Is it the rendering or the calculations between all the objects?

[Dan Dixon]

The rendering is relatively trivial. It's the calculations.

For a direct sum algorithm (what Universe Sandbox currently uses) it's n^2 (where n equals the number of bodies with mass).

So 100 bodies with mass = 100^2 = 100x100 = 10000
(it's actually 100x99 since you don't calculate gravity on one's self)

10 bodies = 10^2 calculations= 100
100 bodies = 100^2 calculations= 10,000
1000 bodies = 1000^2 calculations= 1,000,000

See why it gets slow as you add more bodies?

Dust particles, d, work into the calculation like this:

n^2 + n*d

Order of math operations means you do the exponents first, then the multiplication, then the addition:
http://en.wikipedia.org/wiki/Order_of_operations

So 10 bodies, 1000 particles = 10^2 + 10*1000 = 10,100 calculations

In Universe Sandbox 3 we're adding Barnes-Hut. Which will help reduce the number of calculation required each step.
http://en.wikipedia.org/wiki/Barnes%E2%80%93Hut_simulation
http://iss.ices.utexas.edu/?p=projects/galois/benchmarks/barnes_hut

[vh]

cool

http://en.wikipedia.org/wiki/File:Galaxy_collision.ogv

Barnes-Hut works by grouping far away particles and making them act as a single point of mass to reduce calculations from what i can understand of the wikipedia page.

So will be effective enough to make galaxy collisions with real stars and no dust?

[Dan Dixon]

So will be effective enough to make galaxy collisions with real stars and no dust?

It will help. It depends on what frame rate you're looking for.