Simulating Extreme Spacetimes | Hacker News Reader

Simulating Extreme Spacetimes

black-holes.org

91 pointsMichaelAO17 comments

Somewhat off-topic, but regarding visualization, we're told that Andromeda is on a collision course with the Milk Way, but also that it is very unlikely that e.g. the Sun will actually collide with any other stars -- galaxies are approximately empty over all. I have always want to know what it would look like from Earth to see the stars of the two galaxies 'sliding across' the celestial sphere during the merge. Has anyone seen something like that? I search online for it every few years.

Well I have just the thing for you:

http://www.nasa.gov/mission_pages/hubble/science/milky-way-c...

Ever since I saw this I wish I can live and see how it happens.

Even at closest approach, there won't be any appreciable change in the night sky over a human lifetime. Note that in the gallery at the bottom of http://www.nasa.gov/mission_pages/hubble/science/milky-way-c... the shortest time between frames is still 50 million years.

From the original article:

" it will take 4 billion years before the strike. ... it will take an additional two billion years after the encounter for the interacting galaxies to completely merge"

Also, it's important to keep in mind that galaxies have a pretty low density. If you shrank the sun down to the size of a ping pong ball, earth would be a poppy seed about 10 feet away. Proxima centauri, the nearest star, would be 500 miles away.

Well, you will likely see a lot and lot of stars in the sky and that's probably all. The predictable constellations in different months of the year in the night sky will be redefined. But this is going to be a continuous evolving night sky which will take shape over many many generations of human life span.

https://www.youtube.com/watch?v=qnYCpQyRp-4

I understand what you're going for, but by the time it gets exciting, Earth will either be uninhabitable or eaten by the Sun.

Heh, yes, I hedged that with what it "would" look like (from a position where the Earth once was). Four to five billion years is a long time.

Something the general reader may not know: As late as ~2004-7, calculations of this type were impossible. It was a major achievement when numerical relativists were able to achieve a single orbit of a black-hole black-hole binary system.

Figuring out how to make it work was the culmination of decades of attempts at the problem. Improved computational power didn't hurt, but the major improvement was finding the right metrics in which to do the simulations.

Totally unrelated to anything, but I love the term "numerical relativists". I know it's actually a branch of computational physics, but it calls to mind postmodernists invading mathematics departments and claiming that theorems are socially constructed, and that maybe 2 > 3 under a Marxist-Leninist interpretation.

Can you say more about the 'right metrics'? Is this some sort of transformation akin to working in spherical or cylindrical coordinate systems?

From what I remember about numerically solving differential equations, your approximations will completely diverge from the actual solution near singular points in the equation, so you have to transform the problem to come up with a singular solution. I imagine the interaction of two singularities would result in some interesting singular points. Is that what you are alluding to when you say "finding the right metrics" or am I completely off base?

In either case, could you link to one of those 2004-7 papers? It sounds like a fascinating read.

A tricky part was making sure that numerical instabilities didn't propagate outward away from the event horizon. I believe there are now several solutions to the problem, but the one I find most intuitive is simply extending spacetime ad infinitum inside the black hole in order to allow gravitational waves to propagate inward forever.

More here: https://en.wikipedia.org/wiki/Numerical_relativity

Super cool research going on here. I came across their website after seeing, "What the first LIGO detection would look like up close" https://www.youtube.com/watch?v=Zt8Z_uzG71o

[edit] this might be a better link for visuals of the simulations: https://www.youtube.com/user/SXSCollaboration/videos

This almost feels pornographic, it's so good. In the particular the extreme counter-rotating example is so interesting to consider, but really exciting to see simulated. I just imagine the ergoregions of each body going absolutely nuts in those last few moments of merger.

It would be interesting to do the ray tracing and see how the mergers would actually appear to an observer.

You could probably do it with a handful of test photons, without breaking the bank.

There's a visible-wavelength simulation of a gravitational singularity in free space running through Apache on an old Sun machine at Dal:

https://web.cs.dal.ca/~hannon/

;)

Reply on news.ycombinator.com