What can you do with galaxy sized computers?

[WinkAllKerb'']

well , let's suppose i wrote a movie script for theater, or a fictionnal science book:

the plot i would use eventually:

- the universe is like one of the first molecule in the earth primitive soup, it still float somewhere in the soup waiting to encounter an other molecule
- whµman miniaturize themselve to the point they are able to live on a single electron with a nucleus for sun
- they also for some weird reason found a tecnology that allow them to rescale there own lifespan so they re not "seasick" while living on an electron ; )

Edited April 22, 2016 by WinkAllKerb''

[Turf]

44 minutes ago, swjr-swis said:

We would run KSP with ALL the mods, of course.

It would be capable of total realism and full physics even at maximum time warp. But since there would still be argument about realism vs play levels, and the exact combination of mods at any one time, it'd be multi-instance and customizable on the fly.

And then we'd mind-transfer into it. Because we all secretly want to be Jeb (or Val).

 

Imagine clicking the left right button located in top-right screen to switch mods.... Pointless.... Not to mention conflicting mods.... Deep Space Kraken....

How about me?

I would store Graham's Number on there.

[p1t1o]

7 minutes ago, HebaruSan said:

That rules out options #2 and #3, since it wouldn't be the whole universe, and/or it wouldn't be tracking the motion of each single atom. My argument above addresses whether #1 is possible (my contention being that a universe capable of hosting a galaxy-sized computer requires no less power to simulate atom-by-atom than one that already has one).

I think it can still be done, you just allocate 10 seconds of processing time to every second of simulated universe time, or whatever ratio works.

Edited April 22, 2016 by p1t1o

[RainDreamer]

1 minute ago, p1t1o said:

I think it can still be done, you just allocate 10 seconds of processing time to every second of simulated universe time, or whatever ratio works.

Would that mean for a person from outside observing the process they would see the simulated universe being extremely laggy, but for beings inside the simulated universe time flow normally for them? That is...kind of mind bending.

[HebaruSan]

2 minutes ago, p1t1o said:

I think it can still be done, you just allocate 10 seconds of processing time to every second of simulated universe time, or whatever ratio works.

Note that the argument I made doesn't depend on time; regardless of how much you slow down the simulation, you still have to store all the data somewhere, and there isn't enough room to do it.

[sevenperforce]

8 minutes ago, HebaruSan said:

That rules out options #2 and #3, since it wouldn't be the whole universe, and/or it wouldn't be tracking the motion of each single atom. My argument above addresses whether #1 is possible (my contention being that a universe capable of hosting a galaxy-sized computer requires no less power to simulate atom-by-atom than one that already has one).

Atom-by-atom is easy.

But the universe is splitting atoms all the time. So it really needs to go quanta-by-quanta, which is not so easy.

XKCD #505 is apt...

[WinkAllKerb'']

wich could bring us back to chronos and kronos as metaphor; ) wich is pretty fun when you think about it ; )

both all other god father, both a titan. but it use the same phoenet(h)ics with an ideopictogramm disambiguation ; )

Edited April 22, 2016 by WinkAllKerb''

[sevenperforce]

Also relevant...

[todofwar]

I think whoever said cat videos and porn is probably right. Just some of the problems we could solve: Protein folding, perfect drug interaction simulations giving us all the cures to all the diseases, perfect simulations of the human body so that we know all the side effects and how to mitigate them. Honestly, I think biology has more use of supercomputers than physicists do, biological systems are way more complicated. Still, this wouldn't actually instantly advance our science capabilities. We are limited by our equations and ability to understand said equations. So, in the end we would progress through theoretical physics and computational biology much faster, but it would still take time. 

[p1t1o]

11 minutes ago, HebaruSan said:

Note that the argument I made doesn't depend on time; regardless of how much you slow down the simulation, you still have to store all the data somewhere, and there isn't enough room to do it.

Didn't think of that....

Lossless compression? Thats a thing isn't it?

[HebaruSan]

3 minutes ago, p1t1o said:

Lossless compression? Thats a thing isn't it?

Hmm, interesting. Not all data is compressible (e.g., zipping a JPEG or PNG will generally make it bigger, not smaller), and I'm not sure how well this would work for the quanta in the universe. If we have 10 quadrillion atoms with the same velocity, then that data could compress really well, but if they're distributed evenly across a spectrum, it's less helpful (or potentially makes the problem worse).

[p1t1o]

2 minutes ago, HebaruSan said:

Hmm, interesting. Not all data is compressible (e.g., zipping a JPEG or PNG will generally make it bigger, not smaller), and I'm not sure how well this would work for the quanta in the universe. If we have 10 quadrillion atoms with the same velocity, then that data could compress really well, but if they're distributed evenly across a spectrum, it's less helpful (or potentially makes the problem worse).

It all depends on how strictly you want to stick to the "every single particle with exact properties" rule, also I think the uncertainty principle might give some major headaches.

On typing "headache" I realised I had one, thanks universe.

[Shpaget]

Such a computer would turn into a black hole long before it grew to the size of a galaxy. As such it would not be feasible to extract any meaningfull output from it.

It would be the worst supercomputer ever.

[p1t1o]

3 minutes ago, Shpaget said:

Such a computer would turn into a black hole long before it grew to the size of a galaxy. As such it would not be feasible to extract any meaningfull output from it.

It would be the worst supercomputer ever.

When the idea was first brought up by the OP, I did not assume it was a solid monolithic computer enveloping the volume of a galaxy, but a galaxy-worth of matter converted to something close to "computronium" but still dispersed in the same way as a galaxy.

Interesting question: how many conventional galaxy's worth of matter would it take to fill a galaxy-sized void full to the brim with something approaching the density of "normal" matter? If you had to make a solid, monolithic galaxy sized computer, how many other galaxies would you have to consume?

 

[Shpaget]

10^29

[RainDreamer]

 

10 minutes ago, Shpaget said:

Such a computer would turn into a black hole long before it grew to the size of a galaxy. As such it would not be feasible to extract any meaningfull output from it.

It would be the worst supercomputer ever.

 

3 minutes ago, p1t1o said:

When the idea was first brought up by the OP, I did not assume it was a solid monolithic computer enveloping the volume of a galaxy, but a galaxy-worth of matter converted to something close to "computronium" but still dispersed in the same way as a galaxy.

Yep, I was thinking of computers similar to what I described in the OP, which are actually clouds of machines(probably not nano) in the shape of a galaxy, connecting with each other by perhaps light signals. They can re-arrange themselves for optimal configurations as needed and all that.

[FungusForge]

Well, we can kinda simulate galaxies crashing on our home potatoes, so even with potato-tech I'm sure we could do amazing things. Most likely simulate large-scale space events, like a physically accurate supernova or something. Or a planet being torn apart because we plonked it in the middle or a binary star system for giggles.

 

Really though, I'd use it to play Supreme Commander: FA with 7 other guys and the 1000 unit cap. I'd finally be able to do that in real time.

[sevenperforce]

33 minutes ago, HebaruSan said:

Hmm, interesting. Not all data is compressible (e.g., zipping a JPEG or PNG will generally make it bigger, not smaller), and I'm not sure how well this would work for the quanta in the universe. If we have 10 quadrillion atoms with the same velocity, then that data could compress really well, but if they're distributed evenly across a spectrum, it's less helpful (or potentially makes the problem worse).

I just realized something.

Chemistry is a compression algorithm for a subset of physics. There are certain physical particle interactions which happen the same way every time and thus can be represented in abbreviated form.

Biology is a compression algorithm for a subset of chemistry. Psychology is a compression algorithm for a subset of biology. Sociology is a compression algorithm for a subset of psychology.

A neat way of thinking about it, anyway. The supercomputer could "compress" by simulating chemistry rather than physics when particles were in the correct energy ranges, and so forth.

[Broco]

Pff easy answer.

Run a 1000 part vessel with 120 fps.

[Atlas2342]

Ask it to solve a paradox...even though paradoxes are paradoxes... 

Edited April 22, 2016 by Atlas2342

[p1t1o]

9 minutes ago, sevenperforce said:

I just realized something.

Chemistry is a compression algorithm for a subset of physics. There are certain physical particle interactions which happen the same way every time and thus can be represented in abbreviated form.

Biology is a compression algorithm for a subset of chemistry. Psychology is a compression algorithm for a subset of biology. Sociology is a compression algorithm for a subset of psychology.

A neat way of thinking about it, anyway. The supercomputer could "compress" by simulating chemistry rather than physics when particles were in the correct energy ranges, and so forth.

So basically what you are saying is that I could use it to determine sets of sociological behaviour?

I...I could ask it to teach me to be cool??

[HebaruSan]

2 minutes ago, p1t1o said:

So basically what you are saying is that I could use it to determine sets of sociological behaviour?

Well, only insofar as sociology itself is accurate and has predictive power. (Which is to say, not very much.)

2 minutes ago, p1t1o said:

I...I could ask it to teach me to be cool??

You could certainly ask. 

[p1t1o]

2 minutes ago, HebaruSan said:

You could certainly ask. 

If you really wanted to burn me you coulda said: "Make you cool? Come on, at least suggest something plausible, its only the size of ONE galaxy..."

[kerbiloid]

2 hours ago, HebaruSan said:

Other things you could do with a galaxy sized computer: Run out of power, overheat, and collapse into a black hole.

Intel galaxies usually don't overheat. While quasars were invented by... Another company.

 

Back to the topic: galaxy is just a chip. The Universe is a computer. It keeps all possible information as a superposition of her quantum states.

 

[HebaruSan]

8 minutes ago, p1t1o said:

If you really wanted to burn me you coulda said: "Make you cool? Come on, at least suggest something plausible, its only the size of ONE galaxy..."

Sorry, you were in the wrong place at the wrong time during a sick burn on sociology.