Required Compuations

1. Oct 29, 2009

frankinstein

Has anyone calculated what kind of computational horsepower is required to simulate a universe particle by particle?

Frank

2. Oct 29, 2009

Max™

I think the last time that was tried, well... you remember that thing we call the Big Bang?

:P

http://www.mpa-garching.mpg.de/galform/millennium/

That is the best I know of currently.

Over 10^10 particles as it says there.

I know my computer goes "you got life ****ed up if you think I'm playing that crap" when I try to watch it.

Estimates put the number of particles in the Universe at 10^80 I think?

3. Oct 29, 2009

Fra

4. Oct 29, 2009

chroot

Staff Emeritus
Simulating a single particle requires machinery composed of many particles. It is a hopeless endeavor to simulate the universe, because you would need a computer bigger than the universe to do it.

- Warren

5. Oct 29, 2009

mikeph

even if you had complete information, no computational technique comes without error, and in the chaotic dynamics of the universe your approximation would inevitably diverge... that's my thought, so "impossible", for reasons separate to the above mentioned (which are all pretty good in themselves!)

6. Oct 29, 2009

frankinstein

I agree...

However my question is whether anyone has thought of how much computational effort goes towards simulating just one particle? I'm running a simulation of a standing wave computer and I need some metrics to extrapolate the computational IQ of such a computer to create a universe. Right now the system is simply 3 dimensional with a 3 x 3 X 3 matrix and converges to some interesting patterns that are actually ending up as super imposed logical operations.

The IQ of the system is not just how many bits are processed given the degrees of connection between standing waves but how quickly bits can propagate through the system. The parallelism of this approach improves as I add more dimensions or sides to the standing waves, so as more dimensions are added to the standing waves more information can be processed and communicated per volume of standing waves.

The interesting thing about the whole idea is since the system is chaotic it programs itself, both data and function exist simultaneously within the matrix of standing waves!

If I could get a metric for the computational demand for a particle I could calculate a minimum number of dimensions the standing waves' need.

My gut instinct tells me the number may be in the thousands! Can you imagine a super hyper cube with 5000 dimensions? That's 10,000 sides. Think of it this way a computational unit with 10,000 connections! Biological neurons have about the same connectivity per neuron...

Frank

Last edited: Oct 29, 2009
7. Oct 29, 2009

frankinstein

Not true, a system could self organize and correct its error. Kind of like nueral networks which back propagate the error to converge on a solution to a data set.

Frank

8. Oct 29, 2009

Haelfix

Generically you'd need a quantum computer with at least the energy of the observable universe, and with at least the entropy therein (which is something like ~10^120 bits or qbits).

The computer would have to be very nearly the volume of the observable universe (otherwise you end up creating a blackhole) and one wonders about other thermodynamic issues (heat loss/gain), as well as error correction and so forth.

What you end up with is something greater in size and magnitude than what it actually simulates.

9. Oct 30, 2009

frankinstein

I do generally agree with you but the computer I'm theorizing is composed of standing waves in a matrix of super hyper cubes that interact with one another and vector a state. The idea is the premise for a realizable virtual universe. The idea that our reality could be created by a computer is not a new one but no one has laid down a workable computational system that could come about without intelligent design.

The hint that our reality could be a virtual reality lies in the mysticism of force carriers and fields. Things in our universe move magically, why force can make things move is simply an assumption that it can there is no real mechanism for it and physicists claim there is no need to explain it. Force as applied to particles is really information, e.g. an electron absorbs a charge photon and then gains some kind of momentum from it. The electron seems capable of discerning if the photon is a charge or an electro magnetic photon and changes its behavior accordingly. That kind of decision making indicates intelligence.

With a virtual universe theory based on a standing wave computational matrix there are far fewer assumptions, standing waves either re-enforce one another or cancel each other out. Standing waves are fixed in position so there is no need for some kind of glue to hold them together. The stuff of reality is far simpler with this approach but proving the idea I believe is almost impossible. So why do I bother with it...Because I really like the idea that it is actually quite possible...

Frank

10. Oct 30, 2009

magpies

A human with a good imagination could come close to doing it in there head with some practice I think. I know I have created a universe with my mind once and my understanding of it in my mind was that it expanded by doubling its size about once every second for about 2 hours. I could not figure out how to stop it for so long... So I might advise some caution if you do decide to try to build a universe in your mind.

11. Oct 30, 2009

mikeph

I'm going to take a very annoying standpoint, but any computer with a finite time step would not be able to model the universe which works in continuous time... if I give you the state of a system at t=0, and you find it at t = t1, what was it at t = t1/2? what about at t1/n?

Infinite computations for a single non-analytic system with generalised initial conditions (eg. three body)

12. Oct 30, 2009

frankinstein

With a standing wave computer there is no synchronous timing, in fact all the standing waves are asynchronous bits that influence each other chaotically. But even with your argument in what we call a real universe particles only affine to a state when measured. So at t = t1/n is irrelevant its whether or not particles have interacted with one another and t is subject to relativistic issues. So an emulation that can model asynchronous environments, which would be something like event modeling done with object oriented programs today, transactions or interactions are all that are important not the timing of some global clock.

Frank

13. Oct 30, 2009

chroot

Staff Emeritus
The term "standing wave computer" is meaningless, and has no precedent. You need to explain what you're talking about. If this is some kind of personal invention or theory, please note that it is not welcome on PF.

- Warren

14. Oct 30, 2009

frankinstein

I'm asking a question about the computational demands to emulate a particle. As to how I plan on using that kind of information I thought it useful for those who could answer the question...

Frank

15. Oct 30, 2009

chroot

Staff Emeritus
The answer is that -- with all technology currently known, including quantum computation -- it takes many millions or billions of particles, formed into a computing machine, to simulate one.

- Warren

16. Oct 30, 2009

frankinstein

Well thank you for that information...However I'm not asking how many particles it takes to build a machine that can emulate one particle. I am asking how many bits of information are required to symbolize a particle and its behavior so such a machine of billions of particles can act on those bits?

Frank

17. Oct 30, 2009

chroot

Staff Emeritus
So, you're asking how many parameters you need in your model to simulate a particle? That depends on the kind of particle (not all particles "feel" all of the four fundamental forces).

If all you're tying to simulate are the large-scale structures (like the galaxy formation results shown above), then you just need to simulate gravity. Mass, position, and velocity of each particle are all that you need.

- Warren

18. Oct 30, 2009

frankinstein

Warren, you just gave me an idea of how to get want I need, thanks.

Frank

19. Oct 30, 2009

Q_Goest

20. Oct 30, 2009

Haelfix

If you want a perfect simulation for even pure gravity you need to specify not just the positions, velocities and masses of the particles, but also their quantum numbers as well (charge, color, flavor etc) b/c all field interactions contribute to the stress energy tensor of General relativity.

You cannot *compress* the information or the entropy of what you are simulating b/c in some sense it is already as compressed as it can get and likewise you cannot escape spending energy to *faithfully* reproduce the dynamics.

Now if you want something approximate, then all bets are off of course.