Simulation of the universe

Hi PF

I would like to ask, if it would be possible to simulate the whole universe from the beginning, with the following ingredients:

  • A full theory of quantum gravity (LQG, CDT or whatever theory proves to be correct) coupled to matter- and other force-field (for example LQG coupled to a GUT)
  • A powerful enough supercomputer (like Tianhe-1A, the "Jaguar" computer or perhaps even more powerful)
  • The initial conditions of the universe

If this can be done, we would actually have created a virtual universe within the computer. And further this could mean, that it is possible to create a universe from a computer simulation, meaning that we could live in a virtual reality?

Is this all bollocks or could it actually be reality? I would like some opinions :)
 

marcus

Science Advisor
Gold Member
Dearly Missed
24,713
783
Hi Schreiber,
you must have read some of Ambjorn, Loll et al's writings and so you know that CDT people simulate the life history of small universes.

They have to have a place in memory for every 4D simplex, because the Monte Carlo shuffling is done on the whole 4D history.

Because the computer memory is finite, the lifetime of the universe must be finite. It has to arise, exist (for what is necessarily a very brief time), and then collapse.

If I remember, their largest simulations involve some 300,000 four-D simplices. You can imagine scaling up to your Jaguar or whatever Supercomputer. Even if you used a computer that was a trillion (1012) times better than theirs, it would only be handling universes which are 1000 times bigger and last 1000 times longer.

But their simulations are of universes that are only around 20-30 planck length units wide!
Could be less, I don't recall exactly.
They don't have control of the size, except by adding more simplexes to the sim. They have a statistical method to estimate from the model's behavior what the size in planck terms the simplex represents---it cannot be arbitrarily specified.

So you can see how restricted this method is, scale-wise. It does not seem possible that any computer one could imagine building would be able to simulate a fine-grain universe, even one the size of an electric toaster or a dog or cat.

IMHO We have to have modest expectations and think of very coarse grained simulations in which only very crude structure can exist. As I say, this is only my really humble opinion. I think even very coarse grained simulations at the galaxy cluster level could be very interesting
 
Last edited:
Yes I have actually come with the question from reading CDT by Ambjorn, Loll et al.

You are definitely on to something with the coarse-grained simulation :) Less computer power needed and therefore larger universes could be simulated. So if you could couple CDT to a GUT (or as in the paper http://arxiv.org/abs/0802.0719, where, as I understand, CDT is coupled to String Field Theory), then you should, in theory, be able to simulate galaxy formation, right? If you do it as a course-grained simulation of course.

Also Marcus, how are other QG-approaches doing with respect to simulation? Is CDT the only approach which is capable of being simulated on a computer as you mentioned above?

I really like the idea of being able to simulate the universe on a computer, because it opens up a lot of philosophical questions aswell, like are we living in "The Matrix", and if so, what implications would that have to the way we think of life, the universe and everything? :)
 
Last edited:
Anyone? :)
 

marcus

Science Advisor
Gold Member
Dearly Missed
24,713
783
Hi Schreiber, I apologize for being unresponsive. Got distracted by other stuff and the thread somehow slipped my mind. I don't know much about the simulations in other types of QG. There has been work in Causal Sets, and also in LQG, but not to the extent seen in CDT and I haven't followed it.

There is a key person---the name would help you dig up the papers. I will try to give you a lead. Also hopefully some other more knowledgeable people will respond to this question.

Yes, David Rideout. He is at Perimeter or was when I looked
http://arxiv.org/find/grp_physics/1/au:+Rideout_D/0/1/0/all/0/1

Notice that he works in Causal Sets and also LQG (a little, lately).
Some of his work is analytical (solving equations) but some involves massive computation using a parallel process system called "Cactus".

In other words applying supercomputers/parallel processing to work numerically in Causets and LQG instead of only working analytically.
I think people have to get good at working numerically in QG before they can address the whole job of simulating even a small universe.

Another person is Joe Henson. He may have become adept at the same sort of stuff that Rideout was doing, with "Cactus" if that is the right name for the system.

Here's a possible lead! LYDIA PHILPOTT, young, just got her PhD using Rideout's Cactus software.
http://arxiv.org/find/grp_physics/1/au:+Philpott_L/0/1/0/all/0/1
Causal Set Phenomenology--- http://arxiv.org/abs/1009.1593

http://arxiv.org/abs/0911.5595
Particle simulations in causal set theory
Lydia Philpott
(Submitted on 30 Nov 2009)
Models of particle propagation in causal set theory are investigated through simulations. For the swerves model the simulations are shown to agree with the expected continuum diffusion behaviour. Given the limitations on the simulated causal set size, the agreement is far better than anticipated.
==================
 
Last edited:
Hi again Marcus, and thanks for the reply :) No problem that my thread was unnoticed, seeing that a lot of greater things are going on ATM, for example the QG 11 and Strings 11 :) Lots of stuff to look into.

Thank you so much for the references to these people. I must say it looks really interesting to see that CDT is not the only field trying to do these simulations :) As for my self, I also think that it is an interesting philosophical question about living in a computer simulation and so fourth. But again thanks Marcus, I will be looking into this stuff, and give you a heads up if I get stuck or anything :)
 
This closely resembles a very curious thought experiment me and my room mate constructed some years ago. The conclusion we reached was that to fully simulate the entire universe, one needs a computer with the capacity to store an arbitrarily large amount of data in an arbitrarily small amount of physical space.
The reasoning is as follows:
To include fully functional theories of QG and QED and GR... etc, one needs to program these laws of physics into a computer, which I can't imagine would be too difficult (assuming these laws are inherently algorithmically "simple"). The problem lies I believe with the initial conditions. I'm assuming you mean that to simulate the universe, you must also account for the mass energy distribution of the universe, which is where problems arise. To simulate an arbitrarily small region of spacetime, with even the most efficient memory algorithms, the physical memory itself requires a certain amount of physical space in a memory bank of sorts. Until the technology arrives, this physical memory space required for a simulation of an arbitrarily small region of spacetime will itself be larger than the space being simulated. For example, assume at the most efficient memory algorithms n hydrogen atom can simulated with m bits of memory information. However, the physical size of n bits of information is, say, x [itex]m^{3}[/itex], so as long as x [itex]m^{3}[/itex] exceeds the actual size of m hydrogen atoms, the computer itself will have to be larger than m H atoms. Ad infinitum, a computer simulating the entire universe would in fact be larger than the actual universe, and therefore must be accounted for in the simulation as well, leading to a very absurd conclusion. So as far as I can reason, no perfect simulation can exist.
 
Really nice thought experiment you got there :) As I have been discussing with Marcus, this is also the conclusion we have gotten to. We were thinking of doing course grained simulations rather than the full size / full version of the universe, going from large superclusters down to the quantum level. In such simulation, it would be possible to look into galaxy formation and largescale astrophysics phenomenology.
 

Related Threads for: Simulation of the universe

Replies
4
Views
1K
Replies
6
Views
597
  • Posted
Replies
11
Views
831
  • Posted
Replies
3
Views
773
Replies
1
Views
3K
Replies
3
Views
2K

Physics Forums Values

We Value Quality
• Topics based on mainstream science
• Proper English grammar and spelling
We Value Civility
• Positive and compassionate attitudes
• Patience while debating
We Value Productivity
• Disciplined to remain on-topic
• Recognition of own weaknesses
• Solo and co-op problem solving

Hot Threads

Top