Working Model of the Universe

[pcy]

Hi,


The Perfectly Smooth Elephant (whose weight may be neglected but whose gravitas certainly cannot) has submited a proposal for government funding to construct a Working Model of the Universe

See:
http://forums.anandtech.com/messageview...atid=50&threadid=1971863&enterthread=y
for more informationabout this Elephant.


I'm a bit suspicious about this idea, so I thought I'd find out what you experts think:

Proposal: to construct a Working Model of the Universe. This is possibly not as difficult as you might think. Just code up the laws of physics and the software is complette. OK, OK, loading up the initial data may prove a tad time consuming; but we could always solve that by going back to the big bang when the universe consisted of a point mass.

Anyway.. once you have it running it works like this:

Given a sufficiently fast CPU the model will run faster than real time. Lets assume it runs at twice real time: if not just make the CPU faster and faster until it does. So after 1 sec you look at the state of the universe in the model and you notice (as expected) that it holds the state as of 2 secs. But the model itself is contained in the universe, so this model contains its own state at time 2secs, which of course represents the state of the universe at time 4secs, which in turn conains the model at time 4secs, which...

This process continues indefinately so in fact the model within the model within the model shows the state of the universe at time 2*n and n is arbitrarily large.

So the model is in effect indefinately fast, so it must work with any CPU however slow, indeed stricly speaking without any CPU at all.

Very cheap then...


P. S. Elephant



Can this be right?



Peter

 

[DrPizza]

Just out of curiousity, where/how are you going to have information in your model? Since your model contains information about every part of the universe, you're going to need a memory as big as the universe.

Also, the laws of the universe say that even if you *did* know absolutely everything at some given moment in time, you would not necessarily know what's going to happen in the next instant, nor would you be able to calculate it. The best you could do is give a probability of something happening during the next instant. The quantum world is ruled by probability. i.e. if you have a million particles, you can predict that 3/4 of the million are going to do one thing, and the other quarter are going to do something else; but you cannot calculate what's going to happen for any particular particle.

edit: darn it! I wish I could remember the source; there's something to the effect that a system cannot have complete knowledge of itself; complete knowledge of a system requires more parts than the system has.

 

[Auryg]

How is one of our processors going to handle the physics of the universe? Even right now, just the rocks colliding out in the asteroid belt would be way, way, wayyy too much.

 

[bobsmith1492]

Yeah, um, if you read the other "elephant" thread this is basically HT spam.

 

[silverpig]

You can't use a standard cpu for this job realistically. The best you could do is to use a quantum computer, and this is fine, but in order to realistically simulate the entire universe, you'd need a computer the size of the universe. And this computer would essentially be a second universe...

 

[Xdreamer]

The problem with a model of the universe is that (besides the fact that the model must be contained within the universe that it is modeling) the model reaches the point where it is equally complex as the universe it models.

 

[randay]

wheres the part saying "if you dont post this in 3 other forums your universe will model you in your sleep" ?

 

[blackllotus]

The problem is that the bare minimum memory requirements for a computer, assuming it starts modeling from the present, happens to require all of the matter in the universe because the universe is the most optimal representation of itself (there is no redundancy).

Theoretically the memory requirements for such a computer tend to get smaller and smaller the further back in time we go. For example, if we were modeling the exact instant that our universe began the memory requirement would be 0 because everything was uniform (or so we think) and there was no information present. We should have enough resources in the universe today to model the past but not the future.

EDIT: Yes, what I just said is off-topic

 

[Gibsons]

Originally posted by: DrPizza
edit: darn it! I wish I could remember the source; there's something to the effect that a system cannot have complete knowledge of itself; complete knowledge of a system requires more parts than the system has.

Gödel?

 

[imported_Seer]

The obvious answer is that it's impossible due to infinite recursion . Said Modeling system would have to exist outside of the universe it was modeling in order for it to possibly model said universe.

Or

You might be able to put the computer inside a blackhole or some other suficently informationless zone. Maybe a giant sealed room. But then the computer would draw and emit heat according to the course that its claculations take. Because the actions of the universe follow a chaotic model, the discrepancy due to not factoring in the heat generated by the computer would build up over time, making your model useless for extrapolation. Also, the computer could never communicate its results of the possible future to anyone outside, or those results would become immediately invalid Quite the connundrum.

 

[AeroEngy]

Besides all of the other points already posted why this wouldn't work and the absurdity of it. The current age of the universe is estimated to be 13-14 Billion years old. (There are other estimates but this is reasonable approximation). The model of the universe would not exist inside the first model of the universe at T = 2sec it wouldn't appear until T = 14 Billion years from the Big Bang. So if your CPU could processes all the data at twice real time starting with a t=0 set at the Big Bang, the 1st nested model would not appear until 7 Billion years from now.

T_universe =21 Billion Years
T_model0 = 14 Billion Years
T_model1 = 0

At this time the model within itself would only be showing the universe at T=0 (Big Bang). So it would take an addition 7 Billion Years to show what is happening at the same time event are occurring in the real universe... also this will be when the second nested model appears.

T_universe = 28 Billion Years
T_model0 = 28 Billion Years
T_model1 = 14 Billion Years
T_model2 = 0

So it will take a minimum of 14 Billion years for this model to start showing any kind of theoretical returns in speed and surpass the current real time.

Also the model within itself would not be an actual CPU driven piece of hardware like the original one. Just a simulation of the results of the first one only time shifted. This nested model (and all others that follow) would draw resources form the first and increase its processor load slowing it down exponentially as every nested model appears. Eventually creating a model that was slower than real time and completely worthless.

 

[sdifox]

I would like to see a working, accurate model of the stock market before committing funding for the universe model.

 

[gerwen]

Can we accurately model a single atom yet?

 

[NikolaeVarius]

Accuratly making a model of an atom is impossible because of the uncertainty Principle

 

[tidehigh]

If you could accurately make a model of an atom, then you could accurately make a model of the universe.

 

[cougar1]

Define accurate?

Very accurate models of the electronic properties of the hydrogen atom are available. Significantly less accurate, but still quite useful models of the electronic behavior of all atoms are available in the various Quantum Chemistry software packages (eg. Gaussian, GAMESS, ADF, DMOL, Jaguar, etc...).

Presumably, models of atomic nuclei also exist, although I'm not really familiar with them, since for most cases, nuclear properties can be neglected (or handled in a very simplistic fashion) in most chemical processes.

 

[gerwen]

Apparently we can model atoms for specific purposes. One for chemical reactions, one for electrical properties, and probably one for nuclear processes. Until we can create a single model that accurately does all of these, i wouldn't really call it an accurate model, at least in a discussion about modelling the universe. To model the universe, your atomic model will have to work will in all cases. Nuclear, chemical, electromagnetic, etc.

Of course that's all just an intellectual exercise. Accurately modelling the universe down to the atomic scale is impossible.

 

[cougar1]

An "accurate" model of the hydrogen atom including both electronic and nuclear properties probably wouldn't be that hard to develop and possibly already exists. The problem is that such a model is not terribly useful, except as a pedagogical exercise. For most applications, the coupling between electronic and nuclear effects (the hyperfine interaction) is quite small and can be neglected, so including it introduces a lot of extra calculations with little improvement in overall accuracy.

Of course if one is particularly interested in studying these hyperfine interactions, which might be the case for advanced NMR spectroscopy, or some other highly specialized fields then these effects must be included in the model.

As for modeling the universe, the practical problem is that such complicated models of a single atom quickly becomes mathematically untenable, when extended to interactions between multiple atoms (or even a single atom and additional electrons). Although it should be noted that such a model would not correspond to just the atomic scale, but the subatomic scale.

 

[WildHorse]

Can't accurately model the universe, because the influences upon it are (perhaps mostly) unknown.

For example, consciousness affects the universe.

Can you model the psychology of every sentient being throughout the cosmos, and all stochastic interactions among them, which do indeed exert influence? No. Infeasible.

"Universe" includes everything, that's axiomatic. Universe includes everything physical, all thoughts, all emotions, all everything.

 

[NanoStuff]

Originally posted by: scott
Can't accurately model the universe, because the influences upon it are (perhaps mostly) unknown.

For example, consciousness affects the universe.

Can you model the psychology of every sentient being throughout the cosmos, and all stochastic interactions among them, which do indeed exert influence? No. Infeasible.

"Universe" includes everything, that's axiomatic. Universe includes everything physical, all thoughts, all emotions, all everything.

All thoughts emotions and everything are derived from the physical, poor argument. Regardless, yes it is infeasible to model it to such precision. But it's feasible to model it to a level of precision good enough to get meaningful results.

 

[WildHorse]

Originally posted by: NanoStuff
< cutout >
All thoughts emotions and everything are derived from the physical, poor argument. < cutout >

That's an unfounded assertion.

In the view of many traditions it's the other way around.
E.G.; Plato's theory of knowledge and his "forms" of a higher order of reality, which the lower order of the physical plane tries to approximately conform to, but very imperfectly.

Or, most of the spiritual philosophies I've read into seem to conceive of "reality" as projecting downward from a higher order into the physical level which is near the bottom of the whole works. I think Christianity, Hinduism, Islam, Judaism all fit this, altogether encompassing maybe 2/3 of earth's population who'd disagree with your assertion, as I do also.

The only ones I'm acquainted with that see things the way you put it are the deterministic materialists, the Communists, and maybe some of the Bhuddists, but I'm not sure of that last one.

And I remember Joseph Campbell saying in one of those Bill Moyer videos that he saw life as the "bouquet" of the earth. But I don't think you could interpret that in the "bottoms-up" way you put it, "All thoughts emotions and everything are derived from the physical," because he had a vastly sophisticated concept of what the earth really is.

Bottom line: I think your concept about that isn't right. :sun:

 

[NanoStuff]

I think Christianity, Hinduism, Islam, Judaism all fit this, altogether encompassing maybe 2/3 of earth's population who'd disagree with your assertion

You're right, how silly of me. Let's go ask Ronald McDonald what the answer is.

 

[WildHorse]

Originally posted by: NanoStuff

I think Christianity, Hinduism, Islam, Judaism all fit this, altogether encompassing maybe 2/3 of earth's population who'd disagree with your assertion

You're right, how silly of me. Let's go ask Ronald McDonald what the answer is.

non sequitur

you were not being attacked hey I thought we were up for a discussion & that's my perspective.

 

[imported_iNGEN]

In Soviet Russia, universe model you! Sorry, had to do it.

Anyway, the proposed model is impossible for the very same reason it is infinitely fast.

OK, OK, loading up the initial data may prove a tad time consuming; but we could always solve that by going back to the big bang when the universe consisted of a point mass.

The initial post also suggests every event is the result of a deterministic mechanism. Given human understanding of physics, I think that one HELL of a presumption.

 

[IdaGno]

GIGO