According to Moore’s law the capabilities of processors double about every 18 months. This is exponential growth. Soon we will have very very powerfull computational abilities. My question is:
How many times does the speed have to double before we come up with a PC that can simulate all of the particles in the universe? Were you to have a machine that could do this, combined with a healthy knowledge of physics, you could create your own simulated universe, right? Down to the quantum level how many particles do we need to process? I guess we have to assume the universe is closed for it to work. Infinite systems would be hard to design. I think it raises some interesting design issues. Do you have to program everything, or can you just enter the eqations into a big enough box and run a whole big bang scenario!!! It would be the best version of “The Sims” ever!
I’m hoping it comes in around 2017, I’ll be needing a new universe about then. After I’ve laid waste to this one. . . Damn. . . . I’ve said too much.
You might want to scearch quantum computers, I beleave that Scientific American has done a few articals on them. With out quantum computers Moore’s law will run out of steam long before we have that kind of computing power.
Moore’s Law isn’t a law. Hardly even a theory. All in all, it is soon going to reach a point where we are limited by the size of molecular transistors. Quantum computers are still a twinkle in the eyes of mad scientists.
Measuring the position, size, speed, and direction of every particle in the universe would involve using every particle in the universe to do so. Ain’t gonna happen.
Best we can hope for is The Sims: Supar Boring Even More Detailed Repetitive Action Lucky Happy Extravaganza !!!
Yes, suppose you have a simple universe which only contains ten fundamental indivisible entities; each of these entities has only four distinct and indivisible properties; your system has forty parameters that you must keep track of; how are you going to store those properties in less than forty spaces? not to mention that the system modelling them requires some overhead just in order to exist.
It’s no good even if you suddenly discover that the four properties of each particle are divisible into three subproperties; theoretically this gives you three times more storage space in which to describe the system, but you now have to describe a system that is three times more complex. No win.
Of course it isn’t really a law. It is still referred to as a law even if it is a misnomer. Folks seem to think that it will stop eventually, others think it’ll keep on trucking. Here are some links that show research that suggests we may keep going:
You don’t have to do that. You just have to figure out the laws that govern the particles interactions with each other, and then enter in the number of particles that you want your simulated universe to have, and voila - simulated universe. I’m not looking for a replica of this universe, more like another instance that unfolds in a totally different way based on probabilities, chance, and physical laws.
IANA Progammer, so I don’t really know. I do know that my PS2 is able to hold entire cities in a space smaller than a shoe box: like in the game GTA 3. Granted, the resolution is not down to a quantum level, but it surely takes alot less space than it appears I have access too. The universe will be viewed on a screen, the space that gets taken up is whatever it needs to process the math. The army’s flight simulators don’t require the whole planet to simulate flying over the whole planet.
I think the questions that need to get answered are:
How much matter is in the universe?
How many particles make up that matter?
How many calculations do you have to do to simulate that many particles interacting with each other?
Assuming Moore’s law holds, how many generations until we can do that many calculations on our desktops?
If you want to simulate all the particles in the universe it’s not enough just to input how many there are. You also have to store information about each one (location, velocity, charge, whatever). This is true even if you don’t try to replicate the real universe’s actual state. All simulations work this way: Load the state of the system from memory, apply some mathematical laws, save the state back to memory.
How are you going to store the complete state of the universe in something smaller than the universe?
Consider this: Somewhere in your SimUniverse there will probably be a planet with intelligent lifeforms. If they are sufficiently advanced they will be able to build a SimSimUniverse. And in the SimSimUniverse will be other intelligent lifeforms, who build a SimSimSimUniverse.
Do you see where the problem is now?
IAA Programmer, so I do know. GTA3 isn’t really simulating a city. It’s only simulating a few dozen cars and people in your immdiate vacinity. As soon as you drive away to someplace new they all vanish. The buildings are permanent, of course, but that’s an easier problem because they don’t change (so there’s nothing to simulate).
I believe that the number of elementary particles in the observable universe is somewhere between 10[sup]72[/sup] and 10[sup]80[/sup]. But then you would also need to model all of the virtual particles as well. Good luck!
Allright. Let’s assume the universe is a closed system. There is only a specific & finite amount of matter and therefore a specific and finite amount of particles. If we can glean that number then we can figure out how many calculations it takes to run a simulation of the universe. You suggest that we will never be able to make that many calculations. Fair enough. We’ll see.
For the sake of argument, let’s assume Moore’s law never stops. First let’s plot a point on a chart that shows how many calculations it would take to simulate a universe like this one. Then let’s plot the number of calculations we can (and will be able to) do on the same chart according to Moore’s law. Eventually, our line will pass the point that represents the number of calculations required to simulate the universe - if you chart it far enough off. How many generations is my question? We’ll leave keeping up with Moore’s law to the lab department.* I just want to know, at the current clip (assuming moore’s law), how long until we should be able to do the necessary amount of calculations?
*Whether or not Moore’s law will persist is more like a great debate, I think. There should be a factual answer to how many calculations required - whether we can reach that many is definately debatable.
You may be interested in Frank Tipler’s Physics of Immortality. He touches on this kind of thing in there. It involved nesting exponents to represent numbers so godawful huge that it was the equivalent of just writing “a bazillion!”
It was actually reading Tipler which sent me down this road of thought. One of his ideas is that we can escape the end of the universe by turning the last era of the universe into a subjectively infinite time period by speeding up our perception. So we make a machine that can run out a whole universe’s history 8 billion times a second, and then we jump in. So we effectively experience an infinity of time in a finite amount of time. He called it the Omega Point. Where do I sign up?
Great start! So how many calculations does it take to chart that many particles over the life of the universe (let’s assume a big crunch)? How many calculations per second can we do now? I am not really good at math, so if someone could punch these numbers together I’d be very interested to see how many years we end up having to wait.
It’s not a question of how fast you can do the calculations, it’s a question of how you store the data you’re operating on.
Here’s how all simulations work. You have a set of data that captures the essential features of what you’re trying to simulate. You use that data set as the input for a set of numeric calculations. You write the results of your calculations back out to the data set. Repeat ad infinitum.
The simulation you’re describing needs to store information for every particle in the universe. This means that your storage system must have more particles than there are in the universe, a logical contradiction. Hence such a system is impossible.
Further, is the only number a computer can count to less than the amount of particles in the universe? My understanding is that a computer can count higher than all the particles in the universe. Say we take rsa’s numbers: A computer can consider numbers larger.
What’s different here from a program that lets you view a car at human scale, or zoom in to see the atomic structure of the paint? You should be able to represent the universe matmatically without taking up a whole universe just like you can represent a jet engine in a pc without building the whole engine.
Take the number of particles in the universe and now instead of plotting particles, make a game that has that many cars plotted on a really big highyway. It just takes alot of math - it’s not impossible. The factual answer I’m looking for is how many cacluclations are required to plot out a universe from start to finish. Let’s stay away from how we will achieve the amount of processing power to do it. Once we know how many calculations it should be s simple matter to figure out how many times Moore’s law needs to flip until we get there.
I think you have only two ways to do what you want.
Build your computer in another larger universe that would accomodate its size.
If you are willing to calculate the universe less than perfectly, you may be able to do that. For example, rather than simulating every particle in the Sun I think that you can get a pretty good aproximation using fluid dynamics (or something) instead.
I can’t help much with the math, but will be happy to be your “idea guy” if I get demi-god status in your universe.
So then how can computers work with numbers greater than the amount of particles it takes to make that computer? How can they add subtract and divide huge numbers together if those numbers are larger than the amount of particles contained therein?
Couldn’t a computer multiply the 2 numbers in rsa’s post by each other and give us an answer? The computer could do this without having that many particles in it.
You’re confusing two different ways of using numbers. A computer can certainly represent a number greater than the number of particles in the universe (10^1000, for example). But that doesn’t mean that it can enumerate those particles (i.e. specify each individually).
If you’re going to simulate the behavior of the universe at a quantum level you need to do the latter.
That’s because simulation software ignores the details that don’t matter for the particular domain you’re interested in. A simulation of a jet engine doesn’t model every atom in the engine, instead it makes generalizations based on how we know metal to behave. But this means that each particular simulation can only be used for a specific purpose. A particular simulation of a jet engine may present a very accurate picture of how a turbine blade will fail under stress, but not tell you anything about how radar waves bounce off its surface. Your universe simulation requires that you simulate the engine in its entirety – a much, much, much, much, much harder proposition.
As I keep explaining, it is impossible. Really, truely, impossible – unless you imagine something totally outside of known science (constructing an alternate universe several times the size of ours to use as your calculating device, for example.)
But since you’re being bloody minded … a rough estimate would be the number of particles in the universe (10^80) squared multiplied the age of the universe in Planck time units (10^-43 seconds).
10^18= rough age of universe in seconds
10^43 = number of Planck time units in a second
10^160= number of particles in the universe, squared
Multiply them together … 10^221 calculations … bigger than a googol … .
Of course I’m probably off by an factor of a billion or a trillion, but at this scale that’s trivial … .
Because the amout of storage required to represent a number is much less than the amount required to enumerate it. For example, I can write one million with seven symbols: 1000000. Seven is much less than one million. I can perform all sorts of operations on the number one million with only a few more symbols: 1000000 * 1000000 = 1000000000000. But if I want to enumerate one million – to actually identify each individual number between one and one million – then I have to use more than one million symbols to do it.
Creating SimUniverse requires that you don’t just count the number of particles in the universe. You actually have to enumerate them – to specify each one individually and record its properties. And that would require a storage device larger than the universe itself.