|
On February 24 2017 10:35 Simberto wrote:Show nested quote +On February 24 2017 10:16 Uldridge wrote: I was just thinking from a purely computing standpoint. You need electrons to visualize the attributes of energy, matter, space, etc. You'd also need more electrons than one electron to model an electron doing its stuff. So howmuch energy does one need to model one hydrogen atom, for example? What about Helium? What about space? It's a pretty void question in a sense because absolute numbers for energy can't be accounted for.. Well, no. One electron is obviously a perfect model of an electron. Thus, the maximum theoretical amount of stuff you need to simulate the universe is a universe. We are far worse than that. If you use our current known math and computers to describe even a moderately complex system (Basically, anything involving more than very few elementary particles, you will a) not be exact, and b) use many orders of magnitudes more energy. Lets just say we are talking about something like a carbon atom. That means 6 protons, 6 neutrons, 6 electrons. The C-12 atom has a mass of 12u, which means it has mass-equivalent energy of ~11.2 GeV. This sounds like a large number because of the "Giga", but it really is not. 1J = 6.24*10^18 eV, or 6.24*10^9 GeV. About half a billion times more. Now, 1J is the amount of energy that you would require to power something that needs 1W for 1s. So, with the energy equivalent of a C-12 atom, you could power a 1W computer for around two billionth of a second. So if you manage to build a computer with 4GhZ, that only requires 1W of power, you could do a grand total of 8 calculation operations with the power equivalent of 1 carbon atom. You can not simulate a carbon atom with anything close to 8 basic operations. That shit is hard. Which leads us to the next problem. The atom simply stays there. But to continue simulating it with a computer, you constantly require more energy. So, from our current perspective, the best way to simulate a universe is to just build a universe (If you want togo down to the particle level). On the other hand, if you only care about big picture stuff, that shit is easy. You can simulate a solar system on a single pc easily, if you only care about gravitation and don't need to be too exact. The more exact you want to get, the harder it gets.
Scale is also important. I could put a quarter on the table and say that that quarter was the universe; it wouldn't be a very useful model but it would be cheap to make and require zero power.
|
What about using the limitless power of your imagination!
|
On February 24 2017 12:15 JimmiC wrote: What about using the limitless power of your imagination!
Also only works by using scale. You can imagine a planet, but you can't even imagine earth in detail down to every single person, and that is not even talking about the additional 10 orders of magnitude, and thus ~10^30 times as much stuff when you get down to atomic scales.
Your imagination sucks balls.
|
Maybe yours does, you don't know mine.... Its sweet
|
On February 24 2017 12:36 JimmiC wrote: Maybe yours does, you don't know mine.... Its sweet
In that case, use your imagination to get paid what people pay for time with a supercomputer.
|
On February 24 2017 12:43 Simberto wrote:Show nested quote +On February 24 2017 12:36 JimmiC wrote: Maybe yours does, you don't know mine.... Its sweet In that case, use your imagination to get paid what people pay for time with a supercomputer.
Supercomputers get $0
Its better to get a random degree along with an MBA and convince people to give you money for arbitrary reasons.
|
On February 24 2017 12:43 Simberto wrote:Show nested quote +On February 24 2017 12:36 JimmiC wrote: Maybe yours does, you don't know mine.... Its sweet In that case, use your imagination to get paid what people pay for time with a supercomputer.
I'm imagining that cash right now!
|
Can you simulate anything exactly? No. Not with today's computers. Real things have continuous properties (yes also quantum things), while computers operate in discrete numbers. No supercomputer, no matter what size, will ever be able to exactly describe the momentum of an election.
Can you, in principle, simulate the universe approximately? Well, yes. In principle. Depends on how accurate you want to be. If you want it to be accurate enough to reproduce life naturally, you probably need a computer the size of a galaxy+. With a very targeted simulation tailor made to simulate life you can probably get away with less.
Can we do why of the above today in practice? No, not in a way that we actually learn anything from it. It would need to be so hard coded towards describing life that it would describe how we think life appears rather than how it actually appears. Some are probably trying (seems like something NASA would try for lulz), but I doubt the results are very informative. At most worthy of "huh cool simulation!", which I feel it's what NASA is going for sometimes.
|
I am pretty sure that to simulate all particles in the universe in real time you would need at least several universes worth of matter. That's simply because the universe is absolutely terrible in being a detailed simulation of itself - from a simulation, I expect some outputs, but most of the information in the universe is inaccessible, a lot of it even principially, due to quantum uncertainty and the measurement affecting the system. There is no way to ask an electron to do more than just being an electron and that is already not enough to provide complete information about it's properties in time, so it seems quite obvious to me that for a perfect simulation of N particles, you need significantly more particles. Or, alternatively, you could use less particles, but then you would simulate much slowly than time goes, say one second of time in an hour or something, so that would also not be a simulation of the "whole universe" in a spacetime manner.
|
On February 24 2017 14:45 Cascade wrote:Can you simulate anything exactly? No. Not with today's computers. Real things have continuous properties (yes also quantum things), while computers operate in discrete numbers. No supercomputer, no matter what size, will ever be able to exactly describe the momentum of an election. Can you, in principle, simulate the universe approximately? Well, yes. In principle. Depends on how accurate you want to be. If you want it to be accurate enough to reproduce life naturally, you probably need a computer the size of a galaxy+. With a very targeted simulation tailor made to simulate life you can probably get away with less. Can we do why of the above today in practice? No, not in a way that we actually learn anything from it. It would need to be so hard coded towards describing life that it would describe how we think life appears rather than how it actually appears. Some are probably trying (seems like something NASA would try for lulz), but I doubt the results are very informative. At most worthy of "huh cool simulation!", which I feel it's what NASA is going for sometimes.
Verdict is still out on whether the universe is continuous or discrete when you break it down small enough. There are plenty of discrete space-time models that work in theory. Plus, simulating continuous properties is easy, and we do it all the time. More problematic is that we don't understand the universe well enough to even make a dent. We can't even build a good model of a proton (for starters, what is its size?), let alone enough of them to take a real stab at simulating the universe. Moreover, we don't even know for sure what an electron is... and that's just talking about the stuff we know of and can study with relative ease.
|
On February 24 2017 18:01 Acrofales wrote:Show nested quote +On February 24 2017 14:45 Cascade wrote:Can you simulate anything exactly? No. Not with today's computers. Real things have continuous properties (yes also quantum things), while computers operate in discrete numbers. No supercomputer, no matter what size, will ever be able to exactly describe the momentum of an election. Can you, in principle, simulate the universe approximately? Well, yes. In principle. Depends on how accurate you want to be. If you want it to be accurate enough to reproduce life naturally, you probably need a computer the size of a galaxy+. With a very targeted simulation tailor made to simulate life you can probably get away with less. Can we do why of the above today in practice? No, not in a way that we actually learn anything from it. It would need to be so hard coded towards describing life that it would describe how we think life appears rather than how it actually appears. Some are probably trying (seems like something NASA would try for lulz), but I doubt the results are very informative. At most worthy of "huh cool simulation!", which I feel it's what NASA is going for sometimes. Verdict is still out on whether the universe is continuous or discrete when you break it down small enough. There are plenty of discrete space-time models that work in theory. Plus, simulating continuous properties is easy, and we do it all the time. More problematic is that we don't understand the universe well enough to even make a dent. We can't even build a good model of a proton (for starters, what is its size?), let alone enough of them to take a real stab at simulating the universe. Moreover, we don't even know for sure what an electron is... and that's just talking about the stuff we know of and can study with relative ease. ok, it may be step-wise at tiny scales. Who knows what the future will bring. But as physics is now, it's continuous. And we don't have any continuous values when we simulate, we always approximate.
|
On February 24 2017 18:04 Cascade wrote:Show nested quote +On February 24 2017 18:01 Acrofales wrote:On February 24 2017 14:45 Cascade wrote:Can you simulate anything exactly? No. Not with today's computers. Real things have continuous properties (yes also quantum things), while computers operate in discrete numbers. No supercomputer, no matter what size, will ever be able to exactly describe the momentum of an election. Can you, in principle, simulate the universe approximately? Well, yes. In principle. Depends on how accurate you want to be. If you want it to be accurate enough to reproduce life naturally, you probably need a computer the size of a galaxy+. With a very targeted simulation tailor made to simulate life you can probably get away with less. Can we do why of the above today in practice? No, not in a way that we actually learn anything from it. It would need to be so hard coded towards describing life that it would describe how we think life appears rather than how it actually appears. Some are probably trying (seems like something NASA would try for lulz), but I doubt the results are very informative. At most worthy of "huh cool simulation!", which I feel it's what NASA is going for sometimes. Verdict is still out on whether the universe is continuous or discrete when you break it down small enough. There are plenty of discrete space-time models that work in theory. Plus, simulating continuous properties is easy, and we do it all the time. More problematic is that we don't understand the universe well enough to even make a dent. We can't even build a good model of a proton (for starters, what is its size?), let alone enough of them to take a real stab at simulating the universe. Moreover, we don't even know for sure what an electron is... and that's just talking about the stuff we know of and can study with relative ease. ok, it may be step-wise at tiny scales. Who knows what the future will bring. But as physics is now, it's continuous. And we don't have any continuous values when we simulate, we always approximate. But we can approximate down to whatever number of decimals you could possibly want (given enough time). So it's essentially the same thing. At the moment, our ability to approximate the continuous properties equals or even exceeds our ability to measure them.
|
On February 24 2017 18:07 Acrofales wrote:Show nested quote +On February 24 2017 18:04 Cascade wrote:On February 24 2017 18:01 Acrofales wrote:On February 24 2017 14:45 Cascade wrote:Can you simulate anything exactly? No. Not with today's computers. Real things have continuous properties (yes also quantum things), while computers operate in discrete numbers. No supercomputer, no matter what size, will ever be able to exactly describe the momentum of an election. Can you, in principle, simulate the universe approximately? Well, yes. In principle. Depends on how accurate you want to be. If you want it to be accurate enough to reproduce life naturally, you probably need a computer the size of a galaxy+. With a very targeted simulation tailor made to simulate life you can probably get away with less. Can we do why of the above today in practice? No, not in a way that we actually learn anything from it. It would need to be so hard coded towards describing life that it would describe how we think life appears rather than how it actually appears. Some are probably trying (seems like something NASA would try for lulz), but I doubt the results are very informative. At most worthy of "huh cool simulation!", which I feel it's what NASA is going for sometimes. Verdict is still out on whether the universe is continuous or discrete when you break it down small enough. There are plenty of discrete space-time models that work in theory. Plus, simulating continuous properties is easy, and we do it all the time. More problematic is that we don't understand the universe well enough to even make a dent. We can't even build a good model of a proton (for starters, what is its size?), let alone enough of them to take a real stab at simulating the universe. Moreover, we don't even know for sure what an electron is... and that's just talking about the stuff we know of and can study with relative ease. ok, it may be step-wise at tiny scales. Who knows what the future will bring. But as physics is now, it's continuous. And we don't have any continuous values when we simulate, we always approximate. But we can approximate down to whatever number of decimals you could possibly want (given enough time). So it's essentially the same thing. At the moment, our ability to approximate the continuous properties equals or even exceeds our ability to measure them. Agreed, but when I said "exact", I meant just that: infinite precision, not just close enough. You can use the entire universe to build a computer to store the momentum of an electron, and it'll still not be exact, as in the different is exactly 0, not only negligible. A continuous variable has infinitely many digits. "Close enough" is what I talk about in my second point.
|
On February 24 2017 22:46 Cascade wrote:Show nested quote +On February 24 2017 18:07 Acrofales wrote:On February 24 2017 18:04 Cascade wrote:On February 24 2017 18:01 Acrofales wrote:On February 24 2017 14:45 Cascade wrote:Can you simulate anything exactly? No. Not with today's computers. Real things have continuous properties (yes also quantum things), while computers operate in discrete numbers. No supercomputer, no matter what size, will ever be able to exactly describe the momentum of an election. Can you, in principle, simulate the universe approximately? Well, yes. In principle. Depends on how accurate you want to be. If you want it to be accurate enough to reproduce life naturally, you probably need a computer the size of a galaxy+. With a very targeted simulation tailor made to simulate life you can probably get away with less. Can we do why of the above today in practice? No, not in a way that we actually learn anything from it. It would need to be so hard coded towards describing life that it would describe how we think life appears rather than how it actually appears. Some are probably trying (seems like something NASA would try for lulz), but I doubt the results are very informative. At most worthy of "huh cool simulation!", which I feel it's what NASA is going for sometimes. Verdict is still out on whether the universe is continuous or discrete when you break it down small enough. There are plenty of discrete space-time models that work in theory. Plus, simulating continuous properties is easy, and we do it all the time. More problematic is that we don't understand the universe well enough to even make a dent. We can't even build a good model of a proton (for starters, what is its size?), let alone enough of them to take a real stab at simulating the universe. Moreover, we don't even know for sure what an electron is... and that's just talking about the stuff we know of and can study with relative ease. ok, it may be step-wise at tiny scales. Who knows what the future will bring. But as physics is now, it's continuous. And we don't have any continuous values when we simulate, we always approximate. But we can approximate down to whatever number of decimals you could possibly want (given enough time). So it's essentially the same thing. At the moment, our ability to approximate the continuous properties equals or even exceeds our ability to measure them. Agreed, but when I said "exact", I meant just that: infinite precision, not just close enough. You can use the entire universe to build a computer to store the momentum of an electron, and it'll still not be exact, as in the different is exactly 0, not only negligible. A continuous variable has infinitely many digits. "Close enough" is what I talk about in my second point.
It is hard to tell if that stuff is actually continuous, though. A lot of things that you'd think would be continuous are actually not. Stuff is weird when you look at elementary particles.
(Also, a real number does not have to have infinite decimal numbers. Though in the vast majority of cases, it does)
You could also try find a way to use analog data in your simulation. I don't know how feasible it is to build an analog calculator, though, that is a thematic i have no idea about.
|
On February 24 2017 22:46 Cascade wrote:Show nested quote +On February 24 2017 18:07 Acrofales wrote:On February 24 2017 18:04 Cascade wrote:On February 24 2017 18:01 Acrofales wrote:On February 24 2017 14:45 Cascade wrote:Can you simulate anything exactly? No. Not with today's computers. Real things have continuous properties (yes also quantum things), while computers operate in discrete numbers. No supercomputer, no matter what size, will ever be able to exactly describe the momentum of an election. Can you, in principle, simulate the universe approximately? Well, yes. In principle. Depends on how accurate you want to be. If you want it to be accurate enough to reproduce life naturally, you probably need a computer the size of a galaxy+. With a very targeted simulation tailor made to simulate life you can probably get away with less. Can we do why of the above today in practice? No, not in a way that we actually learn anything from it. It would need to be so hard coded towards describing life that it would describe how we think life appears rather than how it actually appears. Some are probably trying (seems like something NASA would try for lulz), but I doubt the results are very informative. At most worthy of "huh cool simulation!", which I feel it's what NASA is going for sometimes. Verdict is still out on whether the universe is continuous or discrete when you break it down small enough. There are plenty of discrete space-time models that work in theory. Plus, simulating continuous properties is easy, and we do it all the time. More problematic is that we don't understand the universe well enough to even make a dent. We can't even build a good model of a proton (for starters, what is its size?), let alone enough of them to take a real stab at simulating the universe. Moreover, we don't even know for sure what an electron is... and that's just talking about the stuff we know of and can study with relative ease. ok, it may be step-wise at tiny scales. Who knows what the future will bring. But as physics is now, it's continuous. And we don't have any continuous values when we simulate, we always approximate. But we can approximate down to whatever number of decimals you could possibly want (given enough time). So it's essentially the same thing. At the moment, our ability to approximate the continuous properties equals or even exceeds our ability to measure them. Agreed, but when I said "exact", I meant just that: infinite precision, not just close enough. You can use the entire universe to build a computer to store the momentum of an electron, and it'll still not be exact, as in the different is exactly 0, not only negligible. A continuous variable has infinitely many digits. "Close enough" is what I talk about in my second point. But if we're down to infinite precision, then we're back at the question of how our universe works at the tiniest of scales... which may be discrete. And to know whether your model is then right or wrong, you have to know that. Because what is the difference between a model that is right, and a model that is wrong, but we can never prove it, because insofar as it predicts stuff, it coincides exactly with measurements up to the highest level of significance possible (there is a theoretical limit to our ability to observe the smallest things).
Both are reliable models of the universe, but only one is valid. However, we can never distinguish between the two.
This reminds me a lot of a looooong discussion I had with a friend about whether simulated intelligence is intelligence (my stance is that yes, it is).
|
On February 24 2017 23:20 Simberto wrote: (Also, a real number does not have to have infinite decimal numbers. Though in the vast majority of cases, it does)
Actually they do, be it an infinity of zeros. We define them as not relevant, but they are there, and an infinity number of them.
On the same matter, the number 0.9999999... (with an infinity of 9) is (exactly) equal to 1.
|
We have hopped on to the infinity train, destination weird.
|
On February 24 2017 23:20 Simberto wrote:Show nested quote +On February 24 2017 22:46 Cascade wrote:On February 24 2017 18:07 Acrofales wrote:On February 24 2017 18:04 Cascade wrote:On February 24 2017 18:01 Acrofales wrote:On February 24 2017 14:45 Cascade wrote:Can you simulate anything exactly? No. Not with today's computers. Real things have continuous properties (yes also quantum things), while computers operate in discrete numbers. No supercomputer, no matter what size, will ever be able to exactly describe the momentum of an election. Can you, in principle, simulate the universe approximately? Well, yes. In principle. Depends on how accurate you want to be. If you want it to be accurate enough to reproduce life naturally, you probably need a computer the size of a galaxy+. With a very targeted simulation tailor made to simulate life you can probably get away with less. Can we do why of the above today in practice? No, not in a way that we actually learn anything from it. It would need to be so hard coded towards describing life that it would describe how we think life appears rather than how it actually appears. Some are probably trying (seems like something NASA would try for lulz), but I doubt the results are very informative. At most worthy of "huh cool simulation!", which I feel it's what NASA is going for sometimes. Verdict is still out on whether the universe is continuous or discrete when you break it down small enough. There are plenty of discrete space-time models that work in theory. Plus, simulating continuous properties is easy, and we do it all the time. More problematic is that we don't understand the universe well enough to even make a dent. We can't even build a good model of a proton (for starters, what is its size?), let alone enough of them to take a real stab at simulating the universe. Moreover, we don't even know for sure what an electron is... and that's just talking about the stuff we know of and can study with relative ease. ok, it may be step-wise at tiny scales. Who knows what the future will bring. But as physics is now, it's continuous. And we don't have any continuous values when we simulate, we always approximate. But we can approximate down to whatever number of decimals you could possibly want (given enough time). So it's essentially the same thing. At the moment, our ability to approximate the continuous properties equals or even exceeds our ability to measure them. Agreed, but when I said "exact", I meant just that: infinite precision, not just close enough. You can use the entire universe to build a computer to store the momentum of an electron, and it'll still not be exact, as in the different is exactly 0, not only negligible. A continuous variable has infinitely many digits. "Close enough" is what I talk about in my second point. It is hard to tell if that stuff is actually continuous, though. A lot of things that you'd think would be continuous are actually not. Stuff is weird when you look at elementary particles. (Also, a real number does not have to have infinite decimal numbers. Though in the vast majority of cases, it does) You could also try find a way to use analog data in your simulation. I don't know how feasible it is to build an analog calculator, though, that is a thematic i have no idea about. Well, as physics is today, things are continuous. If things are different and we don't know about it, that's just another reason why we can't do exact simulations.
Some weird new physics that replaces today's physics and for whatever reason allows us to store the exact information in very few digits? I guess that is possible in theory. It's not the direction things have moved so far in physics though, but who knows.
And yes, any real value in the nature has infinitely many non zero digits. The probability is zero that an infinite string of random digits are all 0. I guess you can set your coordinates in the simulation so that one particle has zero position and momentum, but apart from that...
|
If reality isn't continuous, you have information theory problems. A given set of matter in a given space requires a certain amount of information to describe it. You can't store that information with a matter-plus-space combination that has less information.
If it is continuous, you can do the old 'store the digits of one decimal in the odd digits and another decimal in the even digits' trick. But good luck implementing that scheme in such a way that you can actually advance the simulation.
|
How do I let birds know I'm friendly?
|
|
|
|