Why you can't simulate an universe - interesting paradox

[Wictorian]

So in order to simulate an atom, you would need more data points than that can be stored in an atom. Thus, you would need bigger servers than the universe itself to simulate it.

[Sarra]

4 hours ago, Wictorian said:

So in order to simulate an atom, you would need more data points than that can be stored in an atom. Thus, you would need bigger servers than the universe itself to simulate it.

There would appear to be a logic flaw in this statement...

[Wictorian]

2 minutes ago, Sarra said:

There would appear to be a logic flaw in this statement...

What is it?

[Sarra]

27 minutes ago, Wictorian said:

What is it?

It is possible to simulate an airfoil in a fluid (air); both the airfoil and the medium consist of quadrillions of molecules, yet a simple CPU can simulate it.

[Syaoran]

You do realize computers will advance higher and higher, and so will science.
So whose to say some kind of alien civilization(s) reached that kind of technology and made this universe we are now in?

[Master Disaster]

5 hours ago, Wictorian said:

So in order to simulate an atom, you would need more data points than that can be stored in an atom. Thus, you would need bigger servers than the universe itself to simulate it.

What? This doesn't make any sense, even at the most fundamental level.

 

Increasing the amount of data points inside a computer doesn't cause the computer to increase in size, you can literally store billions of bits on a device that fits in your pocket.

 

Plus you don't need to simulate the entire thing atom by atom anyway, most desktop PCs can simulate an aeroplane flying despite being smaller than a plane by a factorial.

 

Simulating an atom is impossible, see Heisenberg's uncertainty principle for the reason why.

[straight_stewie]

49 minutes ago, Sarra said:

It is possible to simulate an airfoil in a fluid (air); both the airfoil and the medium consist of quadrillions of molecules, yet a simple CPU can simulate it.

This is quite an interesting proposition.

Is the computer simulating the airflow, or just doing a really good job of emulating it?

One could say that the computer is just emulating the fluid flow, which we know it is: The techniques used in CFD are just approximations of real life. One could also say that the emulation is good enough for the task at hand.

Which of course, ties into the question: How much of an understanding is enough? Sure our understanding of some things is flawed or incomplete, but does that mean that our understanding of those things is useless?

[_princess_]

use bsp and frustum technology, make atoms only exist when you look at them.

[straight_stewie]

43 minutes ago, Master Disaster said:

most desktop PCs can simulate an aeroplane flying despite being smaller than a plane by a factorial.

Even though we call them "flight simulators", that's somewhat of a misnomer. In most of them, the flight characteristics simulations are driven largely by lookup tables of precomputed values. Which is why most absolutely suck at simulating turbulence. X-Plane is a little better at this than FS, Prepar3D, and DCS world, but not that much better. They are getting pretty good at systems simulations these days, but failure simulations are still lacking.

To really rest this topic some discussion needs to be had on the difference between simulation and emulation, and where the line is drawn between the two. Most people use the term simulation as if it means an emulation that is close enough to real life. But that's a very ambiguous definition, as no one is ever going to agree on how close is close enough.

[Master Disaster]

8 minutes ago, straight_stewie said:

Even though we call them "flight simulators", that's somewhat of a misnomer. In most of them, the flight characteristics simulations are driven largely by lookup tables of precomputed values. Which is why most absolutely suck at simulating turbulence. X-Plane is a little better at this than FS, Prepar3D, and DCS world, but not that much better. They are getting pretty good at systems simulations these days, but failure simulations are still lacking.

To really rest this topic some discussion needs to be had on the difference between simulation and emulation, and where the line is drawn between the two. Most people use the term simulation as if it means an emulation that is close enough to real life. But that's a very ambiguous definition, as no one is ever going to agree on how close is close enough.

Oh I know, but those fully boxed off training sims they use to actually train pilots have surprisingly low powered hardware for the simulation and also many profession training simulators do actually use modified "game" simulators.

 

X Plane offer a professional license and combined with approved flight controls can be used by professional pilots to log official training hours in.

 

This was kind of my point, its not necessary to simulate every aspect perfectly to get a good enough result out of the other end. We can cheat and leave bits out that we don't need to reduce strain on the systems.

[Wictorian]

8 hours ago, Master Disaster said:

.

 

Increasing the amount of data points inside a computer doesn't cause the computer to increase in size, you can literally store billions of bits on a device that fits in your pocket.

 

Plus you don't need to simulate the entire thing atom by atom anyway, most desktop PCs can simulate an aeroplane flying despite being smaller than a plane by a factorial.

 

 

It doesn't ? Why is an harddisk bigger than an sd card? There are billion of data points stored in a phone, but there are trillions of data points that are needed to simulate it. 

 

Yeah you don't but, I am talking about a universe as detailed as our universe. Then there is GTA and many other games that can count as an universe. Also if this is a simulation maybe reality is more detailed than our universe.

[Wictorian]

8 hours ago, Syaoran said:

You do realize computers will advance higher and higher, and so will science.
So whose to say some kind of alien civilization(s) reached that kind of technology and made this universe we are now in?

Yeah but for example speed of the light is the maximum speed we can achieve. I can it can be surprassed but that kind of advancement seems much more improbable.

[Wictorian]

8 hours ago, Caroline said:

Tell em to lower the difficulty, countries are unbalanced af, man these lazy devs....

It wouldn't be as realistic, would it?

[Wictorian]

8 hours ago, _princess_ said:

use bsp and frustum technology, make atoms only exist when you look at them.

It wouldn't be realistic enough.there are many things in space we don't see, it would fuck up the orbits.

[Wictorian]

7 hours ago, Master Disaster said:

 

 

This was kind of my point, its not necessary to simulate every aspect perfectly to get a good enough result out of the other end. We can cheat and leave bits out that we don't need to reduce strain on the systems.

Yeah we don't, but that's not what I'm talking about. I am talking about simulating everything. Then we could even accept Xplane as a simulation.

[Wictorian]

7 hours ago, straight_stewie said:

To really rest this topic some discussion needs to be had on the difference between simulation and emulation, and where the line is drawn between the two. Most people use the term simulation as if it means an emulation that is close enough to real life. But that's a very ambiguous definition, as no one is ever going to agree on how close is close enough.

AN emulation that is close enough to real life is entirely possible. That's not what I am talking about, I am talking about simulating everything. Every star and the entirety of the universe. Electrons, protons, quarks and stuff we don't  yet know about.

[Spindel]

Isn't the reasoning in OP flawed?

 

Or at least not fully formulated. 

 

Yes you can't simulate the universe in real time because of the reasons given in OP. 

 

But you could still make an simulation of the universe (theoretically). If we set all initial parameters as they are right now, then run a simulation of what will happen in the next 2 minutes you theoretically will get the answer from that simulation. Only caveat you get the answer 3 days after the simulation started. 

 

But real time simulation might not be what is interesting, to a scientist it might be more interesting to understand why something is what it is and it might be enough to simulate a short period of something that has happened in the past. In which case you don't need to simulate in real time you only need to simulate a short period of time (in the simulated universe) and to do this you just let a cluster chug this time period for a year or so. 

 

 

[Wictorian]

3 minutes ago, Spindel said:

Isn't the reasoning in OP flawed?

 

Or at least not fully formulated. 

 

Yes you can't simulate the universe in real time because of the reasons given in OP. 

 

But you could still make an simulation of the universe (theoretically). If we set all initial parameters as they are right now, then run a simulation of what will happen in the next 2 minutes you theoretically will get the answer from that simulation. Only caveat you get the answer 3 days after the simulation started. 

 

But real time simulation might not be what is interesting, to a scientist it might be more interesting to understand why something is what it is and it might be enough to simulate a short period of something that has happened in the past. In which case you don't need to simulate in real time you only need to simulate a short period of time (in the simulated universe) and to do this you just let a cluster chug this time period for a year or so. 

 

 

I am not talking about computing , but rather the data that needs to be stored. Even if you can simulate it in 3 days you need to store that data.

[Moonzy]

I have 10000 mountains in my backyard

 

This information is stored as a few bytes on few transistors, doesn't need 10000 mountains to store that data

[Wictorian]

1 minute ago, Moonzy said:

I have 10000 mountains in my backyard

 

This information is stored as a few bytes on few transistors, doesn't need 10000 mountains to store that data

yeah but that's not simulating. 

[Moonzy]

4 minutes ago, Wictorian said:

yeah but that's not simulating. 

 

8 minutes ago, Wictorian said:

I am not talking about computing , but rather the data that needs to be stored. Even if you can simulate it in 3 days you need to store that data.

So are you talking computing or storing of the data?

 

The thing about data is that we know that 5 isn't 1 thing, but 5 of 1, that can be used to represent more than 1 object.

[Wictorian]

1 minute ago, Moonzy said:

5 isn't 1 thing, but 5 of 1, that can be used to represent more than 1 object.

I don't really get this.

 

[Moonzy]

3 minutes ago, Wictorian said:

I don't really get this.

Say, we live in a system without maths

'a' is an object

 

I tell you, if I write 'a', it means I have a single 'a'

But if I add an '5' in front, it means I have 'a a a a a', or "five" of 'a' in our understanding

 

So I can use one object, '5', to represent more than one 'a'

 

Basically, compression, in a sense

[Wictorian]

1 minute ago, yaboistar said:

 

 

it also does not take a terabyte of information to emulate a fucking phone

 

 

you can even skip some things out -

That:'s my whole point. It may take a petabyte of information to SIMULATE a fucking phone, or few kilobytes. Depends on detail. I am just talking about simulating a universe as detailed as our universe, which involves simulating the quarks in the atoms and even more.

[straight_stewie]

6 minutes ago, Wictorian said:

That:'s my whole point. It may take a petabyte of information to SIMULATE a fucking phone, or few kilobytes. Depends on detail. I am just talking about simulating a universe as detailed as our universe, which involves simulating the quarks in the atoms and even more.

You can do a complete simulation of a modern cell phone, in near real time (but not quite), of the physical circuits, including those in the integrated circuits, on even a relatively modest computer.

You can even get the tools to do so, for free, from device makers (like Microchip Inc. for example).

You can even simulate/emulate a portion of the hardware, completely accurately including electrical response timing, on a desktop computer while the rest of the circuit is physically built and operating, from a cheap laptop. This is called In Circuit Emulation.

I'm not trying to be to anti-Phil Wheaton here, but your estimations of the amount of data involved in doing things leave a lot to be desired.

And just to add some real science to this, if quantum mechanics is correct, then every possible universe that might exist is nothing but a closed form probability function. In fact, every universe is just different points along the same probability function. And that, is surely something that can be calculated within the universe for which you are doing the calculation.