Hello guys, is anyone here familiar with quantum computing?
Just wait 10 years and I might have a PhD
Key word: might
Key word: might
Sounds interesting, tell me more! Or... don't. I don't wanna go down another rabbit hole. But I also want to...
Is this the superimposition of free will?
Is this the superimposition of free will?
Yes I do.No I don't.somewhere in between.
@Dukedog said in #4:
> Yes I do.No I don't.somewhere in between.
Real quantum knowledge is: both at the same time, no?
> Yes I do.No I don't.somewhere in between.
Real quantum knowledge is: both at the same time, no?
I both do so and not do so!
@MrPushwood said in #6:
> I both do so and not do so!
Seems like you understand the concept of superposition.
> I both do so and not do so!
Seems like you understand the concept of superposition.
I didn't study quantum computing or quantum mechanics so what I am saying here only represents my understanding of things and is to be taken with a big grain of salt. In particular if someone better qualified than me (for instance @Thalassokrator) spots any mistakes, please correct me.
At the quantum scales, particles have the possibility to "interact with copies of themselves", so to speak. This is evidenced by the double slit experiment. There is a source of light which emits one photon at a time, a screen, and in between there is a wall with two slits A and B that can either be opened or closed, at our convenience. To reach the screen from the soutce of light, the photons must go through the slits. If we send one photon at a time in the box when only slit A is open, we observe a certain probability distribution of the photon impact on the screen P(A). When only slit B is open, we observe a certain probability distribution P(B). If each individual photon had no possibility to 'interact with itself', then when both slits are open, the probability distribution should be given by P(A)+P(B). But that is not what we observe, instead we observe interference figures. So in a way each photon 'knows' that both slits are open, contradicting the classical intuition that one photon can only go through one slits at a time.
There are different interpretations of this phenomenon, either ontic (describing what really happens in the box) or epistemic (describing our knowledge of what happens in the box). Among the ontic interpretation, we have Everett's many world interpretation: the photon takes every possible paths in 'superimposed' worlds, and these worlds can 'communicate' between each other. Another ontic interpretation is Bohm-De Broglie pilot wave theory, where, on the opposite, the photon has a single well-defined position at any time, but it is "guided" by the pilot wave, and the pilot wave propagates through the wholes space. So in this interpretation, although the individual photon goes through one slit at a time, the pilot wave is 'going through' both slits and then interfers with itself. And the photon is guided by the pilot wave which 'knows' that both slits are open.
The epistemic interpretations all roughly say "something is going on in the box, we don't know exactly what (but we know it can't be described with local hidden variables), and the wave function gives us the subjective probability of measuring a certain result".
But regardless of the interpretation, it's this capacity of interecting with some 'copy' of yourself which is crucial. Why? When computing, we sometimes need to calculate what happens in different cases. For instance, maybe after I play some forcing move, my opponent has two possible moves and I need to calculate the outcome of each one. In this kind of situation, with a classical computer, you can either calculate one case after another (and then the total time it will take will be the sum of how long each individual case is taking to compute), or you can divide the cases between different computers (or cores, or processors), but then (assuming every case takes roughly the same time to compute), you will need as many computers / cores/ processors as there are cases. If you have 2^n cases that take a time T to calculate, you can either do it in time (2^n)*T with one machine, or in time T if you have 2^n machines, or some compromise in between if you have more than one machine but less than 2^n.
Now with a quantum computer, because each quantum bit can be in two different superimposed state and interact with itself, whenever you have two possible cases, the same quantum bit can handle them in two 'parallel worlds' (if you stick to Everett's interpretation). So now you only need n qubits to treat 2^n cases in time T, which is exponentially better!
Or at least that's so in theory, because as far as I know we still don't really know how to write algorithms for quantum computers.
At the quantum scales, particles have the possibility to "interact with copies of themselves", so to speak. This is evidenced by the double slit experiment. There is a source of light which emits one photon at a time, a screen, and in between there is a wall with two slits A and B that can either be opened or closed, at our convenience. To reach the screen from the soutce of light, the photons must go through the slits. If we send one photon at a time in the box when only slit A is open, we observe a certain probability distribution of the photon impact on the screen P(A). When only slit B is open, we observe a certain probability distribution P(B). If each individual photon had no possibility to 'interact with itself', then when both slits are open, the probability distribution should be given by P(A)+P(B). But that is not what we observe, instead we observe interference figures. So in a way each photon 'knows' that both slits are open, contradicting the classical intuition that one photon can only go through one slits at a time.
There are different interpretations of this phenomenon, either ontic (describing what really happens in the box) or epistemic (describing our knowledge of what happens in the box). Among the ontic interpretation, we have Everett's many world interpretation: the photon takes every possible paths in 'superimposed' worlds, and these worlds can 'communicate' between each other. Another ontic interpretation is Bohm-De Broglie pilot wave theory, where, on the opposite, the photon has a single well-defined position at any time, but it is "guided" by the pilot wave, and the pilot wave propagates through the wholes space. So in this interpretation, although the individual photon goes through one slit at a time, the pilot wave is 'going through' both slits and then interfers with itself. And the photon is guided by the pilot wave which 'knows' that both slits are open.
The epistemic interpretations all roughly say "something is going on in the box, we don't know exactly what (but we know it can't be described with local hidden variables), and the wave function gives us the subjective probability of measuring a certain result".
But regardless of the interpretation, it's this capacity of interecting with some 'copy' of yourself which is crucial. Why? When computing, we sometimes need to calculate what happens in different cases. For instance, maybe after I play some forcing move, my opponent has two possible moves and I need to calculate the outcome of each one. In this kind of situation, with a classical computer, you can either calculate one case after another (and then the total time it will take will be the sum of how long each individual case is taking to compute), or you can divide the cases between different computers (or cores, or processors), but then (assuming every case takes roughly the same time to compute), you will need as many computers / cores/ processors as there are cases. If you have 2^n cases that take a time T to calculate, you can either do it in time (2^n)*T with one machine, or in time T if you have 2^n machines, or some compromise in between if you have more than one machine but less than 2^n.
Now with a quantum computer, because each quantum bit can be in two different superimposed state and interact with itself, whenever you have two possible cases, the same quantum bit can handle them in two 'parallel worlds' (if you stick to Everett's interpretation). So now you only need n qubits to treat 2^n cases in time T, which is exponentially better!
Or at least that's so in theory, because as far as I know we still don't really know how to write algorithms for quantum computers.
@Fmuror02
These videos are quite old (they are from 2013), but they give a brilliant introduction to quantum computing. In part 2, the interviewed physicist actually shows measurement results (oscilloscope readings) of the spin of an individual atomic nucleus in real time! You can see the flipping of an individual phosphorus nucleus while it's happening! That's really exciting in my opinion :-)
Part 1:
www.youtube.com/watch?v=g_IaVepNDT4
Part 2:
www.youtube.com/watch?v=zNzzGgr2mhk
These videos are quite old (they are from 2013), but they give a brilliant introduction to quantum computing. In part 2, the interviewed physicist actually shows measurement results (oscilloscope readings) of the spin of an individual atomic nucleus in real time! You can see the flipping of an individual phosphorus nucleus while it's happening! That's really exciting in my opinion :-)
Part 1:
www.youtube.com/watch?v=g_IaVepNDT4
Part 2:
www.youtube.com/watch?v=zNzzGgr2mhk
@FC-in-the-UK: You basically described one quantum phenomenon (pretty well, at that) and its implications, the double slit experiment. I will try to expand on that to explain what a real quantum computer does and how it does it.
Consider the following: you have a bicycle lock. It opens if you move it into the correct position. The positions are numbered 000-999, so there are 1000 possible combinations. Now suppose you forgot the combination. You still have a way of opening it: brute force. You try the first combination, say 0-0-0, try to open it, then proceed to the next combination, 0-0-1, try again, and so forth. It is easy to see that this will work but will take a lot of time.
Now consider the following: you could, somehow, try several positions at once. Say, you could try 0-0 plus any 0-9 in the last place. Next, you try 0-1 plus any 0-9 in the last place. This would shorten the time it takes to open the lock considerably.
Basically this is how quantum computers work: Say you want to add two numbers: one is fixed, the other is in a certain range: 0, 1, 2, ...., 255. A classic computer would do it this way:
- put the fixed number into a storage place (call that "a")
- put first number of the range in another storage place (call that "b")
- add a + b, giving the result
- put the second number of the range into b
- add a + b, giving the result
- put the third number of the range into b
- add a + b, giving the result
...etc.
A quantum computer would do it like this:
- put first number into a storage place (call that "a")
- put all numbers of the range in another storage place simultaneously (the "superposition")
- add a to the superposition of b, giving another superposition (of all results)
It is obvious that the latter is quite a lot faster (less steps) but also, that the problem itself will have to lend to that sort of problem-solving. Basically a problem has to be able to make use of "parallelisation" to profit from this sort of computing.
A problem like "add a to any number in the range b-c, giving c-b results" can be broken down to a problem "add a to b1", another problem "add a to b2" and so on, and all these can be solved independently. A problem like: take a number a, triple it, calculate the squareroot of the result, then add 9, ..." can only be done step by step and hence will not profit from parallelisation at all.
Consider the following: you have a bicycle lock. It opens if you move it into the correct position. The positions are numbered 000-999, so there are 1000 possible combinations. Now suppose you forgot the combination. You still have a way of opening it: brute force. You try the first combination, say 0-0-0, try to open it, then proceed to the next combination, 0-0-1, try again, and so forth. It is easy to see that this will work but will take a lot of time.
Now consider the following: you could, somehow, try several positions at once. Say, you could try 0-0 plus any 0-9 in the last place. Next, you try 0-1 plus any 0-9 in the last place. This would shorten the time it takes to open the lock considerably.
Basically this is how quantum computers work: Say you want to add two numbers: one is fixed, the other is in a certain range: 0, 1, 2, ...., 255. A classic computer would do it this way:
- put the fixed number into a storage place (call that "a")
- put first number of the range in another storage place (call that "b")
- add a + b, giving the result
- put the second number of the range into b
- add a + b, giving the result
- put the third number of the range into b
- add a + b, giving the result
...etc.
A quantum computer would do it like this:
- put first number into a storage place (call that "a")
- put all numbers of the range in another storage place simultaneously (the "superposition")
- add a to the superposition of b, giving another superposition (of all results)
It is obvious that the latter is quite a lot faster (less steps) but also, that the problem itself will have to lend to that sort of problem-solving. Basically a problem has to be able to make use of "parallelisation" to profit from this sort of computing.
A problem like "add a to any number in the range b-c, giving c-b results" can be broken down to a problem "add a to b1", another problem "add a to b2" and so on, and all these can be solved independently. A problem like: take a number a, triple it, calculate the squareroot of the result, then add 9, ..." can only be done step by step and hence will not profit from parallelisation at all.
This topic has been archived and can no longer be replied to.