Take a source of light that is capable of emitting single photons with linear polarization. Place a detector sensitive enough to detect single photons, and aim the laser at it. Insert one polarized sheet between them and rotate until the photons pass though the sheet. Insert another sheet and rotate it 45 degrees to the first sheet. Now each photon will have a 50% chance of passing the pair of sheets, and there is absolutely no way to predict which outcome you will get for any given photon. There is your random outcome. Want something other than a 50/50 outcome? Change the rotation of the sheet second sheet. You can get any probability you want, with the correct angle. 0 degrees is 100%, 90 degrees is 0%, 45 degrees is 50%; (it is the square of the cosine of the angle, in radians)

This post was edited by Azrad on Nov 1 2013 07:05pm

isnt some quantic mechanic involved here ?

how about flipping a coin from orbit to earth, letting it slap the ground, 50% face or 50% number ?

yes in the case of the photon

That would be a very good approximation for an random outcome.

In principle: the outcome of that flip would be a function of the torque put on the coin, the distance the coin fell, the forces that acted on the coin during its fall (like wind), and lots of other complex stuff. If you had all these numbers, you could predict the outcome of a flip.

In reality: getting these numbers accurately enough would be virtually impossible, essentially making the outcome essentially random. It may not even be possible to get those number accurately enough even in principle, so it is not clear if it would be perfectly random or not.


The key to the experiment I mentioned above is there is no function to predict the outcome of any given photon. So we don't have to worry about all that messy stuff like we do with the coin (where we can't be sure if the flip was truly random or not).

This post was edited by Azrad on Nov 1 2013 09:29pm

Its ALL quantum mechanics.

Another easy experiment is to shine a laser (must be collimated. e.g. light whose rays are parallel, and therefore will spread minimally as it propagates. <- taken from google) through a small slit, compared to the wavelength of the light. This experiment is a direct confirmation of the Heisenberg Principle, which states that it is impossible to know the momentum and position of a particle within a certain time frame.

Quantum mechanics is really cool, but REALLY weird. Like Azrad said, the coin flipping is a good approximation, but the light examples is truly random in nature.

yep Azrad & khemist i know the coin is a suit of events that cant end into a "pure randomness" even if it's far far complex (enough for humans needs usually)

i think we got some kind of issue with quantic mechanic too... okay quantic mechanic is a tool that include the probability/uncertain in itself so it seems clean but

a photon is still something existing isnt it? even if it's a particle that acting like a wave (sorry for my english), this is still something with a following a defined "path" in space and time
btw with this quantic method it will be like a coin with half face and half number in same time or something like this isnt it ?


i guess pure randomness doesnt really exist because it would be the result of something that cant be defined at all hmm

It seems totally reasonable that it has a position at every point in time, and that with a collection of these positions you can build up a trajectory (path), but this does not work. For objects like coins, balls and planets this idea works pretty darn well. For massive objects (coins, balls, etc) these failures are hidden by our ability to preform accurate enough measurements to expose them. Lets look at a bowling ball. Lets assume you know the position of a bowling ball to +/- 0.01 meters, and it has a mass of 7.3 kilograms:



So while you thought your bowling ball was moving at 10 km/h it was moving at some random speed between 10.000000000000000000000000000000000032 km/h and 9.999999999999999999999999999999999968 km/h and you just never noticed the difference.

If you notice the mass of the object is in the denominator on the right hand side, so if you decrease the mass, the unknown part of the speed increases, making the above effect more and more likely to be noticed. By the time you reach very light objects (at the atomic scale), this discrepancy can no longer be safely ignored. Now your very reasonable idea of a definite path will blow up in your face.

Long story short:

In quantum mechanics measuring the system (or even observing it) alters the outcome. In the double (or even single) slit experiment, when a particle (can be a photon, electron, whatever) is sent through the slit, we now know the delta_x (or the range of values where the particle can be) of the system. In other words, when the partcile passe through the slit, at that exact point in time we have very very small delta_x, which leads to a large value for delta_p, which is the uncertainty in momentum. With that large uncertainty, the particle goes off at funny angles and the familiar pattern is seen.

Pure randomness does exist. But computers cannot generate truly random numbers. Refer to http://en.wikipedia.org/wiki/Uncertainty_principle

This post was edited by khemist on Nov 2 2013 11:45am

y i know about this uncertainty in quantic physic, btw what i said about the photon following a "path" was pretty exploited as a fail this is why i quoted it, i knew you were going on that.

it's only a tool to work with, i guess i could wrote this: $PurRandomNumber = PRndFunc(10,100)

now i wrote it, but i dont have the value...

edit omg:

Can a computer generate numbers that are truly "random"?
Yes

26.37% (24 votes)
No

73.63% (67 votes)


This post was edited by Saucisson6000 on Nov 2 2013 01:58pm

The uncertainty principle isn't necessarily a statement about randomness though. It is more a statement about the maximal degree of localization that is possible for two Fourier-conjugate variables or, more generally, two variables obeying a canonical commutation relation. There are other interpretations aside from randomness that can give the same effect. For example, one could view the wave function as being a density of states rather than a probability density--we can think of an electron as simultaneously being in all possible states, where "being" is no longer a binary idea. Your point is taken though that even given the entire current state of a quantum system, we cannot predict with certainty the measurements of the system.

The question of what randomness means really is a profound one--and it is one that we sweep under the rug in the formalism of probability. I don't know of anyone who has a satisfying answer. There is some good work being done on computable randomness. The most intuitive definition I have seen there is that a sequence of numbers is random if no finite time finite state computer can predict the output given the previous outputs. I'm not sure if anyone is anywhere close to that though.

edit: Forgot what words mean.

This post was edited by darkfire on Nov 2 2013 02:31pm

If your looking for a random series of numbers and it's always a number that comes up it is in fact not random.
The only way it could possible be random is if you also got stuff like cats and dogs and blue houses while you looked for just numbers.
Even that isn't random enough imo.