Random to us with today's level of understanding
and abstraction / model of how the world works...
Researchers at the Japan's Nippon Telegraph and Telephone Corporation (NTT) claim they've invented the first high-speed quantum random number generator built on realistic quantum devices. The RNG is detailed in an announcement [PDF] and a paper in Nature Communications. “In this work, by developing an efficient method for …
Detect an electron and its in one place, detect it again and its in another, yet overall it doesn't move, the net effect of all those 'motions' is zero.
So that system is not random, it's just noisy and complex, but not random. Hence this is not a good random number generator.
And it clearly is moving, there is no difference between 'measurement' and any other interaction the electron is doing. It is always interacting, hence it is also being measured, hence it is always moving.
But then so is the electron creating the electric field you're measuring it with, so electric is an *oscillating* force.
Which means you're measuring the difference and the noise is the complexity of that system.
What, you didn't really think you had particles going backwards in time did you? That would be silly and violate all kinds of causal relationships in physics! No, you're diffing two oscillating things, a particle and the force you're probing it with.
But its OK, you have an oscillating electric field over which light can travel, so now you understand why light appears to be constant velocity in relation to the matter you measure it against.... if the field is twice the size in a direction for the electron, then its twice the size for light too! So compare light's 'velocity' to the scale of mass will always return the same result. Light must be an F1 motion.... any apparent frequency it has is the difference between the oscillation in matter and the oscillation in the light.
And its OK too, you don't need to pretend the electron is forming a quasi particle with a hole in the fabric of a vacuum to give the electron orientation with respect to magnetic.... it has an orientation since its in motion, so no impossible magic holes in space appearing whenever you need to fixup magnetism. Lights polarity suddenly makes sense too...
Tick tock tick tock tick tock, what is the smallest number of oscillations that can form a stable particle in an oscillating field that doesn't go anyway? Two! So what's an electron? It's an F2, So what is time? It's a rhetorical question of course, light motion is per - oscillation, matters return to the same place is a multiple of those oscillations, so time is per oscillation, regardless of the evenness or otherwise of those oscillations....
And since magnetic has a relationship to electric it is also oscillating.
As is every force, all F1s.
And that's OK too, because the magic propagation rate of forces, propagating by the same mechanism that light travels will be the same apparent speed as light travels....!
But I digress.
It's not random.
I think the easiest thing to do here is to review the derivation of the quantum optical master equation (e.g. Louisell, 1973, Quantum Statistical Properties of Radiation) so you can see an example of how probabalistic behaviour can arise in quantum systems. Good luck! :-)
What the actual shuddering fuck are you talking about, man?
To those of us who have studied the physical sciences, you might as well be talking about harmonising your chakras through the use of magic crystals. It sounds like your base chakra might need recharging; if you bend over, I'll get my rubber gloves on and insert one for you.
Does a noisy diode give you a truly random stream of 1s and 0s though?
Statistically, if a bit stream is truly random, there is no way of predicting what the next bit will be, or even how likely it is to be a 1 or a 0.
I suspect a noisy diode may not be truly random (but random enough for many applications). For example, if you are sampling at a known interval, you may be more likely to get a series of the same bit than not (for example 0000 or 1111 may be more likely than 0101 or 1010, whereas with a true random distribution, each of these sequences should be equally likely), or if looking at the time between flipping between 1 and 0, the distribution may not be random, or may change over time, or with temperature, so that it cannot be normalised against a known distribution, which would bias the distribution towards more 1s or more 0s.
For most applications, where you probably just want a bit of randomness, but it's not critical that it's truly random, this isn't important. For cryptographic purposes, it is very important. For instance, if it is known that a source of randomness frequently gives long sequences of the same bit, this could drastically reduce the expected time to brute-force a key, by focusing on keys with repeated bits in first. Ditto for if it is known to give 60% 1s and 40% 0s.
A diode doesn't give you a stream of 1's and 0's: all that happens is that electrons tunnel through the NP energy barrier at quantum random times.
I have no idea how real true random number generator devices turn that into data. But whatever it is, it is quantum random number generator and I wondered how the new approach compares. I presume the diode-based approach gives a much lower bitrate.
I would imagine this then gets turned into 1s and 0s by doing something like measuring how many electrons tunnel in a given time interval, and assigning this a 1 or 0 depending on whether it is above or below a certain threshold, which would be at the middle of the normal distribution.
This is all well and good, but such things vary with temperature, with electrons tunnelling more frequently at higher temperatures.
If you can keep them strictly temperature controlled, you could scale them out for a higher bitrate. Diodes are cheap, and you can stick a hell of a lot of 'em on one chip.
I was curious about that too. Semiconductor junctions are a good source of analog noise as long as they're shielded. Gas discharge tubes can do the same. Run it through a ton of bit spreading to blend away momentary external influences and it should generate a lot of random bits. Even if lightning struck outside, the influence it has would be too random to predict any outcome after the bit spreading.