[skinwill]

Can someone explain to me how this is different than a simple noise generator based on a PN junction? As in, isn't this just amplifying noise and aren't there less sensational ways of doing nearly the same thing? Does measuring a photon with this method actually get you better randomness? I have some serious gaps in my understanding here and an ELI5 would be neat.

[bob1029]

Measuring photons in this manner gives you the best randomness. It is effectively a quantum technique. A PN junction is (mostly) classical.

The specific mechanism is mentioned in the article:

https://en.wikipedia.org/wiki/Spontaneous_emission

> Although there is only one electronic transition from the excited state to ground state, there are many ways in which the electromagnetic field may go from the ground state to a one-photon state. That is, the electromagnetic field has infinitely more degrees of freedom, corresponding to the different directions in which the photon can be emitted. Equivalently, one might say that the phase space offered by the electromagnetic field is infinitely larger than that offered by the atom. This infinite degree of freedom for the emission of the photon results in the apparent irreversible decay, i.e., spontaneous emission.

[dragontamer]

I've been told that reverse shot noise from a PN junction is quantum in nature.

It is possible for an electron to spontaneously gather enough voltage to break through a PN junction backwards. This shows up as a very noisy current measured in microamps.

--------

A forward bias PN junction might not be quantomly random. I'll have to research more. But a reverse bias PN junction is almost certainly quantum in nature.

---------

IMO, this is all just PN junction noise. Maybe LEDs are better than Zener diodes for noise. I'm pretty sure that noise characteristics are a guess and check methodology, it's all PN junctions of slightly different shapes after all.

[cubefox]

The question is whether quantum mechanical noise could have a conceivable advantage over classical noise. I strongly suspect: no. Classical noise is already factually unpredictable, so the theoretical unpredictability (assuming no hidden variable theories I guess) of quantum noise doesn't add anything.

[bob1029]

Classical noise is only unpredictable if you are lacking the necessary physical information to make an accurate prediction. Otherwise, it is always predictable.

Quantum noise is not based in any kind of physical information in the same way. It is intrinsically random. The "randomness" isn't merely a side effect of a bunch of physical phenomena. You cannot compromise a QRNG even if you had perfect knowledge of the state of every particle in the system over time.

https://www.jpmorgan.com/technology/technology-blog/certifie...

[cubefox]

> Classical noise is only unpredictable if you are lacking the necessary physical information to make an accurate prediction. Otherwise, it is always predictable.

Since you are always lacking that necessary physical information, it is always unpredictable. If it were otherwise, we would already know whether hidden variable theories of quantum mechanics (which lack intrinsic randomness) are correct or incorrect. But we don't know that. So intrinsic randomness doesn't make a difference to us. So quantum noise is useless.

[charcircuit]

>Quantum noise is not based in any kind of physical information

This sounds like more of a limitation of the model you are using than a limitation of reality.

[amelius]

I suspect that "better randomness" is not what this solves, but rather faster randomness.

A PN junction gives you only megabits/s of randomness at most.

This proposed method, if the article is correct, reaches gigabits/s.

But it could be because they are just using a large array.

[rainsford]

I'm sure I'm overlooking something, but what's the real use-case for true random number generation at that fast of a rate? Even a few Kb/s of random numbers is enough to continually reseed a cryptographic pseudo-random number generator that will generate as much output as you want that's indistinguishable from true randomness. I suppose you aren't reliant on the security of the underlying cryptographic primitives then, but you're still reliant on the particular hardware RNG chip being implemented in a way that's free of bias even if the underlying physics principle is sound.

[amelius]

One thing is for sure, nature has no shortage of randomness. So indeed it seems difficult to find advantages of a new method.

In any case, a PRNG might be a no-go for many applications, out of principle.

And also, maybe a PRNG requires more power and die area than this new method?