3
$\begingroup$

I've been reading a question regarding the uses of quantum computing. There is a broad theme throughout the answers that seems to suggest that randomness generated via quantum means is somehow better than randomness generated via traditional means. I get the impression from some of those answers that the massive processing potential of quantum computers can somehow derandomise traditionally generated randomness.

A quantum computer would use quantum gates. They would be operating via entanglement and superposition to produce random states. A classical physics based random number generator using say a humble diode, would use chaos theory to produce entropy+. Both constructions would in their gross outputs generate large biases which would be whitened via additional hardware /software. So both techniques converge onto a conventional randomness extractor. Both techniques would also be massively affected by electrical system noise which might be ~50% of the total raw entropy in the quantum case as in this 2 Gb/s beam splitter. Electrical noise is generated mainly by classical physics.

Ultimately, a quantum random number generator outputs a sequence of bits that might be "1" with a 50% probability. I thought that random is random is random. Have I misunderstood?


+Diodes also use quantum effects for noise /entropy generation.

$\endgroup$
5
$\begingroup$

Statistically, yes, random is random (read: entropy). It's also worth stating that you can really only approximate how random something is based on samples. This means that you can't measure randomness explicitly, but you can make approximations based on expected outputs compared to conditional probabilities, etc.

Quantum computing does have random properties which are a fundamental part of quantum mechanics, but a strict requirement of any quantum gate is that they are invertible. This means that for any quantum gate, and any quantum algorithm, you must be able to reverse any transformation (page 18). This is achieved through the use of error-correction, as well as the design of algorithms used in quantum computing.

There is a subtle difference between the randomness associated with quantum computing, and the randomness generated by quantum properties. I've submitted an answer to the question you linked, and the reason I made it clear that you don't need a quantum computer to generate randomness using quantum mechanics is because it's a different use of quantum mechanics.

Quantum randomness can be generated by some cameras already built into several models of smart phones. This is possible through measurement on the few photon level. Essentially you use a camera to measure a few photons, and then translate that into a seed for a PRNG. This works because when light is measured as a photon, the distribution follows the double-slit experiment for individual particles. If you were to measure light as a wave, the distribution is fundamentally different.

The main upshot is that randomness is randomness. For cryptographic purposes, this obviously isn't enough of an answer. We need ways to measure randomness, and so we use tools to approximate how random an output is. This isn't measuring randomness itself, but the approximation helps give information on whether we should classify something as an RNG or PRNG.

So while there's a difference between calling rand in C++, or /dev/urandom in Unix systems, in terms of a PRNG generated classically compared to quantum randomness the difference may potentially be less pronounced. Even if there is a sharp distinction between a classical PRNG and quantum randomness, you're right that for the foreseeable future both would most likely rely on classical randomness extractors.

$\endgroup$
  • $\begingroup$ I should also add that "noisy environments" in quantum computing are a problem. This is why there aren't large-scale, universal quantum computers yet. You could argue D-Wave has solved this by using quantum annealing, but D-Wave is restricted to solving specific problems compared to IBM's "quantum experience" and the soon-to-be available 50 qubit computers from IBM. $\endgroup$ – nonce Apr 23 '17 at 1:37
  • $\begingroup$ can you back up your claim that smart phone cameras can detect single photons? $\endgroup$ – kodlu Apr 23 '17 at 4:14
  • $\begingroup$ I'll go through the answer and add links. Apologies for not doing that initially. $\endgroup$ – nonce Apr 23 '17 at 4:30

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.