Researchers Use Tiny Magnetic Swirls To Generate True Random Numbers (phys.org) 72
A group of Brown University physicists has developed a technique that can potentially generate millions of random digits per second by harnessing the behavior of skyrmions -- tiny magnetic anomalies that arise in certain two-dimensional materials. Phys.Org reports: Their research, published in Nature Communications, reveals previously unexplored dynamics of single skyrmions, the researchers say. Discovered around a half-decade ago, skyrmions have sparked interest in physics as a path toward next-generation computing devices that take advantage of the magnetic properties of particles -- a field known as spintronics. [...] Skyrmions arise from the "spin" of electrons in ultra-thin materials. Spin can be thought of as the tiny magnetic moment of each electron, which points up, down or somewhere in between. Some two-dimensional materials, in their lowest energy states, have a property called perpendicular magnetic anisotropy -- meaning the spins of electrons all point in a direction perpendicular to the film. When these materials are excited with electricity or a magnetic field, some of the electron spins flip as the energy of the system rises. When that happens, the spins of surrounding electrons are perturbed to some extent, forming a magnetic whirlpool surrounding the flipped electron -- a skyrmion.
Skyrmions, which are generally about 1 micrometer (a millionth of a meter) or smaller in diameter, behave a bit like a kind of particle, zipping across the material from side to side. And once they're formed, they're very difficult to get rid of. Because they're so robust, researchers are interested in using their movement to perform computations and to store data. This new study shows that in addition to the global movement of skyrmions across a material, the local behavior of individual skyrmions can also be useful. For the study, which was led by Brown postdoctoral fellow Kang Wang, the researchers fabricated magnetic thin films using a technique that produced subtle defects in the material's atomic lattice. When skyrmions form in the material, these defects, which the researchers call pinning centers, hold the skyrmions firmly in place rather than allowing them to move as they normally would.
The researchers found that when a skyrmion is held in place, they fluctuate randomly in size. With one section of the skyrmion held tightly to one pinning center, the rest of the skyrmion jumps back and forth, wrapping around two nearby pinning centers, one closer and one farther away. The change in skyrmion size is measured through what's known as the anomalous Hall effect, which is a voltage that propagates across the material. This voltage is sensitive to the perpendicular component of electron spins. When the skyrmion size changes, the voltage changes to an extent that is easily measured. Those random voltage changes can be used to produce a string of random digits. The researchers estimate that by optimizing the defect-spacing in their device, they can produce as many as 10 million random digits per second, providing a new and highly efficient method of producing true random numbers.
Skyrmions, which are generally about 1 micrometer (a millionth of a meter) or smaller in diameter, behave a bit like a kind of particle, zipping across the material from side to side. And once they're formed, they're very difficult to get rid of. Because they're so robust, researchers are interested in using their movement to perform computations and to store data. This new study shows that in addition to the global movement of skyrmions across a material, the local behavior of individual skyrmions can also be useful. For the study, which was led by Brown postdoctoral fellow Kang Wang, the researchers fabricated magnetic thin films using a technique that produced subtle defects in the material's atomic lattice. When skyrmions form in the material, these defects, which the researchers call pinning centers, hold the skyrmions firmly in place rather than allowing them to move as they normally would.
The researchers found that when a skyrmion is held in place, they fluctuate randomly in size. With one section of the skyrmion held tightly to one pinning center, the rest of the skyrmion jumps back and forth, wrapping around two nearby pinning centers, one closer and one farther away. The change in skyrmion size is measured through what's known as the anomalous Hall effect, which is a voltage that propagates across the material. This voltage is sensitive to the perpendicular component of electron spins. When the skyrmion size changes, the voltage changes to an extent that is easily measured. Those random voltage changes can be used to produce a string of random digits. The researchers estimate that by optimizing the defect-spacing in their device, they can produce as many as 10 million random digits per second, providing a new and highly efficient method of producing true random numbers.
Oh, yes (Score:2)
I played Skyrim myself, such a nice series of games.
Oh, wait...
Re:Oh, yes (Score:4, Funny)
Re: (Score:2)
Re: (Score:2)
Nobody should bother with the parent's link. That foundation is more or less selling snake oil. I'm surprised they haven't yet induced their followers to believe in aliens that live on an asteroid just before getting them to drink the tainted Kool-Aid once their checks have cleared.
Sounds expensive (Score:5, Funny)
This sounds complicated/expensive.
Back in the 90s we used to point a video camera at a table full of lava lamps. It was way easier and very awesome.
https://en.wikipedia.org/wiki/... [wikipedia.org]
Re: (Score:2)
This sounds complicated/expensive.
Back in the 90s we used to point a video camera at a table full of lava lamps. It was way easier and very awesome.
https://en.wikipedia.org/wiki/... [wikipedia.org]
While Lavarand is considered a "true" random generator due to the multi-step process to generate the final number, I thought some of the (unjust) criticism back in the day, was that part of the process, was actually pseudo-random.
Perhaps this solution, is more for the "rando-purists"?
Re:Sounds expensive (Score:5, Interesting)
I actually built a true RNG from a microcontroller by sampling thermal noise. It produces about 8 million random digits per second, with the limit being the USB bus.
So my guess is that this is just an interesting property of skyrmions, not the goal of their research.
Re: (Score:3)
Re:Sounds expensive (Score:4, Informative)
I did experiment with PN junctions for random noise. They are not ideal because you need a somewhat high voltage (10V at least) that is not available from USB, so you have to generate it. The generator itself can introduce non-random noise.
The other issue is ageing and drift over time. The random voltages are analogue, so you either need a fast ADC or a comparator to convert them to digital. I came up with a scheme that uses a low pass filter to set the comparison voltage automatically, but you are still limited in terms of the bandwidth, i.e. the number of bits per second.
That's why I ended up using the LSB of an ADC measuring an on-die PTC thermistor. At those tiny voltage levels the fluctuations are quantum noise. Being all on-chip it is less prone to outside interference, and the on-chip voltage reference effectively decouples it from the supply.
Re:Sounds expensive (Score:4, Informative)
Yeah, if you have a rigorous requirement for true random numbers then an ad hoc repurposing scheme with existing components is not going to serve the need -- unless you did a rigorous validation of your set-up you could not be certain that there is a non-random component defeating the purpose due to secondary effects since this is not what the component was designed/qualified for. Such validation is time consuming, costly, and prone to error if you are doing this as a lone wolf.
Indeed I see that the Wikipedia article on hardware random number generators [wikipedia.org] emphasize that getting genuine randomly distributed bits out is difficult and with multiple failure modes. Also the effective rate tend to be fairly low due the temporal behaviour of the random phenomena.
Re: (Score:2)
Yeah, if you have a rigorous requirement for true random numbers then an ad hoc repurposing scheme with existing components is not going to serve the need -- unless you did a rigorous validation of your set-up you could not be certain that there is a non-random component defeating the purpose due to secondary effects since this is not what the component was designed/qualified for.
Replicate the source circuit two or three times and data slice the difference between two sources to remove common effects. Three sources with three data slicers will allow redundancy to detect failure.
Higher speed operation is possible with shot noise sources however it requires special attention to the source.
Re: (Score:2)
My friend worked on UK telephone test kit, when telephones were run by the Post Office. Microphones and earpieces were tested using a random noise source. The big problem is that random signals fluctuate, and that meant that a unit could pass the test on one day, and fail on another day, due to random fluctuations in the test signal. The solution at the time was to use a record of some random noise in the test kit, so that every test was repeatable. The record was a glass disk, with a waveform engraved on i
Re: (Score:2)
The gold standard for random noise is a zener diode junction
I made one of those. The distribution of amplitudes was so uneven that I could see it on a scope. There was more negative than positive peaks. I found that running the zener diode at low current increased the amplitude of the fluctuations, but that was at the expense of bias.
A better noise source than a zener diode is probably a hot filament run off a DC supply. A useful broadband microwave source is a light bulb in a cut down tin can.
Re: (Score:2)
Can't believe you let the patent expire...
See, that's how we can have nice things after 20 years!
- Yo Grark
true randomness is only a philosophical contruct (Score:1, Insightful)
Re: (Score:2, Insightful)
Re:true randomness is only a philosophical contruc (Score:4, Informative)
True randomness is more of a philosophical construct then anything else. Given input that is highly erratic and unpredictable the outcome may be unpredictable and in practice qualify as random. This will only hold until, I will duck pre-emptively, the input factors are understood and become predictable. That is very unlikely to happen anytime soon, possibly never. But for the outcome to be provably random, we need to prove that the input is truly random. (Thus painting a huge bullseye on my back, I know).
If by random we mean âoewithout any external force/influence acting upon itâ then Newtonâ(TM)s first law comes into play: if a body is at rest or moving at a constant speed in a straight line, it will remain at rest or keep moving in a straight line at constant speed unless it is acted upon by a force
Pseudo randomness in practice may be possible because of our (current limitations). But True randomness is impossible.
So, what you're saying is that you can build a Heisenberg Compensator?
True randomness is not only believed to be possible but has been the foundation of most of physics for more than a century. If you can prove your claim there's a big fat cheque from the Nobel Committee waiting for you.
Re: (Score:2)
Re: (Score:2)
then our understanding should be wrong. There cannot be true randomness.
Why's that, then? Because you don't like it?
Can you explain the mechanism of it?
It's random, duh.
How does that happen? The only cases where there is "true" randomness is if we are a) inside a simulation (randomness provided from outside) or b) created by an ominpotent being.
Neither of those make any difference to your argument, actually. If randomness is impossible then it's impossible to gods or alien programmers and if it's possible for them then it's possible for us. You're just pushing the problem up one level without actually addressing it or even defining it.
Think about it: Every particle in the universe affects in all 4 known forces every other particle, no matter how far and no matter how we round statistically our models. So in my view, that is what produces "randomness". The models may appear to be stochastic, but reality itself how can it be?
If you bother to examine the details of QM, the result is, like the speed of light, naturally arrived at from assumptions which are in themselves pretty un
Re: (Score:2)
Both simulation or omnipotent being can explain "randomness". Because if we are in a simulation, we can experience something that to us looks like true randomness (RNG provided by the simulation), when in fact there is not any. To me it seems the o
Re: (Score:2)
Yes, honestly I don't like it. I get it of course that just because I don't like it, doesn't make it false. "it's random duh" does not prove or explain something. True randomness if existed shouldn't "arise" from some basic property and be thoroughly explained?
Well, it arises from some basic features of the uncertainty principle. Bell's Theorem closed off the only serious attempt (by Einstein among others) to find some mechanism behind it which would explain it (and entanglement) as the result of some hidden mechanism which might prove to be non-random. It certainly gets complicated but in the end "true randomness" can't arise completely from a previous cause, can it? Otherwise it would be in some way predictable. If I roll a die (i.e., place two charged particle
Re: (Score:2)
Re: (Score:2)
From what you say, it seems to me, that you define randomness as non-computable due to complexities. Am I right? Because to me, it seems impossible that if we have say 1 electron/proton only, in a universe of its own (no external forces by other particles, and let's say the photon is a fundamental particle, not made of quark - let's just assume that), that it wouldn't have a constant momentum for eternity. In such case I think it wouldn't never do random stuff ( random quantum phenomena wouldn't exist).
Well, it's hard to know what would happen in such an alien situation, but perhaps this page might suggest some ideas about what other options the electron has than just doing nothing :) Then there's the question of spontaneous decay of the electron into energey (photons) and back again - sometimes particles don't seem to need external forces.
As to the photon - my understanding is that you can't have a lone photon in theory, but that's another can of worms that leads to similar questions of free-will and det
Re: (Score:2)
Here's the page I meant to link to:
https://profmattstrassler.com/... [profmattstrassler.com]
Re: (Score:2)
Re: (Score:2)
The mechanism of true randomness is that it's the fundamental construct of the world. It's determinism that is an illusion. According to many quantum mechanics experiments, this holds, and there's some really complicated stuff like Bell's Theorem [wikipedia.org] that proves that it's not just "hidden variables" meaning actual deterministic functionality that we just can't measure sitting behind the randomness. The wikipedia article is pretty dense, but it does refer to the formal proof and it is not simple.
The only cases where there is "true" randomness is if we are a) inside a simulation (randomness provided from outside) or b) created by an ominpotent being.
What are you
Re: (Score:2)
Yes, I know I'm begging the question, I cannot prove it of course, I'm assuming the conclusion as an axiom. And yes, if nature is fundamentally random, there cannot be dete
Re: (Score:2)
Randomness is not part of the foundation of physics at all, what we call randomness is nearly always simply our inability to measure precisely right to the particle level and it is something we probably won't ever be able to do, that doesn't make it random, just unmeasurable.
You need to look at quantum chemistry. The question of how bonds form and reform is not about uncertainty but about randomness. Quantum tunnelling is another example.
I would challenge degree of enthropy aswell (Score:2)
True random isn't only about our understanding, or lack thereof.
In theory no two snowflakes are alike, the same goes for flowers etc. etc. However, that doesn't mean they are totally random. A flower cant all of a sudden be a polar bear. Randomness is within certain limits. And those limits bring bias. The study didn't actually mention that any had been noticed, but that doesn't mean it's not there. Bias is a HUGE thing in random numbers. If a single number is being generated significantly more often than a
Re: (Score:3)
Re: (Score:3)
The claim that "no two snowflakes are alike" is false, except possibly at the level of individual molecules. There are a LOT of different snowflake forms, and they are quite sensitive to the exact conditions for their creation,, but there's a lab at CalTech where they have made identical snowflakes. https://www.messynessychic.com... [messynessychic.com]
Re: (Score:2)
The claim that "no two snowflakes are alike" is false ...
I suppose it is true that one might find two identical snowflakes, but the likelihood is vanishingly small, and of no practical consequence. According to theories of thermodynamics based on random interactions of particles, it is possible that my hot cup of tea might spontaneously develop an ice cube in a vessel of boiling liquid, but I can't think of any useful theory based on that deduction.
Re: (Score:2)
It sounds like "true" randomness isn't impossible.
It sounds more like finding true randomness, may simply not be derived from motion-based inputs.
And to be frank, this seems to be an exercise in futility when one end of this challenge is labeled Not Random, and the other end is labeled Random...In Theory.
All that said, is "true" or "pure" randomness, a solution without a problem? Our almost/close/yes/maybe random generators seem to have been working for...decades now.
Re: (Score:2)
That's not quite right. Quantum randomness is impossible to predict, at least according to our current understanding of physics. The universe is not deterministic at the quantum level, no matter how much information you have.
That's why many true RNGs use quantum effects.
There is also another useful measure of randomness, which is cryptographic security. Basically can a randomly generated key be guessed in some reasonable amount of time. It used to be the case that it was difficult to produce random numbers
Re: (Score:2)
Actually, no. Many physical RNGs use true randomness as a part of the mix because they get it for free. A Z-Diode at 5.6V, for example, outputs about half thermal noise and half quantum (true) noise. So does a reversed transistor BE junction, albeit with a bit less quantum noise because they tend to run at 8...15V from my observations (depends on the manufacturing process and transistor models).
But thermal noise (e.g. from a resistor or OpAmp input) is completely fine for random number generation even for t
Re: (Score:2)
I'll grant "impossible to predict", but the randomness of quantum fluctuations is not certain. There are theories that the apparent randomness is due to reducing the number of dimensions in which the action is happening. I don't think this has been proven correct, but it's a defensible argument.
Re: (Score:2)
What distinction are you drawing between "impossible to predict" and "random"?
Re: (Score:2)
It's impossible to predict because we can't observe the higher dimension in which the action is happening, but if we could observe in sufficient detail, it would be deterministic, and therefore not random, even though since we can't do the observations it appears random to us. (If this is correct, then I believe that sufficient observation in any particular case would allow us to predict that case...but it would be extremely difficult, as even simple "dark sight"* observations are difficult.)
* dark sight:
Re: (Score:3)
random.org works well enough for me.
From that site :
RANDOM.ORG offers true random numbers to anyone on the Internet. The randomness comes from atmospheric noise, which for many purposes is better than the pseudo-random number algorithms typically used in computer programs. People use RANDOM.ORG for holding drawings, lotteries and sweepstakes, to drive online games, for scientific applications and for art and music. The service has existed since 1998 and was built by Dr Mads Haahr of the School of Computer S
Re: (Score:2)
Nope. It simply means that current physical theory says it stays unpredictable when all starting conditions are known. The only thing in nature that can do that are some quantum effects. As we have no clue how they can do that, it is a pretty important open question about the nature of this universe. It is very much _not_ known whether there are additional input factors that could make the the output predictable and currently known Physics says there are not. You are stuck in the classical mechanics here.
Th
Re: (Score:2)
It is a rigorous mathematical concept - and can be examined in the real world and measured against mathematical tests. There are formal criteria for evaluating true randomness, and for setting strict boundaries on potential deviations from randomness.
Re: (Score:2)
>There are formal criteria for evaluating true randomness, and for setting strict boundaries on potential deviations from randomness
Does that imply there is a difference between philosophically pure randomness, which is being suggested doesn't exist, and good enough randomness, randomness that doesn't over deviate beyond strict boundaries? If I'm following right the latter is referred to as true randomness.
On the philosophical side maybe the use of the word true is best avoided. So avoiding the term my u
Cab't Wait... (Score:2)
Can't wait till someone living in the basement knows better then the scientist....
3...2....3...GO
Re: (Score:1)
You can't even count properly. I wouldn't let you into my basement.
Re: (Score:2)
Can't wait till someone living in the basement knows better then the scientist....
3...2....3...GO
You can't even count properly. I wouldn't let you into my basement.
he tried to count properly but his PRNG messed up the sequence.
A million monkeys with a million typewriters (Score:2)
Re: (Score:2)
Reverse bias a diode... (Score:1)
Leakage noise is pretty much random.
Electron/thermal noise is truly random (Score:2)
So while this is I'm sure academically interesting, as far as true random number generation goes, its too complex and way late to the party.
Though it begs the question of why modern computers don't have true thermal random number generators built in instead of relying on variations of the LFSR.
Re: (Score:2)
Yup, they should skip all the research and just produce truly random numbers. I don't know what is holding them back. You seem to know a lot about the area, maybe you could tell them.
Re: (Score:2)
Yup, they should skip all the research and just produce truly random numbers.
That's not difficult; I think the main claim of the article is the rate of generation.
Re: (Score:2)
https://en.wikipedia.org/wiki/... [wikipedia.org]
There's this service called Google, you might want to try it sometime. You might even climb out of your pit of ignorance eventually, who knows?
Re: (Score:2)
Thermal noise is not "true" random. Not that it matters in practice.
Quantum noise is "true" random.
A reversed transistor BE junction creates a mix of both at very low cost, high reliability and high bandwidth.
Predictable pseudo random numbers are useful too (Score:2)
No practical relevance (Score:2)
A reversed PN-junction already creates a mix of thermal noise (sufficient for anything, not "true" random) and quantum noise ("true" random) at a very low price and with a bandwidth of >100MHz, depending on the details. This research nice but has no practical value in random number generation.
Re: (Score:2)
A reversed PN-junction already creates a mix of thermal noise (sufficient for anything, not "true" random) and quantum noise ("true" random) at a very low price and with a bandwidth of >100MHz, depending on the details. This research nice but has no practical value in random number generation.
Do not be misled by quantum woo.
Thermal noise is quantum noise.
Nobody knows what the hell "True Random" means, but some people use the term when they mean 'fully entropic'.
Sensing of quantum noise is not full entropy and never will be. That 50/50 half silvered mirror is not precisely 50/50. That spin detector angle is not perfectly orthogonal to the electron spins being detected. That single photon source is often a two-photon or no-photon source. So you need to do cryptographic entropy extraction (aka con
Re: (Score:2)
A reversed PN-junction already creates a mix of thermal noise (sufficient for anything, not "true" random) and quantum noise ("true" random) at a very low price and with a bandwidth of >100MHz, depending on the details. This research nice but has no practical value in random number generation.
Do not be misled by quantum woo.
Thermal noise is quantum noise.
No, it is not. It is noise from Brownian motion which, according to current models, is deterministic. The quantum noise from a reverse PN breakdown is tunneling through an insulator + avalanche effect that lifts it to something you can measure with regular equipment. Also, no "woo" here. I actually have a pretty reasonable understanding of currently known Physics and its limitations. Any "woo" in this is pure marketing bullshit that I typically do not fall for. Even deterministic noise from Brownian motion
Re: (Score:2)
>No, it is not. It is noise from Brownian motion which, according to current models, is deterministic.
You don't get a pass from quantum physics when you engage in brownian motion.
I don't think Brownian motion is even a good model for the various electrical noise distributions within transistors.
Re: (Score:2)
>No, it is not. It is noise from Brownian motion which, according to current models, is deterministic.
You don't get a pass from quantum physics when you engage in brownian motion.
I don't think Brownian motion is even a good model for the various electrical noise distributions within transistors.
You get to bypass the _randomized_ part of Quantum Physics. Seriously. Get some basics understanding of things before you shoot your mouth off.
Re: (Score:2)
>No, it is not. It is noise from Brownian motion which, according to current models, is deterministic.
You don't get a pass from quantum physics when you engage in brownian motion.
I don't think Brownian motion is even a good model for the various electrical noise distributions within transistors.
You get to bypass the _randomized_ part of Quantum Physics. Seriously. Get some basics understanding of things before you shoot your mouth off.
Which bit did I bypass? I spent my time studying quantum physics necessary to be able to quantify entropy in entropy gathering circuits for purposes of cryptography and waste my life on post quantum cryptography. The apparent randomness of quantum physics (notwithstanding super determinism which posits a deterministic model for QP) is what underlies electrical noise which is generally not Gaussian and generally doesn't follow the distribution of brownian motion (which may or may not look gaussian depending
Re: (Score:2)
>>Nobody knows what the hell "True Random" means,
>Actually, that is very well known in Mathematics:
Actually the term is used differently all over the place and mathematicians are no different. That's why it is poorly defined. The accepted and standardized properties used in well formed papers and standard include things like deterministic vs nondeterministic, entropic, full-entropy, partial entropy and computational prediction bounds.
You appeared to use the term 'True Random' to mean non-determini
Re: (Score:2)
There is no connection between "random" and "full entropy". Seriously. "Random" always comes with a distribution and absolutely nothing says it has to be the Uniform Distribution. Anybody misusing the term in this fashion is just obviously _incompetent_ and should not be listened to.
Also note that no real-world process has or can have "full entropy", which makes the term completely meaningless except in a theoretical discussion. Which I am not going to have here.
Re: (Score:2)
>Also note that no real-world process has or can have "full entropy", which makes the term completely meaningless except in a theoretical discussion. Which I am not going to have here.
A full-entropy process would be in the class of things you could describe as a random process although it ain't gonna happen.
In the sense of 100% full entropy, indeed there is no process that can get you 100% full entropy. However cryptographic standards define 'full entropy' for the purposes of being good enough for crypto
Re: (Score:2)
Consciousness and probably general intelligence is not possible with currently known Physics.
Oh, gweihir, you and your consciousness woo-woo.
Pray tell, exactly how you came to this conclusion?
And don't bother bringing in philosophy into this when
philosophy is nothing but woo-woo.
There is a better method (Score:2)
Millions! (Score:2)
Your desktop PC has circuitry that produces billions of nondeterministic, full entropy random bits per second by exploiting thermal noise.
I hope they're planning to do physics rather that go into production of RNGs.
Channel 3? (Score:2)
I never understood the inability to come up with a true random number generator. Tune your old school TV to channel 3. It's static. Background radiation from the big bang. Use THAT. (?)
Obligatory xkcd (Score:1)