A little bit about the hardware behind the MED generators

I created this thread as a starting place to have info on - and discuss - the hardware behind the MED devices, the type of quantum phenomena they’re based on etc. I can imagine one day this thread might be the starting point for newcomers with an interest in the hardware (as I am myself knowing very little about the hardware side of things at the moment).

(The idea to create the thread came as I was writing the generator.cpp class for the MeterFeeder library and I wanted a simple, succinct yet detailed one-liner comment to describe the kind of devices this class was intended to encapsulate.)

I would be happy to answer any specific questions about MMI generator hardware (if I can). I am very familiar with the theoretical basis and practical design of these generators. My desire is to pass on my considerable knowledge developed over the past 27 years – I don’t consider any question silly or not worthy of consideration.

I have built every conceivable type of generator hardware using entropy sources from single photon to photon beam shot noise; thermal noise in resistors and CMOS transistors; nuclear decay timing, and semiconductor junction shot noise and avalanche noise.

As an example, Zener diodes (or reverse bias emitter-base junctions in transistors) with Zener voltages above about 7 volts produce primarily avalanche noise. Fundamental modeling indicates this type of noise is one of the least responsive to MMI influence because the rms amplitude of the noise signal is so large. At the opposite end, thermal noise in resistors can be the basis of one of the more responsive detectors of MMI influence.

In general, ranking of commonly used entropy sources from worst to best for MMI use is: Zener avalanche noise, Zener shot noise and thermal noise. Quantum sources producing continuous noise signals provide no advantage over more classical sources.

Thank you. I’d like to start off by getting a better understanding of the concept of noise.

Over the past year I’ve heard terms like thermal, shot, Johnson, Nyquist and just having heard about it for the first time in your last post “avalanche” noise. I’ve read up on some stuff but I come from a background with just a very little understanding of the physics and electronic theory behind those papers I’ve read. My overall understanding is less complete than I know it needs to be.

I think what would help me and others in the future is having an easy to understand definition of the different types of noise, the various types of electronic methods for measuring those types of noise and their place in measuring MMI (some of which I appreciate you already started to answer in your last post).

In particular I’m interested in what true entropy sources we can utilize for future projects with a focus on MMI applications. A lot of the recent discussions have been around your MED generators but I’m also keen to get more of an understanding about things like camera RNGs and PCRNG. I’ve spent many a time implementing applications of the amazing work of Andy and Theo but now I’m at the stage I want to understand more about the underlying concepts and theories.

A Brief Explanation of Electronic Noise

The first point is to distinguish nondeterministic or true random numbers from pseudorandom numbers. Every true random generator uses a physical entropy source to provide its inherent unpredictability. Every computing device, except a quantum computer, is a finite state machine . That means every program it runs will only take on a limited number of states. The result of this is no program can be written to produce true random numbers and pseudorandom numbers can never be inherently unpredictable.

Electronic noise is the usual source of unpredictability in random generators for mind matter interaction (MMI) measurements. Mechanical sources of entropy are not suited to modern, high speed, high quality generators so they will not be considered. Noise sources are sometimes called entropy sources, though in this context they are essentially the same thing. That is, a physical component that produces a measurable output that is inherently unpredictable.

Entropy sources commonly used in MMI generators, sometimes called random event generators (REGs), are of only a few types. These are:

  • Thermal noise, also called Johnson, Nyquist or Johnson-Nyquist noise, is the electronic noise generated by the thermal agitation of the charge carriers – usually electrons – in an electrical conductor. The amplitude of the noise is proportional to the square root of the resistance across which it is measured and is present regardless of any applied voltage.
  • Shot noise, sometimes called Poisson noise because of its statistical distribution, arises because current consists of a very large number of discrete charges, making its flow grainy. The flow of these discrete charges gives rise to essentially white noise. That is, noise that has equal amplitude over a very wide range of frequency. Unlike thermal noise, shot noise is proportional to the square root of the average current flowing and is not dependent on the temperature of the circuit. Shot noise is more apparent in devices such as reverse biased semiconductor junctions, such as Zener diodes or emitter-base junctions of transistors with a breakdown voltage below about 6 volts. Shot noise also appears in optical detection since light is a stream of photons that is grainy in the same way as electrons.
  • Avalanche noise occurs in reverse-biased junctions such as Zener diodes when the breakdown voltage is above about 7 volts. The high electric field in the junction accelerates the charges until collisions cause a chain reaction, generating additional charge carriers. This way the current is multiplied in pulses. This process is random and the intensity of avalanche noise, which is not predictable from basic principles like thermal and shot noise, is usually much larger than any other noise component.
  • Jitter noise is not a noise type, but is a jitter in the timing of rising and falling edges in oscillating signals. Rising and falling edges do not happen instantaneously so other noise sources can add during the rise or fall to cause the edges to be moved earlier or later in time. The distribution of timing jitter is the same as the noise that caused it, so it is inherently random. Jitter noise is sampled by latching one oscillator, usually a ring oscillator, by another sampling signal or clock that may also be produced by a ring oscillator.
  • Pure quantum systems, such as a photon passing through a beam splitter, are inherently random as described by quantum mechanics. A single photon detector in each of the two possible outputs from the beam splitter detects the photon. Each detector output is assigned to represent a 1 or a 0, which is the random output of the detection. Photonic generators are complex and hard to build and have biases that are difficult to make very small. However, unlike shot noise that is considered quantum mechanical, pure quantum systems may provide superior performance when used as MMI detectors. This possibility has not been proven experimentally.
  • Nuclear decay is caused by an unstable nucleus or radioactive material. Timing of nuclear decay is determined by quantum mechanics and is considered inherently random within its expected statistical distribution. However, it is not as simple as a pure quantum system so is harder to make into a random output for MMI detection. One type of generator measures the time between decay events with a high-speed counter and produces a 1 or 0 depending on whether the count is even or odd. Another may compare consecutive decay timings and assign a 1 or 0 if the first is shorter or longer than the second. Finally, a generator may find the average timing between decay events using an integrator and analog sampler and produce a 1 or 0 depending on whether a decay occurs sooner or later than average.

The quantum systems, the last two above, are not commonly found in MMI generators since they require more specialized technical knowledge and equipment. Jitter noise is used extensively in mind enabled devices such as MED100Kxx. Jitter noise is also the basis of the PCQNG, a software-enabled true random number generator that uses components in a PC’s CPU.

Thank you, that was the perfect list up I was looking for.

Are there any portable/small beam splitters that you’ve come across before?

I have used a couple of different types of beam splitters for photonic RNG construction. The first was a 1cm cube that was a polarization beam splitter. The two output ports provided vertical and horizontal polarized photons. Polarized photons were supplied to the input port with the polarization rotated 45 degrees so there was a 50/50 chance the photons would exit either output port. The second was a 5mm cube with a half-reflecting beam splitter. Photons had a 50/50 chance (though not quite since there was some loss) of exiting either output port. The first type produces a very pure quantum state, but is more complex and expensive. I don’t remember where I got these two, but they were from different optics suppliers. You might try Edmund Optics. Their cube splitters are a little expensive, but their 50/50 plate splitters are not too bad. Amazon has a couple for less, but not as small. They also have a 50/50 beam splitter plate for about $15. It’s 30mm square (1.1mm thick), but could possibly be cut into smaller pieces with the proper tools if you need a really small splitter.

I’m a bit confused here, you mention that empirically you see no advantage of quantum-mechanical based sources over classical but you believe pure quantum systems to be superior, why? Did you get some promising results with polarizers/beam splitters?

Since the true nature of MMI phenomena is not clear to me, I can consider two possible mechanisms: micropsyhokinesis, i.e. flipping bits by force of will and that would require at least an amount of energy more that Landauer’s limit, approximately 10^–21 J
Thermal noise power density at room temperature gives us about the same, but 4 times higher estimation 293K * 1.380649×10−23 J/K ~= 4 * 10^–21 J
If it is the mechanism, storage devices approaching Landauer’s limit will become psychoresponsive just by the magnitude of power they need to flip a bit.

The other possible explanation is by influencing the probability itself (let’s assume that probability wave function is an intrinsic property of reality here), this explains why pure quantum would be most susceptible to MMI, but it also assumes that ‘classical’ Johnson–Nyquist noise should be more-or-less MMI-proof which is admittedly not the case. I might also be missing a possible mechanism of quantum-unfuencess-classic here by stochastic resonance or other means. What is your opinion on this?

PS. At some point in the near future Quantum computers will become more or less commercially accessible, any plans on trying Qbit+Hadamard gate setup as a model system to test PSI responsiveness?

I developed a very thorough modeling of responsiveness (effect size, ES) of MMI generators based on the energy it takes to “flip” bits to the intended value. This was used to show the expected ES of most of the REGs and other MMI generators where data was published, plus my extensive database of MMI test results. The model shows it is relatively easy to achieve sub-Landauer Limit MMI entropy sources. However, the common quantum entropy sources (shot noise in Zener diodes, reverse-biased base-emitter transistor junctions or tunneling in small integrated MOS transistors) are no more responsive than thermal noise sources (resistors).

A pure quantum system is one where, for example, two quantum states in superposition can result in only one of two measured results. Those being equated with a “1” or a “0” output. The simple example is a polarized photon being sent into an input port of a polarization beam splitter with the plane of polarization being at 45 degrees relative to the two output ports. The result is, the photon will exit with 50% probability through one of the two output ports. A single photon detector at each of the output ports indicates if a “1” or a “0” resulted from the measurement. The difference in energy between the two possible output states is related to the interference of the photon’s wave function with the quantum vacuum or a virtual photon wave function at the other beam splitter input port. It’s very hard to quantify this energy because the frequency of the output photon is unchanged – or at least has not been reported as being changed – so it is very small. Therefore, I infer from an energetic perspective it takes exceedingly small amounts of energy to mentally “flip” a pure quantum bit of this type of entropy source. Quantum entropy sources are not all equal for MMI use.

If the type of superpositions and measurements made in quantum computers are of the pure quantum type described above, they have the potential to be used as an MMI detector. However, not all quantum computers work the same way and direct access to qubit outputs is generally not available. Just having a quantum computer doesn’t automatically provide an MMI detector, but the technology could be used to build one.

My best idea in case of MED is RAM memory, composed of flip flop triggers and photosensors. So, basically it’s optical matrix, with pixels, which shift their state, when they get photons. I think, it might be best device for MMI research. Because we can precisely choose amounts of random bits, which are generated in given time.Also, we can research spatial patterns in random bits after flash of light.

Yes, many years ago (almost 40) there were a couple of companies that sold imaging systems made from memory chips with either a transparent lid or one with the lid removed. See for example:

One could also use a DRAM and rely on time for the charge to decay enough so many of the bits would appear random when measured (read).

One downside of these approaches is there is a lot of variation in the levels the ram capacitor reaches in a fixed time with or without exposure to light. The result is many of the bits will on average be fixed at a 1 or a 0. Of course the bits could be processed to obtain a smaller number of higher quality random bits. And, it may be possible to make a calibration table to read different bits at different times so many more of them would be indeterminate.

Modern DRAMS have huge capacities and high speed readouts. It would be interesting to see how one of these would perform as a random generator.

It will perform poorly. Because memory cells has bias to 0 or 1 depending on manufacturing imperfections. There was article, which proposed to use SRAM:

But it has same problem. Biases of memory cells.
That’s why it should be memory chip on JK triggers. It should flip state when it was hit by photon. I think it will be random enough.