WHY QUANTUM COMPUTERS WILL NOT WORK

This can actually be summarized in a simple statement. Any quantum particle is probabilistic. This feature generates quantum noise. A wavefunction of the particle is always in superposition with the random quantum noise. Because of that, the state of that particle can't be determined.

In brief, if a quantum particle exists, it is not possible to know what the particle's state is. In even simpler terms, the universe likes to keep its secrets to itself.

The wavefunction of a quantum particle is described by Eq. (1), where α and n(t) are complex coefficients.

This equation must satisfy Born's rule as shown in Eq. (2).

Eq. 2 is simplified to Eq. (3).

There is no solution to this set of equations.

This relation applies to individual particles, such as electrons and photons, and only to a very small number of particles. It does not apply to a large particle aggregate.

This is not a famous measurement issue devised by Bohr about 100 years ago. It is a step before the measurement.

The wavefunction of a traveling electron always spreads in space. Even if its energy stays constant. So far, by any means, nobody has come up with a method to keep the wave packet from spreading. This limits how long a device can be. In addition, if there is a positive or negative step in energy, an electron is going to be scattered by it. There will be some probability of an electron passing the step, say 0.8, and a probability of, say, 0.2, of a wave packet being turned back. This seriously limits how many gates can be utilized. An identical situation applies to photons traveling over material with a non-uniform reflection coefficient.

How to verify the validity of our statement.

Try a single qubit, any quantum gate, or the whole quantum computer.

1) Place any of those in an initial state. It could be anything.

2) Determine that state using any known method.

3) Repeat steps 1) and 2) a fairly large number of times, perhaps 1000-10000. Collect statistically meaningful data.

4) Plot the distribution of acquired data. The distribution is going to be fairly wide. If noise were not in the picture, an extremely narrow vertical distribution would result. However, this never happens.

Since access to cloud-based quantum computers is available, this exercise can be tried, and the theory confirmed.

Comments on steps 1) and 2).

A quantum particle can't be placed in a reproducible state. This can't be done.

We already know that its state can't be read reproducibly. This is bad news. It means that a quantum computer can't be reset to the same value every time it is turned on. If it starts from different initial conditions, its output will reflect that and contain a fair amount of randomness. This has been verified many times. Repeating the experiment will produce a wide distribution of data.

Another example is a Chinese abacus. If a bead in a lower deck is moved up, an operation 0 + 1 = 1 is performed. If someone very carefully performed that operation 1000 times, this would produce a vertical distribution with a count of 1000 located at 1.

On a quantum computer, this would never work. A quantum computer can't handle integers because of the noise issues and the impossibility of determining a particle's quantum state. This almost-trivial operation, 0 + 1 = 1, can't be performed on a quantum computer.

This is why the well-known Shor's algorithm does not work on a quantum computer. It does not work on a binary computer, either. Well, would you expect something else coming from MIT? On binary computers, the Mersenne prime number has over 40 million digits; on a quantum computer, that number is only 2. Because the quantum computer can't handle integers.

What about error correction? Why can't we just come up with a scheme to remove quantum noise? This will not work. Because of the randomness, something that has occurred in the past can't be reversed.

An additional issue in a quantum gate-based quantum computer is that each gate introduces additional noise. Not only can the noise from the previous gate not be removed, but the next gate will only add more noise. After just a few gate operations, the noise will overwhelm the useful signal, and nothing meaningful will reach the output. This is one of the reasons the number of quantum gates in existing quantum computers is only slightly more than 100. For comparison, a binary chip can already have several hundred billion gates. There is a subtle difference between those two numbers. If a very wide but shallow approach is used, such an architecture would be very limited, since many programs can't be run on it. Synchronizing such quantum gates would also be quite a challenge.

Let's briefly discuss several different approaches used in today's quantum computers.

1) A superconducting qubit approach, probably made most famous by IBM.

It has been around since the mid-eighties. So far, it's essentially a failure. The Josephson junction is very noisy due to defects on both sides of the oxide-superconductor interface. Superconducting qubits generate very low-level signals. The piece of wire between the qubit and an amplifier has a temperature gradient, with several mK at one end and about 4K at the other, which will always generate random current noise. This can't be avoided because, at very low temperatures, dopants in a semiconductor are not ionized, and transistors in the amplifier basically no longer work.

A relatively short coherence time is another detrimental feature. Bad for the computer.

A modular design to increase the computing power - bad idea. Quantum states can't be copied and transferred between modules. Quantum entanglement of particles, electrons in this case, separated by a considerable distance does not exist. There is no known force in nature that could link such particles together.

If the distance between particles is small, there is an electromagnetic interaction between them. This basically means that quantum entanglement is a dubious effect. It is unlikely that gravity between those particles could play any meaningful role, although gravity remains a mystery; it is up to you to decide whether this is feasible. Two probabilistic particles with their wavefunctions in superposition with random noise can't be entangled, because noise in two different places can't be identical. The noise is random.

Every time you've seen a piece of equipment with hundreds of those golden wires and gold-plated connectors, you should know one thing - this can't be fast.

Joe Biden got a picture in front of an IBM quantum computer, and look what happened - cancer of the bone.

We must also add that IBM's corporate culture would always kill pretty much anything, even if those technical challenges are somehow overcome.

2) Google with its superconducting quantum computer, analogous in principle to IBM's. Same approach, same results. Google supercomputers can calculate 100 trillion digits of Pi; their quantum computer can handle the first digit of Pi with 0.1 probability. Apparently their quantum computing program had too many martinis. After that, things went kaput. Google CEO Sundar Pichai still claims that a quantum computer can handle in minutes what a supercomputer would take a trillion years to do. Yes, for sure! Did he remember to turn it on? Another one of those luminaries is Michio Kaku. Did he remember to turn the computer on? Maybe it does not matter!

3) Semiconducting approach based on silicon or III-V semiconductor qubits, probably made most famous by Intel. If it worked, it would enable billions of qubits to be integrated onto a single chip. An almost unimaginable number of connections between those qubits would be possible. An enormous computing power would result. In theory, yes; in practice, no.

An electron or hole traveling in a semiconductor, or inside a quantum well, sooner or later is going to collide with the semiconductor lattice. A single act of collision is enough to destroy coherence. Usually, there are many more collisions. An electron's probability is continuous from a very low energy to an emission energy; there are still virtual states that an electron can occupy. An electron can be at different energy levels. Also, the Heisenberg uncertainty principle prevents us from knowing the electron's exact position. If one deals with an electron spin without knowing its location, then good luck with that idea.

In addition, a semiconductor needs to have some kind of contact, either ohmic or Schottky. Because of different work functions between them, random noise is always generated at the interface, as if this could not get any worse. If there is any temperature gradient across the metal lines, they become additional sources of noise due to the Seebeck effect. If two different metal lines are connected, due to their different work functions and interface defects, a random voltage is also generated. Building a complex chip that uses only a single metal is not feasible in advanced chip manufacturing. Electrons or holes need to be moved between qubits with an electric or magnetic field. This can't be done without injecting noise into qubits. If all of this sounds utterly hopeless, it is. The presented data consistently show wide distributions of the output signals due to all noise contributions.

In addition, Intel spent too much on escort services and too little on R&D, and the company almost went under. It is slightly better right now and seems to be wiggling, but the quantum computing there is dead.

HRL Labs, which was conducting research on silicon spin qubits, was recently acquired by IBM. So, this is going to be dead pretty soon.

4) Trapped-ion-based quantum approach. Perhaps best known from IonQ in Maryland. Well, the founder and co-founder left the company after it became clear things were not going to work. Moving ions around with an external field is a way to inject noise into the system. Residual gas particles in a vacuum system will destroy coherence. Lasers required to "entangle" ions are always noisy; this is bad news. The state of the ions will be read using a high-power, noisy laser and a noisy light detector. This combination guarantees ample noise with no clear solution. So far, that has been the case. As we already mentioned, quantum error correction does not exist because you correct something that carries random noise and that has already occurred. You can't reverse time, and you can't reduce entropy to go back to the pre-noise state. This can't be done.

Then, how is it possible for a company that has so little to offer to be worth billions of dollars on the stock market? Well, it is supported by well-known billionaires, and their money can move things around, even if the working product does not really exist.

5) Neutral atom approach. Atoms are held in place almost motionless at an extremely low temperature of several μK. Magneto-optical traps, laser tweezers, and laser cooling are used. A camera captures fluorescence of the atoms. It's very difficult to imagine a computer that could operate at such low temperatures. Plus, the camera is good for landscape photography, but not as good for sensing what Rydberg atoms are doing. This is not very promising. Several companies are working in this field; Google jumped into it after the superconducting computer flopped.

6) Photonic quantum computing. This has been around for quite a while. The only thing that you have to remember is that a photon is going to be scattered by either defects in the light-conducting medium or by the interface between that medium and the surroundings. Any bending of the waveguide is going to scatter photons even more. Photon loss is the result. If 100 photons enter the optical chip, perhaps 85 would exit it. 15 photons got lost somewhere inside the chip. Obviously, those missing photons carried important information, and without them, the output is now very deficient. Traveling photons have a wave spreading wider and wider, which limits the size of the chip. There is no solution to these problems.

7) Microsoft's topological qubits. So far, there is no convincing evidence that Majorana fermions exist. Qubits with zero-mode Majoranas would be immune to noise. In theory, yes; in practice, no. Every single quantum particle is always in superposition with random quantum noise, which is the result of that particle being probabilistic. The noise in all four corners of those qubits is different because noise can't be reproduced exactly in four different places. The presented data is inconsistent, suggesting that the qubits are not working.

Then, why does my laptop work, and why are there so many problems with quantum computers?

Your laptop uses binary chips. In a binary system, there are only two states possible: a low state referred to as 0, and a high state referred to as 1. There is a gap between those two states. The 0 state does not overlap with the 1 state, and the 1 state does not overlap with the 0 state. Binary chips use Boolean algebra. Mathematical operations and functions can be described with that algebra, and only a small number of different gates are necessary to perform many operations. The state-of-the-art CPUs contain hundreds of millions of gates. The most important part of a binary gate is that the noise generated by the gate can't change the gate from one state to another state. A signal reflected from a gate's input does not significantly affect the previous gate. Therefore, a chain of a very large number of gates can still function correctly.

The number of electrons and holes is still reasonably large. However, as MOS transistors get smaller, the number of carriers will decrease, and noise issues will become much more relevant. Eventually, the devices will become unusable. It is not clear whether the device size will reach its limit first or whether the noise will make those devices unmanageable. However, we are very close to that moment.

In a quantum computer, a quantum gate always generates random quantum noise, which is added to the noise from the previous gate and forwarded to the next gate. The next gate injects its own noise. After just a few gate operations, the output basically drowns in noise; the useful information the gates started with can't be extracted from it. And remember that the first gate can't be initialized to a reproducible state. It's the fundamental law of quantum mechanics discovered by me. Fault-tolerant quantum computing is a buzzword that means essentially nothing.

So by now you know why a quantum computer will never work. Yes, you can quote me on that. And you can watch TikTok videos on your binary laptop:

(Preferably this channel https://www.tiktok.com/@vatsek11?lang=en)