Basically (hence fast and dirty), future quantum computers are expected to have massive parallel computing capacities, which would mostly derive from the nonlocal properties of quantum objects.

A local object carrying a classical bit can explore only one trajectory, or one circuit at a time. A nonlocal quantum qubit could ideally explore the whole system in a single pass (another interpretation would indeed be that it would be in a superposition of all possible trajectories)

A key issue is that any measurement would destroy the quantum properties of the qubit. Hence, no classical storage of data, no reinjection in logical loops, etc. The quantum computation will have to be performed as a whole, which mean very different approachs to programming.

Thank you Peter Breuer for your words. What I got from Eric Picholle is what I get from the internet. But like you said, "where's the "compute" in that?"

If I knew how to make a logic gate, I will be satisfied without knowing any more details. (I don't know the details of a classical computer but I assume that after gates are constructed the remaining functions are the same for classical as quantum so I don't need to know those details.) How do we make a logic gate using quantum? Also, how do we store data? Do we write data by preparing a spin system in a sequence of spin-ups and spin-downs, then not disturb the system, then check the sequence of spin-ups and spin-downs later to read the stored data?

I get that quantum computing is not easy to explain to a layman, including someone like me that is moderately competent in quantum mechanics but knows nothing about how quantum computing works. The internet sites I found were pathetically (embarrassingly) useless in those explanations. Yet some people do understand quantum computing. How did they learn it? Not from any explanations that I found. Probably from studying very sophisticated works. Can any of those experts explain it in a way that can be understood by a layman that is moderately competent in quantum mechanics? How do you make a logic gate, or do you? How do you write data and then read it later? These two questions are completely avoided in the internet explanations.

Thank you for the correction, Peter Breuer. So let us forget about gates. But we still need some way of inputting data and either instructions or a substitute for them (like analog machines use), and we need to read answers. So maybe I will be happy enough if someone would explain how those things are done (I don't mean what buttons to push on a keyboard I mean how does it work). At least that would be a first step towards an understanding.

By the way, you are clearly an expert on computing so if you have trouble understanding QC then I am in good company.

There is a very good textbook on quantum computing, authored by Michael Nielsen and Isaac Chuang, called Quantum Computation and Quantum Information.

Its first edition was written back in 2001, if I'm not mistaken, so it's rather out of date from experimental points of view, but still is a must to get the theoretical part right.

Actually, when we talk about QC, it's better to explicitly state what you mean. The so-called universal quantum computer must be built of a set of quantum gates, which act on the input qbits and transform their states in a certain way. Obviously, in order to implement some logical operations, mentioned above, one needs two-qbit gate, the most commonly used in the algorithms is CNOT, which from physical point of view is an entangling operation.

I agree that there is very little scientifically sound, but popular explanation of these concepts, so I guess one has to learn QC the hard way )))

Ok, I see your point, and it's a good one.

I'm no expert in computer science, so computation and calculation seem to be not too different for me.

The point is in which tasks QC can lead to a serious advantage over CC. Quantum algorithms that offer the greatest speed-up (theoretical) over classical algorithms are those where the same operation must be applied to a great number of inputs (such as in quantum Fourier transform, Deutsch's algorithm, which allows to tell if a function is constant or not constant from just a single calculation of a quantum scheme that somehow implements it).

In a lot of practical algorithms, which include branching, conditional operations etc, QC can't lead to any speed-up.

Recently, a lot of attention has been paid to quantum simulators which help predict some complex quantum systems.

So, using your terminology, quantum computing indeed is better be referred to as quantum calculation.

I see that I was too ignorant to appreciate the depths of these discussions, including the meaning of "compute", so I lowered the standards of what will make me happy. I'm never going to understand this topic in depth but don't need to in order to be happy. I just need to understand a tiny little bit. Nikolai Ushakov made a statement that is very interesting to me. The statement is:

"The so-called universal quantum computer must be built of a set of quantum gates, which act on the input qbits and transform their states in a certain way."

If I can understand the physical mechanism, in quatum mechanical language, by which a gate acts on a qbit to transform its state, I will be happy. Unless the answer is "read a book" which will make me less happy.

I am the least sophisticated person on this thread, so my questions are probably the easiest to answer. I only care about the physics. I assume that quantum computing requires at least three physical or mechanical steps. They are used in some complicated way to produce "computing", but I don't care how they are used (because I'm not smart enough to even understand it), I only care about the steps. They are:

1. Prepare a quantum system in some user-selected quantum sate.

2. Make some kind of change to that quantum system.

3. Detect the new state of the quantum system.

I will be happy if I just know how those steps are done.

Again, if you Peter Breuer are unsure of this stuff, then I am in good company. From what I read, the quantum states actually used might not be electron spins, but the concept is the same as it would be for electron spins, so we can use electron spin vocabulary for the concepts. Let us assume that there is a way to prepare electron spin states to be either up or down, and also a way to detect whether each electron in some collection is spin-up or spin-down. Given all of these abilities (I assume the abilities exist for the actual case, whatever it is, in which electron spin is only a conceptual model) then electron spin states (when all are always either up or down) have the same role as bits in classical computers, and my elementary questions go away. Also, quantum computing becomes the same as classical computing, except maybe smaller in physical size, but no other advantage. A problem I have with this picture comes from other things I read from the internet in which a major advantage of QC is from the fact that an arbitrary spin state (e.g., spin-left) is a superposition of spin-up and spin-down states. So, unlike a bit, or a spin state that is either spin-up or spin-down, we might say that the superposition of spin states contains more information than just 0 or 1. My problem is how do we detect that information? The only way to detect a spin state without disturbing it is to already know which axis it is aligned with. A spin-up measurement can detect either spin-up or spin-down without disturbing the state if the state already was either spin-up or spin-down. But if the state was spin-left (for example), the above measurement disturbs the state in an unpredictable way. Probabilities are predictable but so what? How do we do any computing when our observations have unpredictable effects? I don't get that, unless maybe we do repeated trials (like Monte Carlo) and reach conclusions from the statistics of those trials, but that does not seem (to me) to be a very efficient way to compute.

So, I don't see how superposition helps at all. Superposition is supposed to be one of the things that distinguishes quantum from classical and it is supposed to give QC an advantage over classical computing, but I don't see how superposition helps at all. This is what I am struggling to understand. Some people understand this, but I guess they can't explain it in layman's terms, or if they did, it is difficult to find on the internet.

Why can't some expert just tell us the answer? A number of questions have been asked. Pick out whichever question you can answer and please tell us the answer. The fact that it is so difficult to get any understandable answers to any of the questions makes me wonder if I even want to spend any more time on this when I'm still trying to learn about a truck load of other topics in physics that I haven't learned yet.

One thought is that the problem with getting answers here is that this RG thread is not being seen by enough people, so the specialty experts are not seeing it. But another thought is that the discussions easily found on the internet are embarrassingly incomplete. So, I think that only a very few people understand this. To get answers we need to reach those special people. Is there a way to make these questions more public to reach a larger population?