It’s hard to know if quantum computing is the 21st century equivalent of cold fusion (ie- always 5 years away), or if it really is something.
From what I understand (as an interested lay-person), d-wave and the other quantum companies are struggling to get more than a few qubits (5 or so?) working. Another big problem is that problems need to be formulated in a way that is compatible with quantum machines. This will require entirely new algorithms, programming languages, mathematics, and engineering.
Don’t get me wrong. This is an extremely exciting field, but we are still very, very far away from anything practical. I hope we get there, but we may not. It’s still a big, but exciting gamble.
Well, cold fusion didn’t have theoretical support and experimentally was at best … overly optimistic. Quantum computing on the other hand is well supported by established theory and demonstrated in toy situations. So it’s probably better to compare it to “hot” fusion reactors; it should work, but the engineering challenge is massive.
D-wave is supposed to have something like 2000 qubits. It’s just not universal. The existing universal quantum computers have a much smaller number of qubits.
They are making computers out of discrete devices. This is a bit like computers in the fifties. There is a lot of microfab technology they may be able to use, but it does not have anything like the development that silicon semiconductors have.
The quantum uncertainty effects usually stop us from knowing something. It is a bit weird that they can actually work for us in this way. There may be some fundamental limit on how much quantum parallelism they can get. But for now, it seems to work with simple problems.
If we were going to put similar efforts into other technologies, it may get beaten by making circuits entirely out of carbon. This could let us make molecular-sized components. You could get a lot of processing on a mol. of cores without quantum magic.
So: Early days. Much woo. May never happen for various reasons. But the thinking seems sound.
This is also an awful misstatement of the physics, and if someone on the board of an ostensible quantum computing company is spouting such nonsense it speaks volumes.
If you do subscribe to something like a Multiple Worlds Interpretation, that’s great for you. Even with such a model, quantum computing has nothing to do with cooperation between the multiple worlds to operate in parallel. Indeed, if all quantum computers bought you was parallelism, it would be worthless. We have had massively parallel computation for a long time now.
Intereference in quantum states is what makes things work interestingly. If you’ve heard of the famous double slit experiments for foundational quantum mechanics, that’s the same principle going on here. You write an algorithm so that when the states interfere, they cancel in the wrong answer space and interfere constructively on right answers. Simple as that! (Not simple.)
And if he claims to have met physicists who would rather not talk about quantum mechanics, I call bull on that too. Shutting us up is generally the hard part.
An aspect of quantum computing I heard of and find particularly interesting, is the ability to get an optimal solution to an NP-Complete problem in a quasi-linear time. The trick being that you only have a significant probability of getting the solution you want, but demonstrating that it’s optimal only takes linear computing time.
None of the big “Wow!” algorithms can be run on it, since it is not a universal quantum computer. It won’t do any of the magical stuff that a universal quantum computer can do.
In simple terms, for a universal quantum computer, the relevant factor is size (ie number of entangled qbits) not speed. The size determines the size of a problem you can encode in it, and it offloads the computation to alternate universes (which might well not exist the same way this one does, that’s a matter of interpretation) so that it runs every possible permutation of your problem at once. Most quantum algorithms, like Shor’s algorithm, don’t give you the right answer every time, but the answer is trivial to check with a classical computer, so you just keep running it until it gives you the right answer.
The D-wave is not a universal quantum computer. When they claim eg 512 qbits, what they mean is nothing like what a 512 qbit universal quantum computer would do. Rather, it is 512 1-qubit (or maybe 256 2-qbit, it’s been a while since I looked at the details.) quantum computers running in parallel. This means it can’t do any of the crazy stuff you could do with a 512-qbit quantum computer.
That said, the D-wave does use quantum effects to do computation, and far outperforms classical computers at simulated annealing. Since simulated annealing is a very general optimization algorithm, which is not used that often, because it is computationally much less efficient than other available algorithms, there is a lot of potential for the D-wave to be useful. It’s not all smoke and mirrors, it’s definitely for real.
Well in an infinite sea of universes any event however unlikely will happen. So while it seems like a problem of everyone waiting for someone else to peek over the wall and nobody wanting to… someone will have done it first. The problem is figuring out how to talk to them, and if we can do that we probably already solved the problem in the first place.
If the next iPhone is quantum, will it come with a Quantum Twitter (Quitter) app? (Of course, inter-universal leakage might explain some of regular Twitter.)
