Still haven’t heard of a single valid application of quantum computing which is not breaking encryption.
Optimization algorithms, sure. Just make sure your problem is an appropriate graph. Does optimizing railway timetables make building a quantum computer for that viable?
Computational chemistry and protein folding. That’s actually great! However nobody working on such problems has ever been able to explain what is the theory which would allow such calculations. Most of them just say stuff like atoms have quantum properties and thus using quantum computers you can calculate those properties. Even disregarding the theory availability, when I ask of then they’d be able to work on systems comprised of tens of thousands of atoms all of them start panicking as apparently it is probably physically impossible to build a computer that retains coherence at the number of obits necessary to encode such a system.
I mean, this would all be nice; but do we even know what the purpose of such machines could be?
Basically they have all the utility of a supercomputer. Breaking encryption is a use sure. As is computational chemistry and protein folding and interactions. The reason it can do so is because of the information processing. It doesn’t mean new equations but that the equations that are made can be tested faster. There’s various times where equations to help understand bio/chemistry are limited to remain feasible to use but reduce the precision and accuracy.
I’m afraid you did not read the message I wrote or you did not understand it.
A quantum computer is absolutely not equivalent to a supercomputer. There are calculations you can do in a quantum computer which you can not do in a supercomputer, most calculations which run on a supercomputer can not run on a quantum computer. In theory you can run any traditional algorithm on a quantum computer, in practice it is not feasible: you’d end up having a huge quantum computer with billions of qbits and still run slower than a single GPU.
New equations are needed to run that kind of calculations on quantum computers. We have algorithms to solve graph problems on quantum computers, but those equations can not be expressed as a graph. Such algorithms do not exist yet.
In computational chemistry there are basically two types of simplifications: quantum chemistry, which uses the born Oppenheimer simplification, and classical approximations, which use classical physical equations.
None of the two methodologies can be replicated directly with quantum computers. You can not take the Hartree Fock equation and the methodologies used to solve that equation and solve them in the same way on a quantum computer.
Ideally, the option I have seen would be to encode the molecule itself in the computer. This would broadly mean something like each qbit represents an electron of the molecule. Then you somehow connect the qbits in a way in which they behave like the molecule does. In this way you do not have to solve equations, but rather just measure the end state of the system.
However, each carbon atom has 12 electrons and a protein has tens of thousands of carbon atoms. This places the minimum amount of qbits for encoding such a system in the hundreds of thousands, whiteout accounting for error correction.
However, the problem lies in the amount of qbits. Even if you were to build a computer with a large enough number of qbits, you’d have to program them one by one through a set of operations in order to place them in the correct quantum state to represent the molecular system. These operations take some time, that would be quite a long operation over hundreds of thousands of qbits, possibly it could span several days. The amount of time is not the problem itself: solving such a system in just a few days would be great! However, over time qbits lose their state and the system becomes incoherent. You expect it to be in a certain state due to the operations you performed to program it, but it no longer is in such a state.
As long as I understood, it is plausibly physically impossible to maintain coherence over such a large amount of qbits.
Could they devise an algorithm which requires less qbits? Sure, I mean even this algorithm does not exist; it’s just the most viable idea I’ve seen. However as of now no such algorithms exist. And I do not understand why: you do not need a computer to design the algorithms.
I’m probably out of my depth as to how a quantum computer functions but the way you describe it breaking encryption would also be rather useless as well. I had thought it was just a system wherein the code is switched from binary 1/0 to one where you can do +0/1 neutral 0/1 and -0/1 or something like that. If you have to be dealing with individual qubits then it sounds useless.
No, regarding breaking encryption, it can definitely be used. Algorithms to do that already exist. We are missing a large enough computer for that to be feasible, but once it’s available everything which is encrypted with a non quantum safe algorithm can be decrypted.
Luckily they already developed encryption algorithms which should not be broken by quantum computers.
Regarding the working of the computer, yes you can see it a bit as rather than having a bit on or off you can have intermediate values as well. However qbits can also be “linked” to each other. As such you don’t really work on individual qbits but rather on the overall state of the system: changing one qbit will change the value of other ones as well.
Then also you have to be careful about measurement: if you want to read the result you’ll have to measure the qbits, doing that will change the state of the system and you have to account for that.
However, I’m not an expert in quantum computers. I have not studied them or worked on them. Most I know is from bar talks with physicists.
“useful”.