Advances in Quantum Computing

Well I imagine they run a series of tests for logic and accuracy inserting known problems and solutions to test the machine before they get to such a complex problem as the one mentioned...starting with simpler problems and ramping up to most complex and time consuming computations known...The time when a machine would give a wrong answer just for a joke that he could share with his supercomputer bro's is well into the future I trust....

The Chinese quantum computer is "hardwired" for one problem, much as gates and registers would be in a conventional logic circuit designed to perform one task.

"One difference between Jiuzhang and Google’s Sycamore is that the photonic prototype is not easily reprogrammable to run different calculations. Its settings were effectively hardcoded into its optical circuits."

https://www.wired.com/story/china-stakes-claim-quantum-supremacy/

The Chinese quantum computer uses photons, which don't require cryogenic refrigeration like superconductors used in Google's quantum computer, but the downside is the current lack of flexibility.

I understand the words in your statement, but your explanation begs even more questions.

How do you "hardwire" a computer for a specific problem (and, what does that even entail), which would normally take a regular supercomputer 12 million years to solve, and that no one could possibly already know the answer to, and then trust the outcome? I think a more specific honing of that question would be: how would one even know what specific types of calculations are needed to make for something so complex?

It seems to me that it would possibly even be hindering the process by limiting the type of calculations it could make.

Is there no IA software to guide the process that could make that sort of decision on the fly? How could anyone ever make the claim that they had a quantum computer if it isn't guided by some level of AI?

????

How do you "hardwire" a computer for a specific problem

The optical setup consists of 300 beam splitters, 75 mirrors, plus the 100 light sources and 100 detectors. All these are set up and aligned on an optical table for this particular experiment.

"Jiuzhang was working with 50 photons, 100 inputs, 100 outputs, 300 beam splitters and 75 mirrors."

https://newatlas.com/computers/jiuzhang-chinese-quantum-computer-supremacy/

" which would normally take a regular supercomputer 12 million years to solve, and that no one could possibly already know the answer to, and then trust the outcome?"

That's a very good question. The experiment done here is "Boson Sampling". My guess is that the "right" answer is what comes out of their optical quantum computer and the time required for a classical computer to solve it is computed from the complexity of the optical circuit.

"How do you "hardwire" a computer for a specific problem (and, what does that even entail), which would normally take a regular supercomputer 12 million years to solve, and that no one could possibly already know the answer to, and then trust the outcome?"

As a test, you could do what's done in encryption - multiply together 2 large prime numbers, which is easy to do, but the resulting number takes a long time to factorise by an ordinary computer. Try it on the quantum computer. Obviously the answer is known, so you know whether it got it right.

Of course, in most situations the answer is not known up front, or there'd be no need for the computer, so have to accept its answer. But that also applies to current machines.

They know if the answer corresponds to an answer they already knew.

Take a large random number and encrypt it using 2 10^x prime numbers (where x is very, very large). A computer or supercomputer cracks this type of code using a process of iteration which uses so many computational steps it would take years to complete. Feed in the encrypted random number into your quantum computer and ask it to crack the code but don't tell it the primes or the number of digits in the primes. If three minutes later it spits out the original random number it works. You don't have to crack to code with a computer or super computer you just need to know how many calculations would be needed multiplied by how long each calculation takes to complete.

Because it is not programmable, what the Chinese produced is not technically a computer which has the 'ability to be programed' as part of its definition. It is a difference engine, much like the mechanical one Charles Babbage used to calculate tide tables, or the "Bombs" used at Bletchley Park in WWII to break the Enigma code. Both these examples showed spectacular improvements in the time taken to perform repetitive complex calculations.

Your answer is very adroit and quite understandable. It also makes me wonder if the "quantum" qualifier is actually being exercised in the process you described.

I know that 'Quantum Entanglement' and the 'quantum' as applied to computers are not related; but, in my mind, at least some degree of the 'spooky magic' of entanglement has to be inherent in anything that wants to claim 'quantumness'. What you have described here doesn't even come close to qualifying for me, primarily because I can understand it without having to put any spookiness into the mix.

Your 'spooky magic' is fully encapsulated in the sentence "If three minutes later it spits out the original random number it works." I cannot explain how it does it in three minutes but one part appears to involve the ability of Qubits to be both 0 and 1 at the same time. So instead of doing two calculations, one for when the bit is 0 and another for when the bit is 1, the same result can be achieved in a single calculation ie. ½ the time. If you did that for 2 Qubits (2²) together it is ¼ the time. For 10 Qubits (2¹⁰) it is approx 1/1000 the time and for 20 Qubits (2²⁰) it takes approx 1/1,000,000 the time. The Chinese are claiming 48 Qubits so that is quite fast. Another part is to do with strings but whether that is computer strings or quantum strings I am not sure. As for entanglement, that is how my mind got when someone tried to explain this to me.

It seems like you could speed this up with some assumption built in, assuming it is either a 1 or a 0, you could assume they are all 1's and half the computation is already done, you only need to fill in the 0's.... This method is used in graphics...

For anyone interested in learning more about quantum computing, the following course is free:

https://medium.com/qiskit/were-releasing-a-free-hands-on-introduction-to-quantum-computing-course-online-5db213cea79

From what I have read this text looks quite interesting. However, the associated images are missing. Is the loss of those images detrimental to the ultimate intended learning experience?