IBM unveils 433 qubit quantum computer
newscientist.com
newscientist.com
I'm cheering them on!
- Shor's algorithm is about the only productive thing we've figured out where a quantum computer has an advantage
- IBM unveils 433 qbit computer, to factor 4096 RSA we're gonna need 8192 logical qbits
- Those 433 qbits are physical qbits, and there's (to the best of my knowledge) no literature on their measurement fidelity, coherence timelines, etc. These are purely experimental machines, it's not like you can plug your laptop into their giant cryogenic machines and run quantum algorithms without a giant team of literal quantum physicists standing by the thing and monitoring it
- How those 433 qbits actually translate into any amount of meaningful computation is up in the air
- The best we figured out was 5 logical qbits running a specialised Shor's algorithm on the 127qbit computer they unveiled a few years back to factor 21=3*7
- Nobody has, to the best of my knowledge, actually run the full Shor's algorithm on anything of note, even on these systems
Not a quantum physicist; I only have vague memories from a quantum computing seminar I took long ago. Feel free to correct if any of the above is incorrect.
And when we actually build a quantum computer (which, I hope, happens by the end of this century), everybody will be too bored about it.
But let's say it can, and somehow IBM forgot to mention.
Shor's algorithm reduces to a quantum Fourier circuit, with 2N qubit inputs and 2N qubit outputs, where N is the number of digits of the number to be factored. That's 4N qubits only for inputs and outputs.
So, in the best case scenario, this computer would be able to factor a 108-binary digit number.
My computer (2020 Intel-based iMac) can do that in one minute.
Here's the code if you want to try it on your computer:
import sympy
n_digits = 54
p = sympy.randprime(2**(n_digits-1), 2**n_digits)
q = sympy.randprime(2**(n_digits-1), 2**n_digits)
sympy.factorint(N)
Output: {53882560694857690729: 1, 44658549937294814041: 1} N = p * qwhat can this machine do productively?
The one thing quantum computers are (currently and theoretically) able to solve efficiently is the factorization problem, but are very much limited by low qubit number, i.e. engineering problems.