Reply to post: How's factoring coming along?

How's factoring coming along?

It strikes me you can determine if you have an error correction problem, or something more fundamental by trying to find the private key corresponding to a public key.

5000 qubits is a long key! A 2048 will be good enough for such a test (=~256 private key).

You know the drill here, but I'll outline it for others:

The model being optimized is the public key generation from the private key, using any commonly used public key encryption algo.

Run the optimizer to try to get as close a public key as you can by twiddling with the private key values.

Generate the target public key using your known encryption algo, and start the solver off somewhere random (i.e. no cheating by starting it near the private key).

When you've got your result in a reasonable time, examine it. Since you know the actual real private key, you can see if the optimizer found the correct private key.

1) Exact public and private key? Well done, you've succeeded in your quantum aim.

2) Close to the real private key? Still pretty good, you have an error correction problem there. The value you read had some noise in it.

3) Close to public key and far away from private key? Oh dear, you've found a local minimum, that's an analogue optimizer you made there.

4) Far away from public key? Oh dear, you've made a very bad analogue optimizer.

Algos like this, don't work for regular optimizers, but a "quantum" optimizer, a-la Shroedinger would be going through all possible probablistic states and thus *would* work.

You can even compare it to stock digital optimizers, the kind we normally use the real world that you can simply pull from Apache Math lib (i.e. not an algo simulating your hardware pretending to be an optimizer ....no "simulated annealing" marketing crap please).

https://commons.apache.org/proper/commons-math/userguide/optimization.html

And see how they compare. They tend to be 3) or 4).