Hacker News Reader

Comment by helixc | Hacker News Reader

I have a question to the folks who are interested in quantum computation: do you feel getting benefits if you can have direct access to a physical quantum computer? In other words, what hands-on experience you hope to obtain, but cannot make it using a quantum simulator or cloud based quantum programming environment?

View on HN

No, nothing. The only thing that a real machine is useful for now is testing that the machine works. Other than google's recent contrived test, any quantum algorithm ever executed will run just as well (or better) on a simulator as on a real machine.

It's going to be like that for a while, even after we're well into the quantum supremacy realm. First, there's not anything you can do with a real machine until error correction is working, which requires 10k(?) qubits. And frankly, there isn't a whole lot you could do with even a real million qbit machine now if one were to exist. Prime factorization is the big one, but other NP complete problems have no known quantum algorithm to speed them up (and note prime factorization is not NP complete -- so it's possible no quantum speedup for any NP complete problem exists (or it's possible that P==NP in which case....)). The real work is in the math and algo theory to find solutions for these problems. Coding and running them is actually kind of incidental and "cute".

In addition to that, Stephen Wolfram hypothesized that QC won't be possible because with the growth of QC state, the energy required for error correction will grow just as fast, so the theoretical exponential speedup will require exponentially more energy, and this is why the today's QC devices has to split QC bits into small groups.

Many people hypothesize similar things, but there is as yet no real reason behind these hypotheses.

Still, it would be an extremely exciting discovery for physics if it turned out that a QC is not physically realizable. It would prove that quantum mechanics is not the final description of the world, and it would likely be a huge step forward for understanding the measurement problem.

> And frankly, there isn't a whole lot you could do with even a real million qbit machine now if one were to exist.

Aside from being able to break the majority of asymmetric encryption schemes, in particular RSA as you mention with prime factoring.

Mining bitcoin would be a good use, right? Probably would be able to break even on the cost.

I work in a QC lab, and here's my perspective. There is one pretty good reason for executing programs on real hardware: because we're still pre-error-correction, errors are rampant, but they are hard to model correctly. After all, if you could model them, you could simulate quantum mechanics, and then... why were you building the thing? So, simulators have to make some strong assumptions. Running programs on real devices can teach us more about the physics of those devices and about how compile programs for them.

There's one other really good reason: it's fun!

Very insightful! Could you share how many qubits you can use in your lab? I guess one need at least 3 qubits to run some interesting quantum algorithms?

As a matter of fact, we published just this year on a novel algorithm using just two qubits! The trick is that this algorithm is in some sense "generic" over arbitrary data structures, so we were able to claim victory by tackling the simplest possible case. It was still a challenge in practice, since it required a lot of sneaky tricks and record (to our knowledge) two-qubit gate fidelities.

We're also working with larger devices, but I'm not sure if I'm supposed to comment on those right now :)

Super interesting! Could you share the link to your 2 qubit algorithm paper? I'm also doing quantum research, bit on the hardware side, and really interested in knowing algorithm development.

I was interesting in quantum computation / programming about a year ago, but failed to really get into it (I'm too dum, I think :)). Mostly, my interest was in understanding the mental model.

I learned Golang like 1.5 years ago, and it totally changed how I think about concurrency. I think about interacting processes way different now, as a result - and I was kinda hoping the same thing would happen with ~quantum~. Blarg.

Quantum mechanics is certainly mind-changing, but it can't explain yet what our world really is. The most popular many-worlds interpretation is very far from complete.

Many-worlds is definitely not the most popular interpretation among practitioners, Copenhagen is by far.

Just quantum supremacy experiments. Running things on quantum hardware instead of using a simulator will just be slower and more expensive for the foreseeable future.

My 2c. From the programming perspective, QC isn't much different from GPUs or ASICs: a slightly different model, a slightly different language. For this reason, learning QC now, when there are no QC chips available, is like learning an obscure VHDL dialect to program a not yet released chip. If QC chips become available, I should be able to pick up QC programming in a matter of weeks, if necessary,

Do you have any QC experience? Because I strongly disagree with your statement, depending on the level of abstraction that you mean.

At some point in the future certain quantum algorithms will become 'callable' from traditional programming languages, and will be usable by any competent programmer without much hassle. But making the quantum algorithms themselves is quite a different ballgame altogether, very much unlike traditional programming.

I'm a pragmatic person. If QC mental model is so complex that I can't learn it in a few weeks, then very few in the world can and QC will remain a subject of scientific papers understood by a few.

haha i am waiting for a smart nerd to answer this. you elegantly said what i thought "but what can i do with it?"

Reply on news.ycombinator.com