IBM unveils its most powerful quantum processor | Hacker News Reader

IBM unveils its most powerful quantum processor

engadget.com

116 pointsrbanffy23 comments

> the platform can already solve problems that were considered too complex for classical computer systems to handle

Uhm, quantum computers can be simulated by a classic computer with exponential slowdown. So the speedup can be no more than on the order of 2^17 = 131072. But I think it's fair to assume that a single operation on the quantum computer is much much slower than state of the art classic computers. So doesn't it seems doubtful that they can actually do anything faster with this than with a classic computer? I would think that we have to get in the high 20's to low 30's of qbits before it can actually beat a classic computer or a cluster of them.

Actually, best supercomputers can simulate some 40+ qubits https://arxiv.org/abs/1704.01127

As far as I can tell, IBM has not unveiled anything at all, and this Engadget journalist is just misinterpreting an IBM statement on "quantum volume." Does anyone have a link to an IBM release? In particular, it looks like the public can still only access a five-qubit device.

https://quantumexperience.ng.bluemix.net/qx/community?channe...

http://www-03.ibm.com/press/us/en/pressrelease/52403.wss

The public access machine is available through a beta invitation request program - https://quantumexperience.ng.bluemix.net/qx/account/beta-acc...

(disclosure: I work for IBM Systems, not related to Q though)

Just imagine how quickly companies will be able to match your profile with suitable advertisements! The dream is real.

I missed the first release of 5 qubit processor since the IBM website has so much marketing jargon. I find some papers in arXiv where they use the processor. Is there any reference demonstrating that this is a true quantum processor? I'd like to see some non-classical scaling with respect to N, etc.

N=5.

It looks like it's a real quantum processor and not just a simulated annealing engine like D-Wave?

Wait, so how I can write quantum instructions and run them through IBM's cloud? How long before someone gets Shor's algorithm running and destroys the world?

You need around 4000 qubits to break RSA-2048; IBM's system only have 17 qubits so far.

It is much worse than that. You need 4000 logical qubits which are implemented as an error correcting code on top of physical qubits. This has 17 noisy physical qubits. You need millions of physical qubits to implement thousands of logical qubits.

Curious, what about 4096? Where do you get these numbers from?

I'm not the person you were replying to, but this [1] stack exchange answer provides an explanation. I'm not qualified to say whether it is correct though.

[1] https://crypto.stackexchange.com/questions/35137/how-many-qu...

Can you not break the work into smaller pieces? Or am I missing something fundamental here?

superposition, you need to have as many qubits, because quantum magic can only do its exponential wizardry if you have the whole system in a coherent quantum superposition, and then you can do measurements (set up the problem), and then (let the system evolve, and then) measure the output(s).

And following that, what is the likelihood some government agency already have secret projects with larger number of bits already?

links at the bottom of the article?

SDK

https://github.com/IBM/qiskit-sdk-py

Online Editor

https://quantumexperience.ng.bluemix.net/qx/editor

Or DOES it? Sorry, bad quantum joke ;)

I thought it was funny, that and the one where gcc doesn't tell you if your program compiled or not on quantum computers, until you observe the object file it is both syntactically correct and uncompilable.

What's going on with the website? Does any know what the 'amp' in the URL stand for? Is it the same as Google's AMP project?

Here's the non-AMP version: https://www.engadget.com/2017/05/17/ibm-quantum-q-experience...

The markup certainly contains references to Google's AMP project, which makes your guess seem likely.

  <amp-sidebar...

etc.

Found some links to ampproject.org in the source, looking at the website, I think we've found the answer.

Surprisingly I don't think that website is affiliated with Google. I could find only a single reference, except on this page[1] once, and in passing

[1]: https://www.ampproject.org/how-it-works/

EDIT: Scratch that, found this:

https://googleblog.blogspot.com/2015/10/introducing-accelera...

I was about to give a bunch of praise for raising its ability to reach the community and create a very engaging video but then they pulled out an introduction page[0] that does not work on firefox (the SVGs for the superposition states render as blanks).

So close.

[0] - https://quantumexperience.ng.bluemix.net/qx/tutorial?section...

Reply on news.ycombinator.com