IBM unveils first quantum computer in Germany | Hacker News Reader

IBM unveils first quantum computer in Germany

dw.com

42 pointsqwertox51 comments

> Regular modern computers process functions in a binary fashion, carrying out tasks that use fragments of data that are only ever a 1 or 0. Quantum computers use subatomic particles to perform calculations at far greater speeds than existing supercomputers, and data can be both a 1 and 0 at the same time.

twitch

(this is very wrong)

Can someone please ELI5 a better explanation?

This recent link posted on hacker news explains it quite well:

https://news.ycombinator.com/item?id=27439839

"What makes quantum computing so hard to explain?"

A classical bit is either 0 or 1.

A qubit is a complex unit vector, wherein the classical states of 0 and 1 correspond to antipodal points on the unit sphere painted by the range of possible qubits.

Hardly an ELI5

Perhaps some things (thankfully for our employment prospects) can't be explained to a 5 year old ¯\_(ツ)_/¯

I'd imagine the following Google image [1] is sufficiently understandable to a child: the two dots are bits while the sphere is a qubit.

[1] - http://langsrv.astro.cf.ac.uk/research/research_SQDFWM/Bloch...

This of course does not cover everything, but I'll try.

Quantum computers store data as states that are either 0 or 1 when you measure them, but before measurement they are in a superposition of these two. Quantum computers are initialized in a ground state, and are programmed by slowly modifying the system such that the new ground state is the solution to the problem you care about. Then you do the measurements and the state collapses to the solution you want.

There are certain specific problems where evolving and collapsing the system in this way is demonstrably faster than the best classical computer algorithms we have today.

Wait, is this like ML models but instead of generating the models by repeating inputs to get a “best guess” engine, you put in the intended outcome and it.. Magically has an engine?

Maybe the closest analog is: imagine if you have that type of magic hydrophobic sand that doesn't get wet, like you can buy for kids. So you have a sandbox with this, and initially there is a big circular valley in the middle of the sandbox that is filled with water. Then you can slowly move the sand around, making different hills and valleys, and the water will follow the sand always in the minimum energy configuration, until the shape of the lakes in that terrain represent the solution to whatever problem you care about. Then you take a photo and measure exactly the distances, angles, areas etc.

With this analogy you get some feeling for what manipulating the ground state means, and why you need to do so slowly, and also that this is not a tool that will solve all types of problems.

This shit blows my fucking mind.

So do quantum computers give better implementations of languages like Prolog?

https://en.wikipedia.org/wiki/Prolog

No. They are "programmed" in languages entirely unlike anything we are used to (at least for now).

"Quantum computers like the Q System One are still very much experimental devices. They can’t outperform classical computers at useful tasks (in fact, your laptop is probably more powerful when it comes to real-life computation), but are instead supposed to be research tools; letting us work out, qubit by qubit, how quantum devices might work at all." from https://www.theverge.com/2019/1/8/18171732/ibm-quantum-compu...

"They can’t outperform classical computers at useful tasks"

Then calling them quantum computers is a marketing stunt. A quantum computer is something that can factor an integer (or a similarly quantum task) that no classical computer can (in a reasonable amount of time).

Uh no, a quantum computer is a device that can execute quantum algorithms using real qbits. The machines featured in asymptotic analysis don't exist in the real world. It's not like your classical computer has an infinite amount of memory that can be randomly accessed in constant time.

I'm talking about capability, not theoretical properties. Can any current so called quantum computers actually perform any computations faster than a classical one?

To be called a quantum computer, I would expect they should, but last I checked, all claims of quantum supremacy appear to be highly disputed.

I mean calling a quantum computer a machine that uses qbits to compute, completely ignores the reason why we are interested in them. You don't see millions of dollars poured into gear computers.

Theoretical properties of quantum computers are all asymptotic. Asymptotic runtimes don't always correspond to real world performance. The hope is that we will eventually be able to build quantum computers such that the constant factors don't mask asymptotic speedups for problem sizes that we actually want to solve. To figure out how to do that we build research machines.

Why would this matter? Let's say a research company invented a three dimensional computer chip. It would still be a breakthrough, even if the initial samples could only run very slowly.

> Can any current so called quantum computers actually perform any computations faster than a classical one? To be called a quantum computer, I would expect they should ...

But why? A very slow silicion based computer is still a silicon based computer. It doesn't matter how fast, same applies for quantum computers. There is no such definition as IF (> 1000 qubits) THEN "quantum computer" ELSE "nothing".

To super oversimplify this for explanation, let's say our quantum computer runs today with a clock cycle of 50 megahertz.

The quantum computer is able to do certain tasks in fewer clock cycles but since we measure time in seconds a regular 3.2Ghz 8 core CPU today would perform 5,000x the clock cycles per second.

If quantum is going to scale and work is still TBD but saying they are slower in seconds isn't a fair way to grade. They could be 1000x faster in clock cycles and still 5x slower in seconds today.

Link on IBM Quantum https://indico.cern.ch/event/719844/contributions/3019718/at...

No they can't. It's possible they never will. Fortunately, the definition of quantum computer is not "very fast computer" as you appear to think it is so this isn't relevant to the question of whether something is or is not a quantum computer.

Not quite: whether integer factorization is faster on a classical or quantum computer is up to debate iirc.

They said, there _are_ algorithms that are provably faster on quantum computers, such as a random walk on a glued tree.

I am always confused whether the quantum computer are over-hyped like fusion reactors or are in the tipping point of having practical and wide usage.

Would love to read a sane analysis of this question..

Take your confusion about that as your first data point on whether it is over-hyped or not ;)

I don't think there is any way to build a quantum computer on earth with a significant amount of qubits (enough to break ECDSA with Shor's algorithm), where you can entangle the qubits as you want. The system would be too sensitive to the slightest vibration.

Yes same, though I read an argument saying there's no theoretical reason we can't build one, so if we try and can't we'll probably discover some new physics.

like graphene ?

I can't speak to the technical merits of the machine, but I think IBM did a great job with the industrial design. It looks like some kind of alternate history where super-integrated vacuum tubes won the day.

I was surprised to learn that 99% of the shiny parts in the photo belongs to the dilution refrigerator (https://en.wikipedia.org/wiki/Dilution_refrigerator) needed to get the machine down close to absolute zero. The actual computer is a tiny chip at the bottom of the apparatus!

Surprisingly, modern-era IBM supercomputers and mainframes are some of the "sexiest" machines out there in terms of industrial design. Apparently they put a lot of effort into that

https://cdn.wccftech.com/wp-content/uploads/2018/06/Summit_O...

https://images.idgesg.net/images/article/2017/07/ibm_z_4-100...

Over 2 years old at this point, but still highly relevant:

https://scottlocklin.wordpress.com/2019/01/15/quantum-comput...

Indeed, but it didn't age well in one aspect: "If I were in charge of science funding, the entire world budget for this nonsense would be below that we allocate for the development of Bussard ramjets, which are also not known to be impossible, and are a lot more cool looking". IBM made quite some progress on the industrial design front.

With IBM's materials I genuinely can't tell how much is real working technology and how much is salesmanship and hype. Can anyone with more expertise comment on the current state of quantum computing and whether any of it is exaggerated?

IBM is really good at hyping tech and then finding a mark and squeezing real hard. Just look at how they wrung hospitals for exorbitant amounts of money for Watson. If someone bought an IBM solution they should be fired.

Ha, your comment is ironic since I still remember from back in the 70's/80's that "nobody got fired for buying an IBM". The quote might even be older than that, but I had first heard that quote when Microsoft was still on the rise handing IBM their butts in the fledgling PC marketplace.

That's what I was going for.

The current state of quantum computing is very far from actual applications, since adding more qubits results in higher entropy, incoherence and other disturbing effects. Source: Science 2020

IBM and Google are about neck-and-neck with state of the art quantum computers (both 53 qubits, each using fundamentally different technology). So yes, this one is real. However it's all research at the moment, there are no practical applications that aren't better solved on a classical computer.

However IBM is now unable to unveil computer's momentum.

> IBM unveils...

Ugh. At this point whenever is see IBM in the headline I know there is a hook in there with PR spin.

I think I might stop reading about IBM all together. I don't have time to actually find the truth behind PR smoke and mirrors.

> Martin Jetter, IBM’s chairman for Europe, the Middle East and Africa, said IBM is working towards making a stable quantum computer capable of handling more than 1,000 qubits by 2023.

Is 1000 a lot?

If it's not logical qubits then it isn't for practical purposes but I guess it's a qubit record.

Is this new computer intended to prove some new approach? Why does it exist? Why didn't they build the 1000 qubit model instead? After all, they claim to know how to do that.

German government is using this as a PR of "innovation made in Germany" and their successful politics, although it has nothing to do with them.

So today, like in the past IBM is blessed with German money.

Is it just me or these “quantum computers” should be actually called “quantum computer emulators”?

They're using actual Q-Bits, so I would say its an actual Quantum computer. That doesn't make it any good though. Nor does it dismiss the claims of marketing hype. This is exactly that.

"funded by German taxpayers"

As if foreigners living in Germany doesn't pay taxes.

I honestly just read that as "people who pay taxes in Germany"

Yes, that's the correct use of the phrase and how exactly how it's being used 100% of the time.

I’ve referred to myself as a German taxpayer the same way I call myself a permanent resident of Germany, and feel this entitles me to write my local member of the Bundestag - who apparently agrees, responding with a far more direct email addressing the topic than I’ve ever gotten from my US senators or congressman.

I am a bit bothered by things that say “BürgerInnen” have the right to something all legal residents have (like the Covid-19 vaccine), for fear that some of the less foreigner-friendly types are tacking on a “Staats-“ prefix in their heads, but that’s on me.

It's the tax that is German, not necessarily the payer. Also when it comes to foreign citizens paying taxes in Germany it is a bit more nuanced. In general German tax law goes to great lengths to avoid double taxation, so no matter who you pay your taxes to you are treated fairly.

Reply on news.ycombinator.com