New Quantum Computing Paradigm Could Make all the Difference

A new approach to quantum computing -- working with photons, or "photonic computing" -- has recently attracted a lot of attention, not least because I've been talking about it, cough. But there's something I haven't mentioned, which is that photonic quantum computing uses a fundamentally different method -- measurement-based quantum computing -- that could push forward its competitors as well. It's a somewhat technical story, but I've tried my best to give you the summary.

🔗 Join this channel to get access to perks ➜

#science #sciencenews #physics #tech #technews

[HammerOn-bugx]

One of things I like about your work, is that when you don't know, you say so. Thank you!

[Whysicist]

“Pushing around a bump under the carpet” - Priceless! Next book might be, “Saying of Sabine”. More, pls. Thank you for being you.

[strayorion]

What makes me think this was probably a really big achievement for those whose paycheck depends on this technology progressing, is the fact that sabine went from "it will never happen" to "Ok, it CAN happen"

[AutisticThinker]

I didn't really believe Open AI was going to release an AI (you can chat with) anytime soon either; and I'm a senior software developer. One major success and that industry will bloom new startups all over the place.

[aidanclarke]

Hi Sabine. I love your content and your honesty, but could you add some more "fundamental" stuff as well ? Just like in the "good old days" ? 😅 Like "what does this equation mean" and so on.
I would really appreciate.

Lots of love from France ❤❤❤

[jayernoud]

“Quantum technology evangelist” had me in stitches 😂

[edoardolopresti]

The "just checking if you're listening" sincerely worked for me lol

[thob]

For accuracy, the gates shown in the video at 2:06 are classical logic gates, they're not the same used in quantum computers, which are based on a model of computation that is reversible, and as such all have the same number of outgoing units of information as incoming, unlike classical logic gates which collapse bits such as the AND and OR gates which have two inputs and one output. Using this irreversible model in quantum computation would require destroying qubits as the computation is still ongoing, which is not how they're currently built.

[kirilrotan]

I want to add some context. There are two types of qubits: physical and logical. What matters are the logical qubits. While the internet buzzes about 4000 logical qubits (which are needed to solve a 2048-bit RSA), the mention of a million physical qubits doesn't provide much information.

[SuperFredAZ]

As a retired engineer Quantum Computing feels like something that will always be 25 years away.

[lordsqueak]

I love how Winamp is still relevant in this field. @1:05 ish

[gandru]

One thing worth mentioning is that in order for the measurement-based model to be universal (i.e. for it to be able to do any computation that can also be achieved in the gate-based model), one has to do adaptive measurements. This means that you measure one qubit, record the outcome of the measurement and then based on what that outcome is you will measure the next qubit one way or another (in say the Pauli X basis for a 0 outcome of the first measurement, or in a Pauli Y basis for a 1 outcome). This makes implementing the measurement-based model tricky, because you can't just measure all the qubits in parallel at the same time; instead, you will measure some of the qubits, then use the information from the measurement outcomes to measure some of the other qubits and so on. Unfortunately, measurement-based computations which do allow for a high degree of parallelism are usually not that interesting. For instance, a family of quantum computations where the measurement-based model allows for fully parallel measurement is Clifford computation. However, Clifford computation is efficiently simulatable on a classical computer, so it would offer no quantum advantage whatsoever.
That all being said, it does indeed seem that the photonic architecture is the best suited for a measurement-based approach. We'll see how well they handle this adaptivity issue.

It's also worth mentioning, that the other architectures (like ion traps or superconducting qubits) already have some capability to do measurement-based computations. They can do what are called "mid-circuit measurements" or "classically-controlled feedback" where you can measure some of your qubits mid circuit and then, depending on the measurement outcome, decide what gates you'll apply to your remaining, unmeasured qubits. These mid-circuit measurements tend to be very noisy (introduce errors in the computation), so you can't do too many of them without completely ruining the computation. Though this will likely improve as the technology gets better.
I should also say that mid-circuit measurements are necessary for eventually achieving fault-tolerant quantum computation, since these mid-circuit measurements will be the ones used to determine whether errors have occurred and how to correct them (in that context they're called syndrome measurements).

[aaronjennings]

I can say with some degree of frustration that i have understood nothing described here.
Even after watching it twice.

[chrisronan]

Cyberdyne Systems is making some interesting developments in AI.

[davidronbrothers]

This is well written and funny and the fact that you use intelligent jokes is a blast. But the fact that you’re doing this in a language that is not your native tongue is even more impressive. Trying to be clever and funny in a foreign language is an advanced ability.

[carnap]

New Quantum Computing Paradigm is Crazy

[carlbrenninkmeijer]

I once pushed aroud a bump under the carpet, it was my grandma's cat named Schrödinger.

[time]

Interestingly, several countries coordinated recently to establish strict security controls on quantum computers and theur components. My gut feeling at the time was that something happened in terms of a major advance, perhaps it was this as Australia was on the list of countries.

[mm-ytsf]

oh this reminds me of long ago i signed up for a class in college called "laser electronics" and thought this would be a new technology of light based computers instead of electron. i was rather bummed to find out it was about laser light and i think the "electronics" referred to electrons in lasers.... so not at all what i was dreaming about it. but i stuck with it for a bit and like much of science, it turned out to be more interesting when delving into the details...i had thought that a "perfect" laser beam was a beam of light that would travel forever unchanged unless it hit stuff...but learned that the fields in the beam itself would tend to spread the beam so the best one could do is gently focus the beam so that it would try to come together while at the same time tending to spread apart. so...nifty...and i would never have known that if i didn't misinterpret the course name 😀

[yeroca]

So no gates at all? I don't know how you "program" the thing, then. Guess I need to do some reading.