Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Qudits: The Real Future of Quantum Computing?

spectrum.ieee.org

Post reply on HN
Print view View original

21–30 of 36 posts

Re: Qudits: The Real Future of Quantum Computing?

#21

Post by rdtsc »

adamnemecek wrote: post #4

You don't actually want qubits, you want an analog computer with differentiable signals. Most likely photonic. Qubits are a dead evolution branch. I've been recently exploring computational metamaterials for photonic computation. http://users.ece.utexas.edu/~aalu/research%20-%20page%203.ht... (there's quite a few papers on this but unfortunately they are all paywalled. Spoiler alert, they seem to be based entirely on…

> Did you know that on an analog machine, you can do sort in O(n)?

Can it sort names or addresses?

Otherwise for integers you don't have buy spaghetti and just use a computer to do a radix sort which is O(nw).

Another thing with analog computers is to think how they scale. I can sort 1M items on my laptop. What is the practical application of buying 1M strands of spaghetti.

Re: Qudits: The Real Future of Quantum Computing?

#22

Post by adamnemecek »

rdtsc wrote: post #21
adamnemecek wrote: post #4

You don't actually want qubits, you want an analog computer with differentiable signals. Most likely photonic. Qubits are a dead evolution branch. I've been recently exploring computational metamaterials for photonic computation. http://users.ece.utexas.edu/~aalu/research%20-%20page%203.ht... (there's quite a few papers on this but unfortunately they are all paywalled. Spoiler alert, they seem to be based entirely on…

> Did you know that on an analog machine, you can do sort in O(n)? Can it sort names or addresses? Otherwise for integers you don't have buy spaghetti and just use a computer to do a radix sort which is O(nw). Another thing with analog computers is to think how they scale. I can sort 1M items on my laptop. What is the practical application of buying 1M strands of spaghetti.

> Can it sort names or addresses?

Why couldn't it? Ideally they can represent real numbers with infinite precision which are identical to strings.

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

But even in non-real computation this shouldn't be a problem, the same way you can sort strings even though you have only 32/64 bits of memory at a time if that's what you are hinting at.

> Another thing with analog computers is to think how they scale.

I don't think that scale has ever been a problem for analog computers.

Re: Qudits: The Real Future of Quantum Computing?

#23

Post by gizmo686 »

petersjt014 wrote: post #2

I know these guys understand quantum anything more than I ever will, but is having a larger radix specifically a practical way for increasing performance? If a binary device has a hardware fault, bits that are in error still have a 50% chance of being correct. It seems like a single component failing here would make this worse (five times worse if it wasn't quantum). I don't know whether or not the pros outweigh the…

I am not an expert.

The issue that quantum computers have is not so much that individual qubits get errors, but rather that quibits become entangled with the rest of the universe, erasing our ability to take advantage of any quantom effect. To prevent this, you need to very carefully contain the quibits to prevent them from interacting with anything apart from highly controlled manipulations. The more quibits you have, the harder this becomes. If you can replace this with quidits that have more states, you can do the same computation, while needing fewer particles, making containment easier.

Re: Qudits: The Real Future of Quantum Computing?

#24

Post by rdtsc »

adamnemecek wrote: post #22
rdtsc wrote: post #21

Earlier quoted context omitted.

> Did you know that on an analog machine, you can do sort in O(n)? Can it sort names or addresses? Otherwise for integers you don't have buy spaghetti and just use a computer to do a radix sort which is O(nw). Another thing with analog computers is to think how they scale. I can sort 1M items on my laptop. What is the practical application of buying 1M strands of spaghetti.

> Can it sort names or addresses? Why couldn't it? Ideally they can represent real numbers with infinite precision which are identical to strings. https://en.wikipedia.org/wiki/Real_computation But even in non-real computation this shouldn't be a problem, the same way you can sort strings even though you have only 32/64 bits of memory at a time if that's what you are hinting at. > Another thing with analog computers…

> I don't think that scale has ever been a problem for analog computers.

I'd think an array of 1B spaghetti strands might pose a few difficulties above what sorting those numbers on computers would.

Re: Qudits: The Real Future of Quantum Computing?

#25

Post by adamnemecek »

rdtsc wrote: post #24
adamnemecek wrote: post #22

Earlier quoted context omitted.

> Can it sort names or addresses? Why couldn't it? Ideally they can represent real numbers with infinite precision which are identical to strings. https://en.wikipedia.org/wiki/Real_computation But even in non-real computation this shouldn't be a problem, the same way you can sort strings even though you have only 32/64 bits of memory at a time if that's what you are hinting at. > Another thing with analog computers…

> I don't think that scale has ever been a problem for analog computers. I'd think an array of 1B spaghetti strands might pose a few difficulties above what sorting those numbers on computers would.

I can't tell if you are joking or not. You don't actually think there's 1B spaghetti right?

Re: Qudits: The Real Future of Quantum Computing?

#26

Post by rdtsc »

adamnemecek wrote: post #25
rdtsc wrote: post #24

Earlier quoted context omitted.

> I don't think that scale has ever been a problem for analog computers. I'd think an array of 1B spaghetti strands might pose a few difficulties above what sorting those numbers on computers would.

I can't tell if you are joking or not. You don't actually think there's 1B spaghetti right?

Not actual pasta, no but there would have to be an analog equivalent. How do you see a practical implementation of this working that still preserves the O(n) property

Re: Qudits: The Real Future of Quantum Computing?

#27

Post by babyrainbow »

adamnemecek wrote: post #4

You don't actually want qubits, you want an analog computer with differentiable signals. Most likely photonic. Qubits are a dead evolution branch. I've been recently exploring computational metamaterials for photonic computation. http://users.ece.utexas.edu/~aalu/research%20-%20page%203.ht... (there's quite a few papers on this but unfortunately they are all paywalled. Spoiler alert, they seem to be based entirely on…

Is it yet possible to build the basic logic gates, using light only?

Re: Qudits: The Real Future of Quantum Computing?

#28

Post by akvadrako »

adamnemecek wrote: post #16
akvadrako wrote: post #15

Earlier quoted context omitted.

An ideal 1000 qubit system can explore 10^300 solutions simultaneously. A classical analog computer achieving a similar speedup would be larger than the visible universe (10^30 m) with precision below the planck length (10^-35 m). Unlikely.

They they are orthogonal. Digital/Analog vs Non-quantum/Quantum. You can have quantum analog. Note that analog is a better match since quanta are fundamentally analog and you loose all this info when going to bits. That's the thing, it can solve a lot of computations at the same time however an analog computer can recognize the direction of the solution therefore the number of simultaneous solutions doesn't really ma…

I agree they are orthogonal, but that does nothing to support the point that analog computers could be even more powerful than quantum computers.

You haven't really explained your second point, but assuming it's true, you'll only be able to "recognize the direction" within your margin of error. Your noise floor will be way above the required precision to explore a high-dimensional space.

Analog computers never achieved common use because of noise and lack of practical error correction. Even the human brain evolved to be mostly digital and we almost never focus on being exact.

Re: Qudits: The Real Future of Quantum Computing?

#29

Post by adamnemecek »

akvadrako wrote: post #28
adamnemecek wrote: post #16

Earlier quoted context omitted.

They they are orthogonal. Digital/Analog vs Non-quantum/Quantum. You can have quantum analog. Note that analog is a better match since quanta are fundamentally analog and you loose all this info when going to bits. That's the thing, it can solve a lot of computations at the same time however an analog computer can recognize the direction of the solution therefore the number of simultaneous solutions doesn't really ma…

I agree they are orthogonal, but that does nothing to support the point that analog computers could be even more powerful than quantum computers. You haven't really explained your second point, but assuming it's true, you'll only be able to "recognize the direction" within your margin of error. Your noise floor will be way above the required precision to explore a high-dimensional space. Analog computers never achiev…

Can you like google around please? What I'm saying isn't as unsupported as you seem to believe.

You are talking about electric analog, I'm talking about photonic analog.

How is the brain mostly digital?

Re: Qudits: The Real Future of Quantum Computing?

#30

Post by adamnemecek »

babyrainbow wrote: post #27
adamnemecek wrote: post #4

You don't actually want qubits, you want an analog computer with differentiable signals. Most likely photonic. Qubits are a dead evolution branch. I've been recently exploring computational metamaterials for photonic computation. http://users.ece.utexas.edu/~aalu/research%20-%20page%203.ht... (there's quite a few papers on this but unfortunately they are all paywalled. Spoiler alert, they seem to be based entirely on…

Is it yet possible to build the basic logic gates, using light only?

You don't want logic gates.
Post reply on HN