12 ms·

There are some interesting integration opportunities between high-performance computing centers and on-premise quantum computers. In part it's interesting becau

There are some interesting integration opportunities between high-performance computing centers and on-premise quantum computers. In part it's interesting because in many aspects the quantum computing industry is already somewhat being productized, and there are almost no standards or even common practices, protocols or languages to define the operations and integrations. It's like the 60s again, and we don't even know how wide and flexible the standards should be. Source: working for a European quantum computing startup.

credit_guy 3y ago
Since you are in the know, what exactly is a practical application of quantum computing? 25 years ago I took a class in quantum computing and at the time it looked to be just around the corner, and that look didn't change. I concluded during my class that quantum computing depends on keeping qubits in a delicate balance. Although some quantum computing algorithms (quantum Fourier transform, and by extension Schor's algorithm) offer exponential speedup, adding new qubits to a quantum processor is also an exponentially hard problem, so it's a wash. People then redefine the concept of qubit to be weaker, but then it's not clear how many algorithms still enjoy an exponential speedup. As far as I can tell, they only enjoy a quadratic speedup, which is nothing to sneeze at; but there are generally classical options that offer the same quadratic speedup. I'd love to hear from you that the picture I just painted is wrong, and there are actual concrete application that will work in the next, let's say, 3 years.
- freetonik 3y ago
  The big theoretical use case of quantum computing is simulating quantum systems, with applications in chemistry, biology, medicine/pharmacology, material science and quantum physics. In theory, doing things like calculating the ground state of a complex molecule would be viable on a quantum computer while a classical computer would require enormous amounts of energy, memory and time. Popular science outlets love to talk about "quantum computing breaking encryption" and "instant search with Grover's algorithm", etc., but from what I've seen, barely anyone in modern quantum computing "industry" is really interested in those. Companies like Volkswagen have been interested in QC for its potential in improving EV batteries; agriculture, medicine and pharma are interested in things like protein folding and other simulations with complex molecules and chemical processes, which aren't just viable to a good degree of accuracy on any classical computer nowadays. Another application, which is a bit closer to the popular view of quantum algorithms, is logistics, where potential speed ups in NP-complete problem solving may be useful. (Oh, and there is a so-called post-quantum cryptography with quite fascinating real-world tests)
  - credit_guy 3y ago
    > The big theoretical use case of quantum computing is simulating quantum systems And in practice, how close are we to being able to solve n-body quantum problems on actual quantum computing hardware? Can we do that with D-Wave type of computers, or we have to wait for perfectly entangled qubits (which will never happen)?
    - freetonik 3y ago
      We don't need perfectly entangled qubits, but we do need a large number of physical qubits in order to achieve quantum error correction (and in practice get a smaller amount of perfect logical qubits). We are not very close to that level today, but the progress in the last decade was quite radical.
      - credit_guy 3y ago
        This is really cool, thank you for providing the details. I did a bit of digging, and I think the crucial result that I did not know 25 years ago is the "threshold theorem" [1], which to me sounds like an immensely important result. However, it appears you need quite a bit of noisy cubits, wikipedia mentions 1000 to 10000 for a single logical qubit, but it appears that that's for a level of noise well below what's currently achievable. In any case, quantum n-body problems with tens of entangled electrons are doable with classicla algorithms, so to make an impact you'd need many millions of noisy quantum cubits, maybe even billions. After (admittedly fairly shallow) googling, I found out that IBM plans to have a quantum computer with 100000 noisy cubits ... sometime soon. On the other hand it appears that solid advances happened on the theory side, according to this editorial [2] in the Journal of Chemical Theory and Computation. Still, I can't shake the feeling that it's all just another case of nuclear fusion. When it will happen it will change the world, but that's going to be in 30 years, 30 years from now. [1] https://en.wikipedia.org/wiki/Threshold_theorem https://en.wikipedia.org/wiki/Threshold_theorem [2] https://pubs.acs.org/doi/epdf/10.1021/acs.jctc.3c01043 https://pubs.acs.org/doi/epdf/10.1021/acs.jctc.3c01043
        3y ago

credit_guy 3y ago

Since you are in the know, what exactly is a practical application of quantum computing? 25 years ago I took a class in quantum computing and at the time it looked to be just around the corner, and that look didn't change. I concluded during my class that quantum computing depends on keeping qubits in a delicate balance. Although some quantum computing algorithms (quantum Fourier transform, and by extension Schor's algorithm) offer exponential speedup, adding new qubits to a quantum processor is also an exponentially hard problem, so it's a wash. People then redefine the concept of qubit to be weaker, but then it's not clear how many algorithms still enjoy an exponential speedup. As far as I can tell, they only enjoy a quadratic speedup, which is nothing to sneeze at; but there are generally classical options that offer the same quadratic speedup. I'd love to hear from you that the picture I just painted is wrong, and there are actual concrete application that will work in the next, let's say, 3 years.

freetonik 3y ago
The big theoretical use case of quantum computing is simulating quantum systems, with applications in chemistry, biology, medicine/pharmacology, material science and quantum physics. In theory, doing things like calculating the ground state of a complex molecule would be viable on a quantum computer while a classical computer would require enormous amounts of energy, memory and time. Popular science outlets love to talk about "quantum computing breaking encryption" and "instant search with Grover's algorithm", etc., but from what I've seen, barely anyone in modern quantum computing "industry" is really interested in those. Companies like Volkswagen have been interested in QC for its potential in improving EV batteries; agriculture, medicine and pharma are interested in things like protein folding and other simulations with complex molecules and chemical processes, which aren't just viable to a good degree of accuracy on any classical computer nowadays. Another application, which is a bit closer to the popular view of quantum algorithms, is logistics, where potential speed ups in NP-complete problem solving may be useful. (Oh, and there is a so-called post-quantum cryptography with quite fascinating real-world tests)
- credit_guy 3y ago
  > The big theoretical use case of quantum computing is simulating quantum systems And in practice, how close are we to being able to solve n-body quantum problems on actual quantum computing hardware? Can we do that with D-Wave type of computers, or we have to wait for perfectly entangled qubits (which will never happen)?
  - freetonik 3y ago
    We don't need perfectly entangled qubits, but we do need a large number of physical qubits in order to achieve quantum error correction (and in practice get a smaller amount of perfect logical qubits). We are not very close to that level today, but the progress in the last decade was quite radical.
    - credit_guy 3y ago
      This is really cool, thank you for providing the details. I did a bit of digging, and I think the crucial result that I did not know 25 years ago is the "threshold theorem" [1], which to me sounds like an immensely important result. However, it appears you need quite a bit of noisy cubits, wikipedia mentions 1000 to 10000 for a single logical qubit, but it appears that that's for a level of noise well below what's currently achievable. In any case, quantum n-body problems with tens of entangled electrons are doable with classicla algorithms, so to make an impact you'd need many millions of noisy quantum cubits, maybe even billions. After (admittedly fairly shallow) googling, I found out that IBM plans to have a quantum computer with 100000 noisy cubits ... sometime soon. On the other hand it appears that solid advances happened on the theory side, according to this editorial [2] in the Journal of Chemical Theory and Computation. Still, I can't shake the feeling that it's all just another case of nuclear fusion. When it will happen it will change the world, but that's going to be in 30 years, 30 years from now. [1] https://en.wikipedia.org/wiki/Threshold_theorem https://en.wikipedia.org/wiki/Threshold_theorem [2] https://pubs.acs.org/doi/epdf/10.1021/acs.jctc.3c01043 https://pubs.acs.org/doi/epdf/10.1021/acs.jctc.3c01043
      - 3y ago