Short explanation:
Qubits encode information needed for quantum computers. Noise from the environment can destroy that information. By introducing a driving field to the qubit they made the qubit+driving field the new qubit ("dressed" qubit) and it was more robust to noise.
It also gives some different levers to change the information (namely, the driving field).
Lots of encryption is based on the prime factorisation being hard to do. Specifically, that it scales significantly as you slightly increase the key size.
There's a quantum algorithm (Shor's algorithm) that doesn't scale quite so badly as the numbers get bigger. That means that with a fast quantum computer you could solve this specific problem much much more quickly than is possible on a classical computer.
However, other crypto types (elliptical curve cryptography EDIT - ECC is not a type that meets these criteria, see the responses to me below) doesn't depend on prime factorisation, but other things. Those other things have no known nice fast quantum algo. I'm not sure where this sits on "there's provably no fast quantum algorithm" and "we don't currently know of one" however (EDIT - ECC is in the "there is definitely a fast quantum algorithm" category!).
Most generally, quantum computers are not just fast regular computers. For some (not all) problems, they scale better for solving the problem. So for example, finding an item in an unordered list. If you have 100 items, you need to on average check 50 items to find it on a regular computer, and if you have a million items you need to check 500,000 items. A quantum computer can run 10 iterations to solve find something out of 100, but just 1000 to find something in a million. Other differences scale better or worse.
Quick simple overview: Some problems that we thought were intractable turn out to be quite possible on quantum computers. But not every problem.
I tried to keep this simple as my understanding is also quite simple, and I don't want to post things that are wrong.
> However, other crypto types (elliptical curve cryptography) doesn't depend on prime factorisation, but other things. Those other things have no known nice fast quantum algo.
For your clarification, Shor's algorithm just involves encoding every possible value of a function (say f(x) for all values of x) into a bunch of qubits, taking the fourier transform, and measuring, which will tell you the period of that function (some r such that f(x + r) = f(x) for all x). Thus, you can find the period of any function that you can define a fourier transform for. For breaking RSA, your function is on the domain of integers modulo n, so you use the discrete fourier transform. You can also define a fourier transform for functions on the domain of an elliptic curve as well! In fact, any finite abelian group has a unique definition of fourier transform (see https://en.wikipedia.org/wiki/Pontryagin_duality for some of the math behind it). That means we can't just upgrade our crypto to some more complex curve space, because all of them will be vulnerable to Shor's algorithm.
The mathematics behind Shor's algorithm is actually a really interesting read if you enjoy pure, abstract algebra and topology.
We know how to do a lot of number theory stuff quickly on quantum computers that we don't know how to do quickly on classical computers. Some of those things underpin all asymmetric key algorithms in use today. We have to develop asymmetric key algorithms that don't use things we know how to compute quickly on a quantum computer.
ECC (including Curve25519) is not secure in a quantum computing world, Shor's algorithm solves discrete log efficiently.
I am not a deep expert in this field but there is a lot of work in post-quantum crypto that derives security from the hardness of some lattice-based problems instead of discrete log (E.g. New Hope being used in an experiment in Chrome is based on RLWE, I think).
Just to highlight the presence of mistaken information below (which others have already corrected further down in the thread):
Traditional public key cryptography is based on modular exponentiation and related number theoretic operations. Next-generation public key cryptography is based on elliptic curve operations. Both of these are highly vulnerable to quantum computers. The attempt to get people to adopt ECC will not protect well against quantum computers, which is one interpretation of why NSA is now recommending against putting resources into that transition.
There are other ideas for how to do public key cryptography that would (edit: seemingly, thanks n4r9) not be vulnerable to attacks from quantum computers. These techniques are called "post-quantum", and they're quite distinct from the elliptic curve techniques. A nice benefit of elliptic curve algorithms is that they normally have very small keys, compared to earlier schemes like RSA and Diffie-Hellman, while a disadvantage of post-quantum approaches is that they generally have extremely large keys. But that might become necessary for security, unfortunately.
It's important to note that post-quantum cryptographic techniques are generally ones for which we have not yet found ways of breaking with quantum algorithms.
Surely you jest? For AI (QC or not) to bring such a result, even with perfect AI, would require delegating governmental controls on a wide scale. Hard research requires hardware, people, money, etc. So does your AI first solve the problem of how governments make decisions? And how do you conclude that an AI would decide that researching fusion is the best choice?
Energy. Fossil fuels are finite and are getting expensive to extract. Fission is a good stop-gap, but again finite. Breeder reactors are supposed to help with that, but they don't exist yet. Renewable options are limited to geothermal and hydroelectric unless you want to invest in a lot of high capacity batteries.
AIs will be dependent on electrical energy for survival, what else would they care about?
with a proper algorithm which may well not exist it's in theory possible to e.g. perform deep learning on all training examples in a single pass or something like that.
While the OP is wrong about being able to "do everything in parallel" via quantum superposition (we have plenty of no-go theorems when it comes to quantum computers being faster simply due to parallelization) some interesting algorithms that are relevant to data science in general have been developed. For example, this[1] one provides an exponential speedup (over known classical algorithms) for calculating persistent homology, a useful method for identifying scale-independent patterns.
Likely huge, I'm already aware of two QC companies in Canada who are staffing up to look at building a quantum cloud and offering quantum compute workloads "in the cloud" - so to speak.
Fundamental research takes time. One cannot discard such research as "all talk". It is the basis on which better results will be achieved and the possibility of a useful computer will be realized.
I am fascinated by the ideas to produce quantum computers... I want to know all the iterative improvements as they come... there is joy in the journey.
While this is impressive compared to earlier systems, just citing the T1 & T2 values does not actually tell you much about the usability of the system as a qubit. What really counts is the number of operations you can realize within the coherence time of the qubit, and as far as I can see from the paper this seems to be around 5-10 (which is two to three orders of magnitude below what e.g superconducting qubits currently offer). Also, the next step would be to show two-qubit operations with the system, which is another challenge that isn't easy to overcome. So only after increasing the coherence time by another two to three orders of magnitude and realizing a robust coupling mechanism would it make sense to think about scaling the system and building an actual quantum computer with it. Still, semiconductor qubits offer several interesting possibilities as Silicon technology is something that is significantly more robust and scalable than e.g. superconducting circuit technology (currently at least).
43 comments
[ 3.3 ms ] story [ 95.1 ms ] threadShort explanation: Qubits encode information needed for quantum computers. Noise from the environment can destroy that information. By introducing a driving field to the qubit they made the qubit+driving field the new qubit ("dressed" qubit) and it was more robust to noise.
It also gives some different levers to change the information (namely, the driving field).
It's what we believe to be quantum computer secure.
There's a quantum algorithm (Shor's algorithm) that doesn't scale quite so badly as the numbers get bigger. That means that with a fast quantum computer you could solve this specific problem much much more quickly than is possible on a classical computer.
However, other crypto types (elliptical curve cryptography EDIT - ECC is not a type that meets these criteria, see the responses to me below) doesn't depend on prime factorisation, but other things. Those other things have no known nice fast quantum algo. I'm not sure where this sits on "there's provably no fast quantum algorithm" and "we don't currently know of one" however (EDIT - ECC is in the "there is definitely a fast quantum algorithm" category!).
Most generally, quantum computers are not just fast regular computers. For some (not all) problems, they scale better for solving the problem. So for example, finding an item in an unordered list. If you have 100 items, you need to on average check 50 items to find it on a regular computer, and if you have a million items you need to check 500,000 items. A quantum computer can run 10 iterations to solve find something out of 100, but just 1000 to find something in a million. Other differences scale better or worse.
Quick simple overview: Some problems that we thought were intractable turn out to be quite possible on quantum computers. But not every problem.
I tried to keep this simple as my understanding is also quite simple, and I don't want to post things that are wrong.
This is wrong. https://en.m.wikipedia.org/wiki/Elliptic_curve_cryptography#...
The mathematics behind Shor's algorithm is actually a really interesting read if you enjoy pure, abstract algebra and topology.
I am not a deep expert in this field but there is a lot of work in post-quantum crypto that derives security from the hardness of some lattice-based problems instead of discrete log (E.g. New Hope being used in an experiment in Chrome is based on RLWE, I think).
https://pqcrypto.org/
(down right now, so here's the cache: https://webcache.googleusercontent.com/search?q=cache:1JkIi2...
)
Traditional public key cryptography is based on modular exponentiation and related number theoretic operations. Next-generation public key cryptography is based on elliptic curve operations. Both of these are highly vulnerable to quantum computers. The attempt to get people to adopt ECC will not protect well against quantum computers, which is one interpretation of why NSA is now recommending against putting resources into that transition.
There are other ideas for how to do public key cryptography that would (edit: seemingly, thanks n4r9) not be vulnerable to attacks from quantum computers. These techniques are called "post-quantum", and they're quite distinct from the elliptic curve techniques. A nice benefit of elliptic curve algorithms is that they normally have very small keys, compared to earlier schemes like RSA and Diffie-Hellman, while a disadvantage of post-quantum approaches is that they generally have extremely large keys. But that might become necessary for security, unfortunately.
AIs will be dependent on electrical energy for survival, what else would they care about?
Buzzword1+Buzzword2 = interesting.
Why? What part? What application?
It's a bit like saying that QC will be awesome because thanks to it people could in theory find the cure for cancer, in some way, somehow.
[1]: https://arxiv.org/abs/1408.3106
Feel free to ignore the stories until it reaches whatever point interests you, but don't downplay the work others do to get you there.