Just how much will nuclear fusion and quantum computing change the world? I’ve heard them hyped as game changers that will shake the world up. Just how will they really change the world? No hype, what will their impact really be?
I’ll give the TL;DR version before someone jumps in with a more comprehensive answer.
Nuclear fusion is in theory unlimited energy with little to no dangerous waste. This wouldn’t just be a drop in your energy bills, it could make a lot of projects feasible that are not right now.
However, in practice, since we haven’t actually implemented this form of energy production yet, we don’t know what limitations or drawbacks for engineering reasons may prevent it being this perfect panacea.
Quantum computing doesn’t actually have that many use cases right now – in the short term, it will break encryption as we know it, and probably require us all to use some sort of quantum computing based service or hardware instead.
It will also be useful for modelling real world quantum phenomena.
Beyond that, it will certainly have many uses, but those will become clear over time, rather than things that seem essential to us right now.
Also, it’s not going to replace classical silicon computers.
Also, if we’re talking technologies likely to change the world dramatically, there are many things like deep learning, CRISPR, graphene (and other carbon allotropes), 3D printing etc that I would consider way more likely than fusion or quantum computers.
If we assume nuclear fusion to mean the success of the current efforts - that is, much the same technology, just refined and scaled to the point of economic viability, the answer is: not all that much. They won’t be producing electricity “too cheap to meter” and they won’t be free of significant issues managing radioactive materials. There is no expectation of sudden new insights to fix these problems. The cost of building a power plant will be enormous, and the cost of energy produced may never make it worthwhile against other production methods - even notionally non-polluting ones.
Quantum computers are a hard call. They can’t and won’t compete with conventional computers. They solve totally different problems. If someone built one that could factor huge primes, it would cause some pain and a flurry of effort to avoid the total breakdown of critical encrypted communications. But they won’t wipe out encryption. Just make it harder. The ability to solve certain classes of intractable or very difficult new problems will open up new areas of research. Just which areas is hard to predict. Just working out how how to program and make good use of them will take very significant effort. There is little chance there will be an effect on society commensurate with the way digital computers in silicon have. Not in any sort of foreseeable future.
I don’t see either technology as game changers.
Making any computer that can viably solve protein folding would be game changing. Maybe a quantum computer could be convinced to do that. But it is the game changing medical and biological technologies resulting that are the actual game changers, not the computer.
A 2018 National Academies report found no near-term, commercially viable applications for QC.
That said, the hype in the chemistry community is that it’ll eventually enable faster and more accurate modeling* over current, digital methods (e.g. DFT).
I can’t speak much more to this, having minimal knowledge of QC, and knowing just enough DFT to be dangerous (getting Gaussian to output numbers isn’t the hard part.) But I can dig up some likely paywalled articles from C&EN (the weekly chemistry rag) if people are into that.
*This is useful for catalyst design, drug discovery, improving materials properties, etc.
I would quibble just a little with that last line. In the modern world, most new technologies require a coming together of breakthroughs in different fields (and often one or more of those breakthroughs will be decades old, so some people will write off the whole thing as not being genuinely new).
If we do protein-folding on QC first, then that’s a gold star for QC in my book. No-one will care how the killer app came to be.
Yeah but paradoxically, that’s often the sign of a real game changing technology: one where we can’t even imagine the uses right now because it’s a whole new world to us.
When the iphone was released it was ridiculed on a number of comedy shows as being utterly pointless (yes, I know some Dopers will still say that it is, but this is a really old-fashioned community ).
Not to say the same will happen with QC, just to say that it’s pretty hard to predict this kind of thing.
IME the NAs are a bit fuddy about up-and-coming tech, but I’ve only interacted with them off and on.
funny you should ask…
Right now, an interesting hypothetical application for quantum computing is nuclear fusion.
One of the really major drivers of super-computer development was analysis of nuclear fission. But that was because we already understood fission: we just needed to do a lot of calculations to predict how it would work, then compare results to refine the process, and it’s ok if it takes a couple of months to run the simulation.
Unfortunately, we don’t understand fusion containment that well. They build expensive fusion experiments, but they’ve never been able to scale up in the way they hope. And top-level super-computers just haven’t been able to handle the scaled-up simulations of minutes of contained fusion cheap enough and fast enough to just do millions of simulations just to see what happens when you try different things.
Which is where quantum computing comes in. It looks like quantum computing might be really good at running fusion simulations involving more that a few atoms for more that a few fusions.
I don’t know when this kind of simulation becomes practical, and nobody knows what this kind of simulation might find. But it is a problem looking for a solution, and a solution looking for a problem, and if quantum computing enables contained nuclear fusion, it will be really big news.
Quantum computing exists right now, just only on a very small scale (i.e., small enough that it’s still completely impractical). No quantum computer has yet been built that can solve any problem faster (in terms of real time) than a conventional computer. That will probably eventually change: In principle, quantum computers can solve factoring problems, for instance, much faster than conventional computers, if we can just scale them up. Which is significant, because the most commonly-used encryption algorithm is based on factorization, and requires that it be difficult. In other words, if some tinkerer managed to build a large (thousands of qbits) quantum computer in his garage right now, then he would be like unto a god, in the digital world.
But that’s probably not the way it’s going to shake out. It’s probably not going to be some tinkerer who suddenly creates a full-scale computer; it’s much more likely to be a gradual increase in capacity that everyone can see coming. Which means that people will know that they need to switch over to other forms of encryption, and will have the time to do it. And those other forms of encryption already exist, and are just as hard to crack on quantum computers as on classical computers (harder, actually, because quantum computing itself is hard). Eventually, when everyone has access to quantum computers, RSA will be just as fragile as Enigma is now, and just as irrelevant.
Practical, commercially viable fusion? Sky is literally the limit on how game changing it would be, once it’s in the pipeline. Of course, that’s the rub, but basically you’d have unlimited energy that would not only profoundly impact life on the Earth but it would make space exploration and exploitation achievable and practical, even interstellar travel possible.
As for quantum computing, that’s a bit less understood. In theory, you could solve problems that currently take a lot of conventional computing power very difficult or even impossible fairly easy, though it’s a matter of strengths and weaknesses, as some things conventional computers could and would do better, some things quantum computers better. I don’t think it would or will be really world changing, but then, we are still in such early days that it’s hard to gauge the impact. Personally, I think strong or general AI will have a bigger and more profound impact, and you don’t need quantum computers for that.
Google claimed to have achieved quantum supremacy with their 54-cubit Sycamore processor last year, but I understand there was some disagreement about the claim. And I don’t pretend to understand paper:
10.1038/s41586-019-1666-5
A very long time ago, I was told that nuclear reactors would make electricity too cheap to bill.
Is this another one of those?
So what’s a cubit? What does it look like? Is there a graphic novel “Cubits For Dummies”? I might not understand the mystical processes but I could look at the pictures.
IBM’s caveat, as I understand it, is that a classical computer with a lot of disk storage could generate and use lookup tables to speed up the calculation to the point where it takes days instead of millenia to do the classical calculation. Of course days is still longer than the minutes the quantum computer took.
Nuclear fusion looks like a big dud as far as changing the world. If someone figures out a way to do it cheaply, easily, with naturally occurring fuels, and safely with no radioactive waste and no easy way to use it to make weapons-grade fission materials, that would be great, but so far it looks like that is entirely science fictional. An improved version of what’s currently in the works would be a good new power source, but not revolutionary, and would probably be at least as tightly controlled as fission reactors. This article goes over some of the problems with real-world fusion:
Lots of luck in factoring primes. But factoring very large numbers would break many trapdoor codes. Whether it would break codes based on abelian varieties is something I have not been able to find out.
OHhhh… it’s Qbit!
So, where is the advantage? Is it just that electron spin or Photon phase provide very fast components for numerical computers? Or is there some Quantum magic that pops out primes?
Yeah sorry, I wrote cubit instead of qubit.
It’s not so much that fiddling with the bits is faster, it’s that superposition and entanglement enable different kinds of calculations that traditionally would take a long time. Per (*):
To properly model these chemical systems, a computer must calculate the positions and energy levels of electrons. These properties are described by mathematical functions called orbitals. And that’s why quantum mechanics is key.
“If you have 125 orbitals and you want to store all possible configurations, then you need more memory in your classical computer than there are atoms in the universe,” says Matthias Troyer, who develops algorithms and applications for quantum computers at Microsoft Research in Zurich. But a quantum computer could model such a system with just 250 qubits.
The difference comes down to how each type of computer represents the states of electrons. Conventional computers use ones and zeros to encode an abstract description of all the possible states of an electron. Meanwhile, the qubits in quantum computers can represent these states as they exist in the molecular systems—as superpositions.
*https://cen.acs.org/articles/95/i43/Chemistry-quantum-computings-killer-app.html
It’s strictly speaking not known what exactly provides the quantum speedup—indeed, it’s strictly speaking not known that there is a quantum speedup: it could be that there are classical algorithms that work as fast as the quantum counterparts, which however nobody has found yet. However, virtually nobody believes this to be the case, and for good reasons.
So, the safe bet is probably that yes, quantum computers do provide a systematic speedup for certain problems. First, it’s important to know what’s meant by ‘speedup’ here: without going into the, uh, complexities of complexity theory, you can in general measure how efficient an algorithm is in doing its job by working out how it performs on different-sized inputs. Any computer algorithm will quickly multiply small numbers, and take longer the bigger the numbers are—how, exactly, that time increases with input size is taken to be a measure of the efficiency of the algorithm: a more efficient algorithm will take systematically less time to solve problems of the same size.
Typically, you consider some input of size n (say, n bits in length), and the algorithm will take a time that’s some function of n to complete. That might be some power of n, say n^2 (I haven’t yet figured out if there’s any sort of math formatting, even rudimentary, that can be done with the new board software), it might grow as the square root of n, or it might even increase exponentially with n.
The quantum speedup then means that there are some problems for which there exist algorithms that can be implemented on a quantum computer that confer a systematic speedup—their execution time growing much slower with increasing n—as compared to any classical algorithm. That is, we don’t simply do things much faster on a quantum computer, but the way the time taken increases with increasing problem instance size differs—it’s a qualitative difference, not merely a quantitative one, as when you essentially perform the same calculation on a much faster computer. That, and the fact that some of these problems for which a speedup exist are highly relevant—like prime factoring—is what makes them so potentially valuable.
Thus, there’s some real quantum magic that’s needed for the quantum computer to outperform classical computers. What, exactly, provides this quantum magic, however, isn’t clear as of yet. Often, you’ll read in the popular press something about how qubits can be ‘both one and zero at the same time’, or that quantum computers ‘perform every computation simultaneously’, but, while we don’t know what provides the magic, we know that that’s not it. If it were, after all, you’d have computed every possible output—but then, how would you tell which one’s the right one? How would you get the machine to spit out that one, but not any other?
Typically, while a quantum computation does make use of quantum superposition—which isn’t really ‘many possibilities at once’, but also isn’t quite not that—the result will be encoded in the properties of the total state afterwards, and not in any of the superposed components.
So why can quantum computers do things that outstrip the capacities of classical machines? Well, as I said, nobody really knows. Various quantum properties have been proposed as candidates. Superposition, as I said, plays a role, and so does interference. Entanglement used to be a hot candidate, but it’s since been shown that a quantum speedup can be achieved with arbitrarily little entanglement. Afterwards, a more abstract quantity known as ‘quantum discord’—essentially, the difference between two ways of calculating the (relative) information in a system that agree in classical settings, but may disagree in quantum mechanics—was proposed, but that’s such an ubiquitous property that it’s not clear if one is actually saying much thereby.
Perhaps the best intuition that quantum computers should be able to do certain things effectively that are hard for classical computers comes from the fact that it’s extremely hard to simulate quantum mechanics on a classical device. The reason for this is, ultimately, that a quantum state is a very complex object: you’ve got 2^n parameters you need to keep track of, for an n-qubit state. This will be very costly, computationally—hence, any classical computer will have trouble simulating even simple quantum systems. But what we can do is to just take the quantum systems themselves and have them ‘compute’ their own time evolution—that, after all, is just what they do. Hence, it’ll be a breeze to them!
So, as far as intuition goes, quantum computers outperform classical computers, at least in some tasks (for one, simulating quantum systems), because simulating quantum systems on classical computers is a very hard problem, that’s however easy if you use quantum systems from the outset.
I want to point out (like the article I linked above says) that this ‘in theory’ is completely abstract theory and very far from anything anyone is working on or thinks is remotely likely to work in a realistic timeframe - none of the methods of fusion being tested provide ‘unlimited energy’ with ‘little to no dangerous waste’. The best reactors require tritium, which we currently only get in significant quantities from fission reactors (it’s radioactive and has too short of a half-life for natural deposits to hang around). Deuterium-only fusion reactions are about twenty times as difficult as Deuterium-Tritium and no one has come remotely close to breakeven with non-tritium fusion. Protium-Protium (regular one-proton hydrogen) reactions are completely impractical with current theory and technology.
On the waste front, most (80%) of the energy from fusion is released in the form of neutrons, which ends up turning the superstructure of the vessel and any material used to collect heat into radioactive waste. There’s some hope of processes that will make it ‘short lived’ radioactive waste, but that’s still dangerous waste to deal with. Also, because actual fusion reactors release so much of their energy via neutrons, you can trivially use a running reactor to enrich regular uranium into bomb-grade plutonium just by putting it in the reaction chamber. This means that they’d have to be as tightly controlled as breeder fission reactors unless the major powers in the world change their position on nuclear proliferation.
None of the current reactors would do this if they progressed to being practical and commercially viable. They are reactors not space drives so don’t really do much for space travel, they are large devices that sit in place and generate heat, they don’t generate thrust and aren’t especially suited to mobile applications. Also, as a drive they would be very limited because they require a huge amount of power to start up (largely to heat the plasma enough for fusion to take place) - you’d either have to run them continuously or have a second power source to be able to turn them on. And fusion power is less ‘unlimited’ than fission power is - as I pointed out above, D-T reactors require a fission reactor to make the ‘T’ in their fuel. If you can make deuterium-only fission work, then it’s about as ‘unlimited’ as fission is, since it’s not especially difficult to extract the uranium dissolved in seawater for fission fuel.
That’s why I put in the caveat. Also, I wasn’t thinking so much of star ship drives as energy for large scale habitats or colonies. Like I said, IF we had ‘Practical, commercially viable fusion’ then the sky is the limit, and since the OP is asking for that, not how close we are or aren’t to having it, that’s what I answered.