14 ms·
Avoiding fusion plasma tearing instability with deep reinforcement learning
- seydor 3y agoHow about we add transformers to the reactor? But not that kind of transformer
- baq 3y agoTurns out paying attention to plasma is all you need.
- api 3y agoI remember having this idea when I was studying machine learning in college. I'm really happy to see that it occurred to someone else in a position to actually look into it, because it "felt like something might be there" to me. The basic idea I had was that fusion plasma containment involves containing a turbulent, dynamical system, so it might require some kind of actual intelligence learning or co-evolving with the system. I wondered if this might be the only way to achieve over-unity fusion outside gravitational confinement (stars, black hole accretion disks, etc.). This would mean there are two fusion mechanisms in nature: gravitational confinement and cognitive confinement. The latter can only be a product of a living system. When a living system achieves this, its biosphere "ignites" and becomes something I termed a "biostar." Biostars could be potential SETI targets -- biospheres that have harnessed fusion and so emit anomalous amounts of optical and infrared radiation on their night side. This moment of ignition would be an event in a biosphere comparable to the evolution of photosynthesis-- a fundamental change in the energetic dynamics of life. In the far future life the that achieved fusion could settle things like rogue planets in deep space, so that would be another potential SETI target. Find objects emitting anomalous infrared in the interstellar void. The advantage would be being far from destructive events like solar storms.
- baq 3y agoAssuming fusion drives are a thing we should be able to see them as they'd be glowing rather bright. https://www.projectrho.com/public_html/rocket/torchships.php https://www.projectrho.com/public_html/rocket/torchships.php
- gs17 3y agoUnless, of course, all the ships are coming straight towards us.
- tsimionescu 3y agoFusion drives wouldn't allow these ships to get anywhere near the speed of light, so we would have ample time to see the light they emit long before they arrive.
- piotrkaminski 3y agoThat would actually make them easier to spot, as they'd need to flip around for a breaking burn roughly equal to their acceleration burn, pointing their engines straight at us. (Assuming they want to stop by and say hi, of course. If not, then there's not much to worry about.)
- gpderetta 3y ago> not much to worry about. Unless they are literally coming straight ahead at us!
- piotrkaminski 3y agoIf they intend to go straight through Earth at relativistic speeds (to establish a hyperspace bypass, perhaps?), then there's really not much we can do about it anyway. :p
- p1mrx 3y agoYou could lie down, or put a paper bag over your head.
- jetbooster 3y agoIn Peter F Hamilton's Commonwealth series of books, true artificial general intelligence 'woke up' from the computers designed to handle the incredibly complex calculations required to make and maintain long-distance wormhole connections. Quite analogous to this situation I feel
- Sabinus 3y agoAnd then formed a treaty with humans, providing Restricted Intelligences for high performance computing without self awareness creation, and then retreating to it's own planet. What a series, I really need to read them again.
- SubiculumCode 3y agoSounds like the basis of a new Star Trek episode. Captain's Log: Since we came to orbit Venuuil III to host talks between the Klingons and the Venullians, there has been a increased incidence of unpredictable fluctuations of plasmas in the warp drive containment field. We are now devoting all available power to increasing our computer's ability to track and predict this rapidly changing phenomenon. Geordi reports that at the current growth factor we can maintain containment for 22 hours, 47 minutes, 17 seconds. To support the peace talks, we will remain here as long as possible. Captain's Log: Intriguingly, the fluctuations are beginning to reveal an embedded temporal distortion that exhibits language like patterns. Data has begun working on an interface between the containment field and his positronic neural net.
- tekla 3y ago> This would mean there are two fusion mechanisms in nature: gravitational confinement and cognitive confinement Da fuck? The mechanism is compression. This is like calling "generating electricity via a generator", "cognitive energy". What is ITER? Black Magic that happens to cause Fusion?
- api 3y agoFusion is reasonably easy. You can do it in your garage with an electrostatic confinement fusor. I’m talking about fusion that generates significantly more power than it takes to run the reactor. Only that kind is useful as an energy source. That so far has been elusive. Inertial confinement has sort of achieved this but only on paper. If you tally up the total input to set up and run the system it’s still way in the red. ITER has the potential to run just a bit over unity but it’s really just a research platform.
- londons_explore 3y agoWhile deep learning might be a good control strategy for such systems, I very much doubt it is the only control strategy that works. Control strategies in nonlinear systems are an effectively huge search space, and deep learning is just one way to find a good-ish solution faster.
- AtlasBarfed 3y agoExcept that fusion on earth will likely never be cheaper than solar/wind. I mean that is a cool scifi story, but economics seems to hate cool things. There's this "big lie" that fusion people imply that it will be cheap, clean, and limitless. Cheap is doubtful, clean is undermined by the reality that fast neutrons from fusion degrade the reactor to radioactive isotopes, and ok the fuel is pretty much limitless Now, if we can get scalable fusion as viable load levelling, to develop it to the point it can be used in space then that's some real scifi.
- thechao 3y agoFusion will be necessary if we plan to colonize the solar system beyond Mars, and not at Jupiter. It's cold, dark, and scary, out there.
- pfdietz 3y agoPower can be beamed out to interstellar distances, so fusion isn't necessary. For that matter, if a space colony is equipped with a mirror for concentrating the sunlight needed to illuminate the inside as if it were Earth, and we place the limiting distance as that at which the mass of the mirror is equal to the mass of the space colony, the distance is about 1 light year.
- tsimionescu 3y agoWe won't, not with anything resembling current technology. So, if we were to imagine a colonized solar system, there is a good chance it's not fusion that gets us there, but some currently unknown technology.
- AtlasBarfed 3y agoJeez, they invented the core tech in the 60s. An orion pulse nuclear ship can get a small city to Jupiter in a month. You don't need a wonder technology to cart around the solar system, well, unless you are talking ECONOMICAL technology. I'm not sure how expensive H-bombs are to make at scale.
- tsimionescu 3y agoCurrently imaginable fusion power plants generate nowhere near enough power for the excess to be visible from outer space. They would not even be a blip compared to the largest already existing hydro power plant, for example. Edit: to add some numbers, the "planned" DEMO power plant (the hypothetical successor of a successful ITER experiment) would produce something like 750MW, while Three Gorges Dam produces 22,500MW. Even if DEMO could be scaled up (which is hard, given that it would already be beyond the limits of today's material science), it definitely couldn't scale up 30 times.
- leereeves 3y agoWhat stops fusion from scaling up by building more plants instead of larger plants?
- tsimionescu 3y agoNothing except the costs, but then it's not a single reactor anymore and other power plants scale as or more easily. So nothing is really special about fusion power if we just want to scale horizontally.
- leereeves 3y agoI thought fusion is special because the inputs are more abundant and the outputs less toxic than other fuel power plants.
- tsimionescu 3y agoFusion can be compared like this: - vs fossil fuels, almost limitless fuel with no greenhouse gas emissions - vs nuclear fission, more fuel compared to current designs (though breeder reactors could essentially use any piece of rock as fuel), and shorter lived but even more toxic outputs (at least for D-T fusion, all pieces of the reactor become highly radioactive materials after 10-20 years) - vs renewables, it has the advantage of being decoupled from weather and day/night patterns, but it is more expensive, it requires fuel, and it produces much more toxic outputs
- eganist 3y agoThe concept of "cognitive confinement" (your term I presume. It's neat, I hope it sticks) was explored in Spiderman 2. I'm being entirely serious. Even though the visualization of the reactor was wildly off the mark, this specific concept fits what the writers had in mind with the AI mechanical limbs.
- fancyfredbot 3y agoDeepmind worked on reinforcement learning for plasma control back in 2022 and this also led to a paper in nature. I don't really understand the differences between their earlier work and this paper but deepmind don't seem to be involved in this one: https://deepmind.google/discover/blog/accelerating-fusion-science-through-learned-plasma-control/ https://deepmind.google/discover/blog/accelerating-fusion-sc...
- jeffbee 3y agoThe DeepMind paper in turn cites the authors of this paper, previously. One of the big differences in the current paper is that the experimental device is much larger and more powerful, and the duration of the shot is longer as well.
- londons_explore 3y ago> nuclear fusion is rapidly emerging And has been rapidly emerging for 60 years...
- aqme28 3y agoSame with AI, and yet they both keep getting closer
- klipt 3y agoIs it economically competitive with solar though?
- exe34 3y agoAt night it is.
- pfdietz 3y agoNot with LFP cells falling to $54/kWh.
- jeffbee 3y agoSolar power is gravitational confinement fusion.
- aqme28 3y agoIs it? Definitely not. Could it be? Maybe. Who knows
- londons_explore 3y agoI kinda get the feeling there are a core of scientists who want to work on this stuff. They'll work on it even if there is no hope of commercial fusion power, simply for academic kudos. Government's will fund it simply because there isn't much other blue skies physics research to fund.
- ben_w 3y ago
- SkyMarshal 3y agoAnother team from Japan and the US published a paper on this recently too: https://www.nature.com/articles/s41598-023-49432-3 https://www.nature.com/articles/s41598-023-49432-3 Sabine has a good review of it: https://www.youtube.com/watch?v=4VD_DLPQJBU https://www.youtube.com/watch?v=4VD_DLPQJBU
- devaiops9001 3y agoTokamak is dumb. I'm sick of hearing about Tokamak. The plasma in the Safire reactor has self-containing magnetic fields and doesn't need the $20billion+ super-magnet infrastructure. A Safire reactor costs under $20-million to build, and probably much less these days. The Safire reactor can keep the plasma lit and going for hours if not days without interruption. The Safire reactor has been around for over seven years now.
- aunty_helen 3y agoThere was a front page hn article this week about institutions keeping the reason for their creation going. ITER has always felt this way for me.
- pfdietz 3y agoIt's not a coincidence that the ITER project will be about one career long.
- devaiops9001 3y agoTokamak would fall in this category too.
- dralley 3y agoITER has the same problem as the NIF, had they been designed a few years later it could be a fraction of the size and cost due to improvements in technology But since ITER was designed decades ago, we're stuck with a massive, expensive, outdated beast that's taken so long to build it will likely end up being lapped by other projects. At least NIF has something to show for it. ITER feels like building the Vasa.
- pfdietz 3y agoITER never made sense at all, even without improvements in fusion technology. And the improvements have not made tokamaks sensible. Even higher field magnets don't rescue the tokamak concept from practical irrelevancy.
- smaddox 3y agoTokamaks are 1960's technology. The future of economical fusion appears much more likely to be based on the field-reversed configuration (FRC). Helion expects to produce net positive energy production from a reactor designed primarily for He3 production in the next few years: https://www.helionenergy.com/ https://www.helionenergy.com/
- raverbashing 3y agoRemains to be seen. Helion would have to ship something that gives positive net energy first. Sure, they have prototypes, but we don't know how well they are currently performing
- smaddox 3y agoAgree, that will be the proof. But tokomaks have yet to do the same despite decades of investment. We know that helion is able to recover 95% of the energy of every pulse. And they've measured the scaling laws. There doesn't seem to be anything that will prevent net positive energy in the reactor they're building right now.
- jszymborski 3y agoI'm not sure the date of invention is relevant. The wheel is stone-age tech. What matters is how we build Tokamak's has changed. A huge notable difference are the magnets used for example.
- smaddox 3y agoIt's relevant when it doesn't work after all that time. As I argue further down the thread, the trajectory of FRCs looks much more promising.
- theptip 3y agoThis seems like spurious reasoning. It’s like saying “rockets are 1950s technology therefore SpaceX is going to fail”. Commonwealth are using cutting-edge high-temp superconductors which can generate much stronger fields: https://cfs.energy/ https://cfs.energy/ The idea is old, the tech is new.
- Muromec 3y ago“Tell me comrade, how doez a tokamak reactor explode?”
- Agingcoder 3y agoI’m not sure I’m ready to trust an ml system that will control a fusion power plant. The potential of the the mistake making a ( bizarre, ml-like ) mistake seems very high to me.
- chrisgp 3y agoWhy doesn't simpler control theory work here, like PID controllers?
- itishappy 3y ago> ... Calculating tearing stability requires massive computational simulations based on resistive magnetohydrodynamics or gyrokinetics, which are not suitable for real-time stability prediction and control during experiments. This suggests the need for AI-accelerated real-time instability-avoidance techniques.
- theptip 3y agoPID controllers are for linear or mildly-nonlinear systems. Plasma fluid dynamics are about as nonlinear as it gets.