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

AI for real-time fusion plasma behavior prediction and manipulation

control.princeton.edu

Post reply on HN
Print view View original

121–130 of 167 posts

Re: AI for real-time fusion plasma behavior prediction and manipulation

#121

Post by LargoLasskhyfv »

throwup238 wrote: post #102
Borg3 wrote: post #58

Earlier quoted context omitted.

Thats least of your problem imo. Neutron corrosion is bigger problem. There is trick to use Lithium shielding, with create Tritium needed for Fussion. But not sure how effective it is, especially for long term reactor lifetime. Those reactors are very expensive, not sure if its worth to shut it down every year and replace entire Li shielding...

I think beryllium is a better candidate. It can be grown as a single crystal and there’s lots of research into using it for shielding in nuclear lightbulb reactors.

Who is researching nuclear lightbulb reactors?

Re: AI for real-time fusion plasma behavior prediction and manipulation

#122

Post by defrost »

LargoLasskhyfv wrote: post #121
throwup238 wrote: post #102

Earlier quoted context omitted.

I think beryllium is a better candidate. It can be grown as a single crystal and there’s lots of research into using it for shielding in nuclear lightbulb reactors.

Who is researching nuclear lightbulb reactors?

Thomas Latham was, in 1991 ... today, not so much.

https://ntrs.nasa.gov/api/citations/19920001892/downloads/19...

https://arc.aiaa.org/doi/10.2514/6.1991-3512

Re: AI for real-time fusion plasma behavior prediction and manipulation

#123

Post by LargoLasskhyfv »

perihelions wrote: post #41
lupusreal wrote: post #36

Earlier quoted context omitted.

The only hard part of dealing with nuclear waste is the social aspect. If not for that, you can simply and safely dump it into the ocean. Water is excellent shielding and the amount of uranium/etc already dissolved in sea water is absurd. Put it in a stainless steel vessel first if you want most of it to decay before coming into contact with the water, but that's not even necessary.

That doesn't really work because marine life is good at filtering and concentrating a subset of the elements that are in spent nuclear fuel. There are already ocean fish that are too poisonous too safely eat because of (coal-emitted) mercury pollution—and that's only 100,000 tons of mercury, total, in the history of human industry [0]. If you dig in to the hard numbers surrounding spent fuel, it's a much, much more t…

s/ocean/subduction zone/

aka the solution to pollution is magmatic delusion err... dilution.

Re: AI for real-time fusion plasma behavior prediction and manipulation

#124

Post by madaxe_again »

noname120 wrote: post #5

TIL: it's not just buzzwords. https://en.wikipedia.org/wiki/Fusion_power#Machine_learning

Yeah, this is precisely the kind of stuff I was talking about 48 hours ago, when everyone was telling me that ML will never find any kind of practical application.

There are so many fundamental fields - engineering, chemistry, biology, physics - which stand to have absolute quantum leaps in knowledge and capability with this technology.

Re: AI for real-time fusion plasma behavior prediction and manipulation

#125

Post by madaxe_again »

asdfman123 wrote: post #44
dekhn wrote: post #35

Earlier quoted context omitted.

When I used to work in grid computing almost 20 years ago, we were already running fusion experiments, realtime streaming the data to the grid, which would rapidly analyze it and compute some new parameters for the next run (I think they had a 20 minute downtime). I don't think it was considered machine learning at the time, though (and was certainly not deep learning as we practice it today).

Remember, AI is just procedurally generated data analysis.

You’re just procedurally generated data analysis.

Re: AI for real-time fusion plasma behavior prediction and manipulation

#126

Post by rajnathani »

noname120 wrote: post #5

TIL: it's not just buzzwords. https://en.wikipedia.org/wiki/Fusion_power#Machine_learning

AI is not a buzzword, try beating Go without machine learning. Just ignore the whole enterprise speak, and you'll see a lot of really cool things which are possible almost only by means of neural nets.

Re: AI for real-time fusion plasma behavior prediction and manipulation

#127

Post by phreeza »

noobermin wrote: post #89
chriskanan wrote: post #74

There is a lot of AI research in the nuclear fusion space. For inertial confinement fusion (a competing technology to magnetic confinement fusion, e.g., tokamaks) the National Ignition Facility (NIF) used it for their experiment that resulted in "ignition." My lab is collaborating with researchers at the Laboratory for Laser Energetics to use AI to improve inertial confinement fusion (ICF). We recently put out this p…

Here we go with the CS people saying "the old fortran codes are terrible." Yeah, the "high error" between code predictions and laser shots is because LPI is inherently noisy and it's essentially impossible to fully control conditions. I would work on expensive sims for days to weeks and the experiments would see differences that would be off from that from an order or mag because their focal point is off by a micron.…

I understood the comment as saying the fortran code maybe used some sort of inefficient numerical scheme and/or some inaccurate approximations? Doesn't seem completely outlandish that more modern methods could help there? Of course fortran is not a problem in itself, you are right.

Re: AI for real-time fusion plasma behavior prediction and manipulation

#128

Post by albertzeyer »

spenczar5 wrote: post #15
HPsquared wrote: post #7

Earlier quoted context omitted.

Neural networks have been used in industrial process control for many years. This is just another industrial control problem, perhaps a difficult one.

That’s really interesting! Do you have other interesting specific examples? I would have guessed that most industrial control problems were simpler sets of differential equations that could be directly estimated.

(Small) neural networks have been discussed to be used in computer chips for branch prediction. I don't really find a good source whether that really landed in production though. Here was some discussion:

https://news.ycombinator.com/item?id=12340348

Re: AI for real-time fusion plasma behavior prediction and manipulation

#129

Post by gauge_field »

phreeza wrote: post #127
noobermin wrote: post #89

Earlier quoted context omitted.

Here we go with the CS people saying "the old fortran codes are terrible." Yeah, the "high error" between code predictions and laser shots is because LPI is inherently noisy and it's essentially impossible to fully control conditions. I would work on expensive sims for days to weeks and the experiments would see differences that would be off from that from an order or mag because their focal point is off by a micron.…

I understood the comment as saying the fortran code maybe used some sort of inefficient numerical scheme and/or some inaccurate approximations? Doesn't seem completely outlandish that more modern methods could help there? Of course fortran is not a problem in itself, you are right.

In addition to the capabilities of language, usually it depends also on the ecosystem than the language itself (unless you want to write the entire algorithm by yourself for some reason). This is especially true if the algorithms used in the project are well-studied and the efficient implementations are provided. That is probably the reason why using modern/popular tool/language will more likely to perform better.

One of the most prominent example of this gemm. Usually, the state of the art code base on gemm are written C/C++, in terms of implementations in academic papers/github, see e.g. openblas, blis, blasfeo. The same situation applies to CUDA code or accelerator agnostic code e.g. using MLIR. I think it's more a result of how the language allowed people to create an ecosystem + ecosystem created by the people using the language. Sure, you can write Fortran, but if I see more tooling and more other people benchmarking, I can be more sure that the software will be tested more and higher performance etc. For instance, if you look at benchmarks-game, the top results are C/C++/Rust. Instead of making claims like this code is not wrong/right, we should look at the concrete/quantitative results like benchmarking/number of users. As another example, you can check blasfeo paper where they used C code.

I would welcome Fortran benchmarking results. But, I just dont see it being tested enough (in open source/papers/benchmarks) to prefer it over C/C++/Rust.

There are other consequences of this networking issue: availability of docs, finding an question and answer for a problem that you experienced.

Re: AI for real-time fusion plasma behavior prediction and manipulation

#130

Post by adrian_b »

rnhmjoj wrote: post #88
adrian_b wrote: post #65

Earlier quoted context omitted.

Deuterium is also not renewable, even if it is more abundant than uranium. The H1-B11 reaction would be a much better energy source than anything else, but for now nobody knows any method to do it. There is no chance to do it by heating, but only by accelerating ions, and it is not known how a high enough reaction rate could be obtained.

I'm curious, what are you considering for stating that deuterium is not renewable? AFAIK there's an essentially limitless supply in the form of HDO in the oceans[1] and there are cost effective methods[2] to isolate it. [1]: https://en.wikipedia.org/wiki/Semiheavy_water [2]: https://en.wikipedia.org/wiki/Girdler_sulfide_process

If you are able to say that there is a limitless amount of deuterium in the oceans, than you can say the same about the amount of uranium in the oceans, even if the amount of dissolved uranium is about one thousand times less.

Both the amounts of deuterium and of uranium in the solar system are finite and smaller than of the abundant elements. Moreover, the natural processes that create deuterium and uranium within a normal stellar system are slower than those that destroy them, so there is no chance of their quantities ever increasing.

Unlike using other chemical elements to make some stuff, using deuterium or uranium for producing energy destroys them without any means to regenerate them, so it is by definition a non-renewable process.

The hydrogen (protium) in the Sun is also non-renewable, but its quantity is enormous in comparison with the amount of deuterium existing on Earth (and the amount of energy that the Sun produces per proton is greater than the amount of energy that can be produced per deuteron).

Like deuterium is extracted from sea water, uranium can also be extracted from sea water, where it is one of the most abundant metals, except for the alkali metals and the alkaline earth metals. However the energy required for extracting uranium is significantly higher, due to its much lower concentration than deuterium (though deuterium is difficult to separate due to its similarity with the lighter isotope of hydrogen, while for the uranium ions much more efficient chemical reactions would be possible, which would bind uranium ions without being affected by the other dissolved ions).

Post reply on HN