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

Electronic Structure of LK-99

arxiv.org

Post reply on HN
Print view View original

221–230 of 448 posts

Re: Electronic Structure of LK-99

#221

Post by dogcomplex »

sheepscreek wrote: post #169
dogcomplex wrote: post #86

Earlier quoted context omitted.

What makes you think we're lacking fundamental plumbing? I am personally still quite convinced it will simply take a bit of time and development to plug in the current AI into sufficient self-testing structures (basically the next versions of AutoGPT) and let it run long enough to map out an entire domain with accurate measurements. Seems more like just a matter of time and a bit of funding - but mostly inevitable an…

I completely agree with you. I failed to articulate it well. Like the current top HN post suggests ( https://eugeneyan.com/writing/llm-patterns/ ), we’re still discovering patterns that work well with LLMs. That said, anecdotally - they already excel at being logic engines. Capable of filling in the gaps between instructions. Using their worldly knowledge or “common sense” to do so. But ever so often, they’ll miss an…

Yeah with their own QA from a variety of different personas/perspectives/concerns/contexts I reckon you'll get a very decent accuracy - or at least self-assessment of inaccuracy. All these can ever do is propagate the data they know so far to the context/prompt you desire, but I don't see obvious limits there. And GPT4 is already a superb conversation partner as smart as nearly any person - so it's really like piecing experts together. If we run into any fundamental limitations from piecing these all together, it's gonna be the same limit that any group of humans trying to make a coherent/consistent organization of knowledge encounters, I think.

Coincidentally, that appears to be how GPT4 was made - apparently it's actually about 8 personas with designated roles running GPT3.5 trained together ("Panel of experts"? there's an AI name for the technique). Makes you wonder how far that one trick scales.

(P.S. great link. Gah - another long read on the todo list)

Re: Electronic Structure of LK-99

#222

Post by SergeAx »

SamPatt wrote: post #47
jeffbee wrote: post #16

The most interesting phenomenon revealed so far is that the room-temperature superconductor people and the generative-LLMs-are-God people draw on the same limited global pool of credulity, such that HN cannot be flooded with both ChatGPT worship and superconductor speculation at the same time.

The least interesting phenomenon revealed so far is that HN is split into a group who are entertained by watching and discussing emerging tech / science news, and a second group who are entertained by disparaging the first. I enjoy watching floating rocks and talking to computers.

It seems to me that the second group is also entertained by discussing money, tech salaries in particular.

Re: Electronic Structure of LK-99

#224

Post by montecarl »

NotSuspicious wrote: post #72

One of the things that has most pissed me off about these papers is that VASP is proprietary and requires a license to use.

Some context on VASP from someone who worked with it for about 10 years.

VASP is a very common projector augmented wave (PAW) planewave DFT program in the solid state physics/chemistry community. I used it for about 10 years when I was doing computational chemistry. It is distributed as a tarball of FORTRAN90 files, so in some sense all researchers that use it have access to the source code. The research group I was in maintained a set of patches against the source code to implement additional functionality for transition state searches (useful for modeling solid state reactions).

Opensource alternatives exist, but are not as widely accepted (or somehow as fast in my experience). GPAW[1] is one example. It is unfortunate that it is not an opensource program, however, among the large community of scientist with access to the software the source code is available and is well understood and accepted. It is more or less the defacto standard against which other solid state DFT programs are tested.

[1] https://wiki.fysik.dtu.dk/gpaw

Re: Electronic Structure of LK-99

#225

Post by mjan22640 »

keepamovin wrote: post #54
jeffbee wrote: post #16

The most interesting phenomenon revealed so far is that the room-temperature superconductor people and the generative-LLMs-are-God people draw on the same limited global pool of credulity, such that HN cannot be flooded with both ChatGPT worship and superconductor speculation at the same time.

Lol thank you. That’s hilarious. Also, I think it kind of sucks that all that UFO reverse engineering didn’t yield a high temperature superconductor before this… but I guess what can you expect when you sequester away all your materials, silo your research and in effect keep your neural network of nodes (as scientists) really small. This open shit, as flawed and maybe annoying as it is to some people, I think it’s a…

The object in the IR Tic-Tac video is colder than the environment. High current supraconductivity will perhaps always require a low temperature.

Re: Electronic Structure of LK-99

#226

Post by seydor »

jeffbee wrote: post #16

The most interesting phenomenon revealed so far is that the room-temperature superconductor people and the generative-LLMs-are-God people draw on the same limited global pool of credulity, such that HN cannot be flooded with both ChatGPT worship and superconductor speculation at the same time.

Therefore proved that credulity is a conserved quantity, and HN is now searching for the underlying symmetry

Re: Electronic Structure of LK-99

#227

Post by irthomasthomas »

bluecoconut wrote: post #12

I got my PhD studying band structure of high-tc superconductors (experimentalist, ARPES). These Cu d-d interactions right at the fermi energy give me huge hope. Feels very familiar to other superconductors (re: all the cuprates). (Note: I specifically worked in a lab that was measuring a lot of the d-wave character / gap-energies of various superconductors) All in all, I'm now much more bullish on LK-99 being real su…

I think we're being trolled. Or trawled. Not sure which, yet. But the mention of Salvatore Cezar Pais in the patent should be enough to put this to bed for serious folks.

https://news.ycombinator.com/item?id=36967333

In case you aren't familiar with the man's work, here are some highlights of his discoveries over the last decade:

Conditional possibility of spacecraft propulsion at superluminal speeds

High frequency gravitational waves-induced propulsion

Piezoelectricity-induced room temperature superconductor

Craft using an inertial mass reduction device

On the Existence of the Superforce–the possible fundamental Force of Unification

https://scholar.google.com/scholar?hl=en&as_sdt=7%2C39&q=Sal...

I can't wait for my cheap antigrav FTL iron man suit to arrive next year.

Re: Electronic Structure of LK-99

#229

Post by bratbag »

mjan22640 wrote: post #207
rel2thr wrote: post #46

Earlier quoted context omitted.

Theres some crossover. Isn't this exactly what a rogue super AI would do first? - run some simulations and discover a superconductor - then write up a paper on it ( the paper has some gaps but humans take it and run with it) - human technology and computing power rapidly accelerates. - the rogue AI's powers increase exponentially

The oneshot or short session manner in which AI are typically used does not have a rogue potential. The AI would need to have a virtually unreachable goal, and run in an endless loop, to be able to drift its goals to be at odds with humanity.

As you say it, I'm tempted to try it.

Doubt I'm the first.

Re: Electronic Structure of LK-99

#230

Post by bratbag »

dylan604 wrote: post #98
dogcomplex wrote: post #88

Earlier quoted context omitted.

lol same as it ever was

not really. before, fraud claims were from the non-believers. now, the fraud claims are from the SEC and other government agencies. one can be swatted away with clever tweets, er, xes (what are tweets called now?), the other sees people going to jail.

Tweets are now Xcretions.
Post reply on HN