2023 will go down as a miracle year not unlike 1905. Between AI, this, and… what’s next?
Blockchain blockchain blockchain blockchain
Electronic Structure of LK-99
61–70 of 448 posts
Re: Electronic Structure of LK-99
#62My understanding is that the "99" in LK-99 is the year it they first synthesized the material, i.e. 1999. Assuming this is all true, why is it just now coming to light? Did they just not know what they had? (I have not been following this closely, maybe this has already been explained)
Uniformed opinion. It takes a week to bake/whatever a test sample. Which is why this is playing out so "slowly". So, they had an idea and have been baking/testing/refining samples for decades. Science and funding takes a while. What idea they started with and why they kept at it for 20 years is beyond me, that's a long time to chase an idea of you didn't have results or a hint that your idea would work. Maybe they ha…
Speaks to a supreme confidence in the theory, surely? I'm a layman but imagine everyone is going to want to hear every wild idea these guys could think up if this works out. They might have been sitting on this for 20 years? Probably why I'm not, but If I were any vc person I'd be busting down their door trying to give them all my money
Re: Electronic Structure of LK-99
#63Earlier quoted context omitted.
Sure it’s post hoc, but these are ab initio calculations, there’s not all that much wiggle room that you could make anything you like come out looking like a superconductor. Plus now I’ve seen three calculations with three different functionals and reasonable convergence criteria and they all find similar electronic structure
Well, then why spend all this time searching in a lab? Instead, run a bunch of simulations until the compounds with properties you desire just sort of pop out as the top hits in your simulations? (I say this as somebody with a decade+ of experience running large ensembles of classical MD simulations, but not so much experience with inorganic DFT)
If you run a bunch of simulations to brute-force-find what you're looking for, you run into a problem of infinitely many things to simulate. It is better to use domain knowledge to eliminate things that don't work and narrow down things that would have high chance of working in theory to give some sort of directional guidance for your research.
Re: Electronic Structure of LK-99
#64More DFT? Haven't we concluded that has little predictive power?
Re: Electronic Structure of LK-99
#65Earlier quoted context omitted.
the singularity isn't going to implement itself, you know.
The next breakthrough we need is with fusion technology. With unlimited fusion energy, room temperature superconductivity, and (soon) artificial general intelligence, the singularity won't be long off.
Yes, a combination of fusion, superconductivity, and AGI would indeed be transformative. It's hard to anticipate what the results would look like.
Good reading: https://en.wikipedia.org/wiki/Post-scarcity https://en.wikipedia.org/wiki/Kardashev_scale then follow it up with a bit of 1993, https://edoras.sdsu.edu/~vinge/misc/singularity.html and https://web.pa.msu.edu/people/yang/RFeynman_plentySpace.pdf
(oh, huh, I just noticed that There's Plenty of Room at the Bottom starts with a reference to the discovery of superconduction)
Re: Electronic Structure of LK-99
#66Earlier quoted context omitted.
Sure it’s post hoc, but these are ab initio calculations, there’s not all that much wiggle room that you could make anything you like come out looking like a superconductor. Plus now I’ve seen three calculations with three different functionals and reasonable convergence criteria and they all find similar electronic structure
Well, then why spend all this time searching in a lab? Instead, run a bunch of simulations until the compounds with properties you desire just sort of pop out as the top hits in your simulations? (I say this as somebody with a decade+ of experience running large ensembles of classical MD simulations, but not so much experience with inorganic DFT)
There are high throughput DFT projects (e.g., the materials project) but (1) the calculations are very computationally expensive and (2) the search space is really large. People are doing cool stuff with ML and generative models, but it’s a pretty open research area still
Also, at the end of the day, DFT is still an imperfect approximate model. Relative trends are generally more reliable than exact correspondence with experiment, and it can have system-specific systematic errors that are hard to account for in a high throughput setting
Re: Electronic Structure of LK-99
#67Earlier quoted context omitted.
On a related note: I am deeply disappointed by how OpenAI has managed to continually dumb down ChatGPT. I’m sure the shortage of NVIDIA H100s and the demand for their APIs shooting through the roof has something to do with it. Perhaps they’ve switched to a quantized model? Who knows. And yes, LLaMA 2 is awesome. Progress is being made. But it’s slowed down. The novelty’s worn off. Now we’re lacking some fundamental p…
It's not dumber. You can verify this on their models with any prompt you ran at 0.0 when it first came out, I did. It's the exact same model run the same exact way. They've repeatedly confirmed this.
The ChatGPT webapp is not the same workflow as the ChatGPT API.
Re: Electronic Structure of LK-99
#68[flagged]
Re: Electronic Structure of LK-99
#69I 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…
> These Cu d-d interactions right at the fermi energy give me huge hope. Feels very familiar to other superconductors (re: all the cuprates).
As someone who is versed in semiconductor band structures but not superconductor band structures: What is it about Cu d-d interactions that causes the superconductivity?
Degenerate energies in semiconductors don't give rise to electron pairing, so I'm a bit out to sea with the proposed mechanism here.
Re: Electronic Structure of LK-99
#70Earlier quoted context omitted.
On a related note: I am deeply disappointed by how OpenAI has managed to continually dumb down ChatGPT. I’m sure the shortage of NVIDIA H100s and the demand for their APIs shooting through the roof has something to do with it. Perhaps they’ve switched to a quantized model? Who knows. And yes, LLaMA 2 is awesome. Progress is being made. But it’s slowed down. The novelty’s worn off. Now we’re lacking some fundamental p…
It's not dumber. You can verify this on their models with any prompt you ran at 0.0 when it first came out, I did. It's the exact same model run the same exact way. They've repeatedly confirmed this.
> It's not dumber
So I absolutely agree with this. And yet this meme persists. I wonder what's creating the feeling that it's "dumber" in so many? Perhaps they're just noticing the limits that always existed previously? I'm not sure, and am interested in others thoughts on it.