denhaus

the final one is not AI, it’s a glorb video from years ago: https://m.youtube.com/watch?v=NkYSK-_hVDQ

Short answer: It’s a way to generate structured databases for (most) scientific topics. Why? Apply data driven methods to these databases. So what? It’s a powerful way to ask and investigate scientific questions/trends otherwise hidden inside a million scientific papers.

Example: Consider what PDB has done for our understanding of protein folding, as well as the ML/computational techniques they’ve enabled (eg, Alphafold). Most scientific questions and properties are not as data-rich as protein folding. What if they could be?

Longer answer: The last 15 years in computational/ML + science have shown that structured databases open up entirely new frontiers in discovery (eg Protein Data Bank, Materials Project). But most scientific topics/properties are NOT in structured DBs, they’re scattered about in millions of papers. It’s especially a huge problem in some topics in materials science. It’s not that these problems are data scarce, but that it’s hard to actually collate their data in a structured format. You literally cannot use most ML methods because structured DBs do not exist.

This paper is a way to generate massive structured databases of specialized, intricate, and hierarchical knowledge graphs from scientific literature. Fine tuning works, prompt engineering does not (at the time, perhaps this has changed). Once you have a database, you can analyze an entire subfield or topic in science with ML or stats methods.

As a clarification, we used fine tuning more than prompt engineering because low or few-shot prompt engineering did not work for our use case.

Regarding point 3, my colleagues and i studied this for a use case in science: https://doi.org/10.1038/s41467-024-45563-x

I share your viewpoint on this, that DFT is a poor proxy model for ML to approximate.

However, the alternative of using, for example - experimental data, is that the synthesis procedures, measurement parameters, sample impurities, and even differences between experimental apparatus means training datasets of even modest size are insanely heterogeneous. So models either are either trained to predict differences between materials due to experimental discrepancies, trained on very small datasets, or must have a slew of post-hoc physics-based adjustments added to get reasonable numbers.

Higher order computational methods (including simply more intensive, non-high throughput DFT) are accurate but expensive as you know. Some of them have systematic error in the way DFT does, and are essentially based on user choice of (many!) parameters. Charged defect calculations are on example of this. Finding large (>10^4) training sets with similar parameters for computation is difficult. “ML” for these kinds of calculations usually consists of like, calculating a hundred (or 10) crystals within a narrow chemical system, doing a linear regression on one variable (eg, valence of cation on some site), and getting numbers +\- 10% of a “true” number.

GGA/meta-GGA DFT, on the other hand, can be applied at a sufficient fidelity to get real(ish) numbers in a homogenous way across huge numbers of crystals. So you are correct, you are predicting an approximate number for a property in many cases. But if we know the approximate number is wrong due to systematic error (and we can, in some situations) we can apply corrections or higher order methods to get the right(ish) answer. More, it’s highly dependent on which property you’re interested in. Some properties, like band gap, can be off by a lot. Others, like formation energy, can be calculated pretty accurately even with run-of-the-mill GGA DFT. Elastic moduli are generally ok.

in summary, approximating DFT with ML is just the least messy way to get real-ish answers across a large number of materials. Of course, there’s a point at which low-fidelity DFT calculations are - (1) so cheap and (2) so inaccurate, generally - that having an ML model approximate them is pointless. Most large DBs of materials now use good enough DFT that the numbers they calculate are not pointless for ML to learn from.

In the future, I think models trained on large numbers of DFT calculations will have to be applied to narrow sets of higher fidelity calculations by tuning. Much like you can fine tune a generalized LLM to do specific things. That might be where ML can actually bring real value to materials design.

Also, it’s worth considering that synthesizing novel materials can be insanely difficult. So 1 in 4 is not bad in my opinion.

i agree. in principle this is totally fine. in reality, highly paid lawyers throw the law around to accomplish whatever their objectives are. if they’re paid by a media company, and the media company’s objective is to reduce “illegal” reproduction of their content, principle is less important than winning.

More than half of the PhDs I know (in technical fields) have families. Actually, almost all of them.

Well I did as a racer and still do on my firm snow skis. But there is a slight performance trade off depending on what kind of skiing you like to do. I keep my edges on my pow skis dull because it’s slightly easier to slash and dash in crud and makes very little difference on icy terrain. For fully cambered directional skis being ridden very aggressively, it would make more of a difference. Every once in a while i’ll do the edges on my firm snow skis though if I’m bored, since a few years ago I bought a full diamond edge tuning kit and feel like I need to get some use out of it.

Point being, I’ve never found sharp edges matter much for most recreational riding. None of this has any relation to being uncool. Maybe in your wifes scenario, having razor sharp edges would have saved her from breaking her arm; almost certainly not.

On most firm conditions, I absolutely think you can rip it on fat skis with awful edges. Especially on groomers. I do it every year on 120mm waist skis. It is highly dependent on skill level. For an example, watch freestylers carve switch down some rock-solid melt frozen 45deg park feature on detuned $200 skis they picked up at a swap meet 8 years ago. The result is much more dependent on the rider, not the equipment. I’ve watched pro dudes rip harder than 99.999% of skiers on joke trash skis from the 90s (rusting, holes in the base, chunks missing from edges) and broken snowblades.

On true ice, NO ONE is ripping it except racers or ex-racers with good equipment.

I laughed out loud at the “fly off a cliff edge”

On the real however, getting down the mountain safely after dusk etc with dull vs. sharp edges will likely only affect intermediate skiers. Beginner skiers are going to crash no matter what they’re riding if on steep and icy terrain. Expert skiers know when and how to ride conservatively and can basically ride anything in any conditions “safely” (even if that means just sliding a firm patch rather than carving it), as long as they’re aware of the limitations of their gear. 90% of the year I ride pow skis in any conditions (including melt freeze etc) with super dull edges - it’s totally fine. The other 10% is just to have a little more fun on very firm days.

Intermediates, on the other hand, will be overly aggressive beyond their capabilities. They’ll bounce their helmet off a melt-frozen knoll at first opportunity, similar to what you said!

It’s a racing thing, mostly. Most shops offer a basic service for a machine wax + edge sharpening so that’s when most people get it done (even if they don’t really know what they’re paying for)

For anyone interested, we wrote a paper on a similar topic: https://www.nature.com/articles/s41467-024-45563-x

I am sure that aphorism is quite apt for thesis defenses, critical debates, shareholder meetings, peer review, and forums of discussion.

Not to be a hater but what exactly is the point of posts like this

Nothing factual presented. Not a unique, stimulating, or even really coherent viewpoint. Just a self-confident stream of consciousness with no real takeaway.

Eye-catching title though

it depends on if you’re looking for an accurate calculation of atoms and the number of atoms. if you’re just interested in how a lattice of spheres interact, force field software might be your best bet. if you’re interested in electronic properties of real materials you will need higher levels of theory such as DFT or beyond

i work alongside this group ( one of Ceders PhD students), but not on this project. I believe OP is referencing a different paper to that of the tweet author

this is why i still drive my old corolla. don’t have to deal with smart/network/access features, and it’s easy for shops to fix. hope i dont blow a head gasket now that i’ve said that…

i’ve been following Jake’s story since one of his posts was here on HN a few months ago. Both him and his wife are excellent writers. I cannot emphasize enough how horrified I am reading every word. What an utterly ghastly experience. I watched my uncle die from cancer as well. I am all too familiar with how this (usually) goes in the end stages.

good luck, jake. may your days be filled with appreciation and love and tranquility.

I wouldn’t say a “no”, just a failed replication. Though it’s not encouraging

It may mean the structure of LK-99 found by Kim et al is highly metastable. It does not mean the DFT calculations were useless. Also, for clarity, the original DFT calculations by Griffin do not purport to have “verified” superconductivity, but rather just report to have found a band structure (electronic energy levels) which are encouraging.

Wow, I can’t believe it. The death-dealing Nissan altima driver meme supported by numbers

Oops, seems like author actually has studied superconductors in the past so first part of my comment should be retracted. However I stand by the part in that I don’t think his explanation is good

This is a first-principles calculation known as density functional theory. it’s a useful basis for understanding physics in solids. it won its inventors the nobel prize in chemistry in 1998 iirc

This is a calculation with density functional theory (DFT). Props to the tweet author for mentioning this work, but it doesn’t seem like he knows much about the physics of materials because the explanation re: band theory in the tweet is (imho) not good. For those not familiar with materials physics, the fact that bands cross the fermi level does not make it a superconductor, it makes it a metal; the interesting finding here is that there are isolated (ie, far from other bands’ energies) and flat (ie, non dispersive) d-band states which is related to superconductivity

Anyways props to Sinead for doing the work here. Having a theoretical understanding of why LK99 behaves why it does (if it does) will be important for its optimization and commercialization. Briefly reading over the methods it seems(?) like it was quickly done with a GGA functional (given there is a +U correction) rather than a hybrid functional so this was probably not “heavy compute” as the tweet author would have you believe. OP it may be productive to just link directly to Sineads publication rather than this tweet

For reference I also work at LBL on computational materials, I have met Sinead on several occasions and she is very legit.

two questions:

- where do you go, and why?

- how do you get your US insurance to cover/work with providers outside the US? does it just automatically work or…?

Sorry to hear that. I had an uncle die from lung cancer (smoker, but otherwise healthy) about a decade ago. I know how horrifying the process is. Hope you have been able to process your grief.

legitimately the shittiest, most boring, and most unimaginatively dystopian use for AI I have ever seen.

Am I the only one who just cannot STAND MFA? Having to get a notification text etc. Like what if I don’t want to give an app capability to notify my phone? What if I want something totally NOT connected to my phone?

I just envision a future where there is some near-circular dependency of passwords/phrases/notifications/authenticators/keys/email verifications etc across different devices and services - the end result is that it is absolute PiTA to log into anything or recover any account if anything is ever lost. Sort of an endless personal bureaucracy for authentication. It’s a future I am personally trying to avoid at all costs

Yubikeys etc seem like something I could potentially get behind, but it still doesn’t seem perfect persay… anyway, maybe I am just a geezer

As someone who doesn’t know anything about this particular data, how would I go about obtaining that Slovak republic data for free?

I think this usually happens if your wifi/cellular is still on and connected but there is no actual data connection (ie AM thinks it can contact the server but it actually can’t, so it waits forever). Usually can be fixed by just turning on/off work/cellular and making sure it’s off. I download all my music on AM in lossless and haven’t had this issue once I turn off wi-fi/cell