rsfern

14 ms·

151.
by rsfern 3y ago
That C-H bond is part of an aldehyde functional group, normally those hydrogens are not removed (I think because it helps for pattern matching the functional group) Usually it’s just hydrogens on hydrocarbon chain backbone or aromatic rings
152.
by rsfern 3y ago
I think it’s unfortunate how the headline frames it like the original claim is somehow separate from the observe-hypothesize-experiment-explain loop of science. It’s also bit overconfident IMO - we totally still don’t have a handle on what’
153.
by rsfern 3y ago
Samples they synthesized. At the end of the paper they recommend the LK team do the same measurements. As far as I know there are still no independent reports on the original sample
154.
by rsfern 3y ago
Two reports have increased my patience factor a lot on this First, a theoretical study found that the phase that’s reported to be superconducting is likely to be metastable based on some density functional theory calculations. That implies
155.
by rsfern 3y ago
The steps are simple, but it seems to be quite finicky and there are almost sure to be sensitive to important factors that we don’t have clarity on yet. If it turns out to be definitively shown to be real
156.
by rsfern 3y ago
You are just early, the stream of papers are getting posted within hours of their release which is not really enough time to make it to wiki or explainer blogs I try to post a short summary of why I found a particular article interesting. N
157.
by rsfern 3y ago
it’s just a hypothesis, but they are saying that maybe the LK-99 samples people are making look like their figure 4 with tiny superconducting islands of lead That doesn’t mean we won’t be able to make wires and devices in the future. If it’
158.
by rsfern 3y ago
The lead films are a model system that show some parallels to the behavior people have been observing with LK-99. Their point is that if this hypothesis is correct, it should be possible to grow fully dense LK-99 material that has much more
159.
by rsfern 3y ago
Cool study - they made a Pb thin film and compared its superconducting properties to the recent reports on LK-99 reproduction attempts They’re hypothesizing that people may be getting superconducting nanocrystals embedded in maybe amorphous
160.
Some remarks on possible superconductivity of composition Pb9CuP6O25(arxiv.org)
179 points by rsfern 3y ago|92 comments
161.
by rsfern 3y ago
Ah, interesting - this NIST page uses °R, which is at odds with the explanation in the other page on Kelvin (which doesn’t mention Rankine). I’m not sure if that’s because there’s a strong convention to use the degree symbol with Rankine or
162.
by rsfern 3y ago
I agree with your first points, but the methods section states (somewhat unclearly) that they used this 4 point probe setup https://www.qdusa.com/products/ppms.html I have more questions about that, like why not calcul
163.
by rsfern 3y ago
The SI unit is just “kelvin” because it’s an absolute scale. Usually “degrees” is only used with Celsius and Fahrenheit scales because they are relative. Not sure about Rankine… https://www.nist.gov/si-redefinition/kelv
164.
by rsfern 3y ago
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
165.
by rsfern 3y ago
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 functio
166.
by rsfern 3y ago
GNNs and graph transformers are the current state of the art methods for this kind of crystal property prediction task. One drawback is that they don’t seem to capture long range structure all that well, and the current generative models (w
167.
by rsfern 3y ago
Arxiv releases preprints at midnight UTC, so if you’re in the US you might catch a paper dated Aug 1 on the evening of July 31
168.
by rsfern 3y ago
JOSS does something like this and it’s awesome. I’m not sure about the pull request workflow though, how would that work for review? Is your concept more for collaboration? https://joss.theoj.org/
169.
by rsfern 3y ago
The original preprint mentions bulk samples, but I’m not sure what size they were. I would guess at least a few millimeters scale, but really bulk just mostly means “not a thin film”
170.
by rsfern 3y ago
It seems the paper isn’t in sandia’s publications database yet, but it’s only been a few days since the journal version was published. I’m not sure how it works at sandia, but at NIST one of the authors has to get around to uploading the ma
171.
by rsfern 3y ago
Rubber bands are actually kind of complicated materials, they have more than one elastic regime before you have any plastic deformation. Maybe that’s what they are referring to. It’s not a perfect analogy, but applied strain changes a prope
172.
by rsfern 3y ago
I tend to agree, but it’s hard to know exactly how it went down from our outsiders perspective. They’re implying that they asked for changes before submission, and that they weren’t given a final call to ask for their names to be removed if
173.
by rsfern 3y ago
He’s an associate professor at BU, and I guess a lecturer at MIT
174.
by rsfern 3y ago
Sure - the point is that the preprint wasn’t posted by a student without permission of their advisor, it was posted by a professor without (apparently) permission of their collaborators, who aren’t part of a monolithic lab for what it’s wor
175.
by rsfern 3y ago
The person who submitted to arxiv (and promoted the paper on Twitter) is one of the senior authors (listed last in the author block), so I don’t think that’s an excuse for publishing to arxiv before doing due diligence on methodology Appare
176.
by rsfern 3y ago
I wasn’t sure what the units were on that one, since the chunk size is more than one byte. So NUMBER is the iterations, but I guess they’re looking for an answer in bytes, and it’s much less than 100M iterations?
177.
New NIST Public Working Group on AI(nist.gov)
3 points by rsfern 3y ago|0 comments
178.
by rsfern 3y ago
If you really can’t get away with a perceptually uniform colormap, have you tried turbo [0] before? It’s a rainbow colormap that tries really hard to have a smooth lightness curve 0: https://ai.googleblog.com/2019/08&#x
179.
by rsfern 3y ago
Agreed, my read of the abstract is that they’re doing near edge x-ray absorption spectroscopy (so XANES) and they make it a single atom technique by tunneling the signal into an STM tip
180.
by rsfern 3y ago
A counterpoint - Some of the US national labs have summer internship programs that can lead to legit publications, if the project is scaffolded well and the student works hard. I’m a coauthor on such a paper with a student from a local high

More ›