HN user
https://statmech.stanford.edu
My research group at Stanford has been alpha testing Tinker, it's both very useful and also really technically impressive in my opinion. It's a unified framework for post-training models and it abstracts almost all of the complexity of managing these jobs across resources. That it manages to do this while also allowing a lot of algorithmic flexibility is pretty unique.
In fact, the shoe was not banned and it is one of the most widely used marathon racing shoes out there. Nearly all other running shoe companies followed suit and made more efficient racing flats with carbon plates and high energy return foams. Some restrictions did come out, including limits on the "stack height" racing shoes and more stringent restrictions on the track. The current generation of these, the alphafly 3, were just used to break a world record in the Chicago Marathon by Kelvin Kiptum. Those are unreleased as of yet.
Statistical physicist here. Negative temperatures occur when a system has a finite number of high energy states. On average, when temperature increases, both the energy and the "randomness" or entropy of a configuration increase as well. Of course, if there are only a few high energy states available, then the randomness will not increase, it will decrease. That's negative temperature! Because we define temperature as the rate of change of the energy with respect to the energy, in systems like the one I described, this rate of change becomes negative.
Convolution is in fact multiplication in Fourier space (this is the convolution theorem [1]) which says that Fourier transforms convert convolutions to products.
Came here to make the exact same comment, agree completely!
I follow court news pretty closely and was genuinely curious about this decision. Here is a brief description of what I understand the decision to mean (I am not an expert by any means). Under the federal "Major Crimes Act" (MCA) members of various Native American tribes are not subject to state prosecution for crimes committed in "Indian territory". This case was brought by McGirt, a member of the Creek Nation, who challenged his conviction based on the original treaty establishing the Creek reservation in eastern Oklahoma. The headline is a bit misleading: for purposes of the MCA, eastern OK is now considered a part of the Creek Nation. This means McGirt must be convicted with the reservation's justice system or in federal court. It also has the consequence (discussed extensively in Roberts' dissent) that many existing convictions could potentially be vacated. It will be interesting to see how the Creek nation works with Oklahoma to address these changes.
I follow SCOTUS news pretty closely; the discussions below are a bit misguided. Audio transcripts of oral arguments are already widely available---you can even subscribe to the Oyez podcast feed and find them in your podcast queue a few days after the court hears the case. The new thing here is "live", so as a practical matter, it probably doesn't constitute a huge change. If there were an incentive for the justices to produce "sound bites", it would already exist. C-SPAN coverage will probably increase the visibility a little, but, having listened to many of the arguments this term, I'd say that most cases are too technical to be of much general interest.
A second point: oral arguments are performative. The cases are argued via written briefs and oral arguments provide a venue for the justices to question the petitioners about their arguments and air their responses to what the believe the other justices are thinking. Streaming the arguments, as opposed to making available courtroom audio after the fact, doesn't seem to change the dynamic much.
Many court watchers have taken this as an optimistic sign that perhaps the court will allow video. However, this is one thing that the court has strongly resisted. Some of the justices are known to prize their relatively low public profile and there's been speculation that maintaining that pseudo-anonymity is perhaps a reason for the hesitance.
As many have pointed out, this rediscovers Newton's method. The reason that this type of approach with a "hessian pre-conditioning" is not widely used in practice is that computing the hessian is costly. Avoiding that additional computation is the idea underlying "quasi-Newton" methods like BFGS and (more loosely) popular methods like Adagrad.
In the chemistry / molecular biology community, that is definitely part of the role played by a journal like Cell. It is considered a very prestigious journal and tends to publish only what the editors and reviewers perceive to be very important findings. Of course, we (scientists) are imperfect reviewers and sometimes don't see the value in great work and likewise sometimes imagine greatness in mediocre work. Editors also feel pressure to select work that they believe will be rapidly and frequently cited---as the number of citations per paper in the first two years after publication affects the impact factor of the journal.
Unlike math and physics, many disciplines have only recently adopted preprint servers (like bioRxiv) and there's still some bias against posting preprints within the community.
This is a fairly bizarre way to rank academic institutions. First of all, the methodology almost entirely neglects computer science because its metric of success is papers published in top tier journals; computer scientists tend to submit to conferences (e.g., NeurIPS) and hence get no credit for their work in this count. However, other fields seem over-represented from the list of journals---this is likely why Cold Spring Harbor, a very good biological lab where probably the vast majority of the papers are published in Nature-approved venues seems to be so elite.
The "normalization" they use divides the proportional count of authors that have contributed to an article in the "Nature Index" to the total output of the institution in the sciences, measured by a company called Dimensions. This has the odd effect of penalizing institutions for publishing outside their listed journals.
Finally, as an academic, there are some journals on the index that I have published in, but many venues I have published in did not make their cut. Sometimes more specialized journals are necessary---one cannot easily publish, for example, a detailed proof of a theorem in Nature, even if the result is very important.
List of journals: https://www.natureindex.com/faq#introduction1
As a statistical physicist, the title of this article looks the way "Two new papers explore the complexity of neural networks" might to a working ML researcher. Bubbles and foams are, and have been, a topic of intense research in soft matter physics for decades. Foams, in particular, have myriad uses in industry and have many well-established models.
It's a good question: you can see a fairly granular breakdown of the workforce by industry [here](https://www.bls.gov/cps/cpsaat11b.htm). There are about 5.1 million people working in "Computer and mathematical occupations", the majority of whom are developers.
It's probably worthwhile to think about scale in this context: There are only 85000 H-1B visas issued each year by the US; 20k of those are reserved for people with advanced degrees [1]. Given that there are an estimated 3.87 million software developers working in the USA [2], this seems unlikely to be the root cause of the inability of some graduates to find entry level software jobs.
1. https://www.uscis.gov/working-united-states/temporary-worker... 2. https://en.wikipedia.org/wiki/Software_engineering_demograph...
In NYC, I would say the average at an upscale place is about $3 for a cup of coffee, more like $5 for a cappuccino or similar espresso drink. With tip, I often end up spending $4-$5 for an espresso, though it pains me to admit that.
I have been using Ubuntu on and off since about 2005. Last year I switched from OSX on a dated macbook to Ubuntu on a newer Dell XPS. On the whole I am happy with the switch: with many of the apps that I use based on electron (slack, atom, mailspring) or directly in the browser (google docs), it is hard to notice the difference. I do most of my work related writing in latex, so this has also not been very noticeable.
I don't like latex slides and I give presentations frequently for work. The only thing that has proved consistently annoying is that LibreOffice Impress is nowhere near as good as keynote.
It's an interesting and complicated question, but it seems like the coinbase business model should be pretty dramatically affected by the massive sell-off. First, as prices go lower, the average value of a transaction is (probably) declining, so the money they make on fees is (probably) going down as well. Second, as the market looks worse and worse, the number of people casually entering or trading is likely declining too. The have a lot of correlated risk with the decline.
The article concludes that it is acceptable to stay on Facebook, but I take issue with its approach.
Underlying the logic is a premise: that because Facebook is used by some to spread propaganda and promote hate, it is (at least in part) to blame for the illiberal trajectory of some Western democracies. For many, it is more simplistic: there is a belief that Facebook is to blame for our election outcome. Of course, it is extremely difficult to know how much "to blame" Facebook as opposed to traditional media, and conclusive facts will be impossible to obtain. Both lines of reasoning, it should be noted, neglect the possibility that the election was determined by the will of people under a non-democratic system via the electoral college.
To me, the question is whether or not Facebook is a unique actor in this capacity. Does Facebook do something substantially different from other social media platforms or content providers that particularly undermines democratic values? I don't believe it does: Twitter spreads conspiracy theories, Snapchat is used for bullying and harassment, Google has comparable troves of data for sale to advertisers. While it has certainly had some public relations snafus, on the whole, as a platform it isn't that unique.
Currently, for "hybrid" journals, the researcher must pay a publishing fee, typically on the order of $2000-$3000. I have an independent postdoctoral fellowship, which means I have some funding, but not enough for it to make sense to pay. These fees essentially lock me out of open access publishing.
Of course, I could simply choose to publish only in open access journals (running the risk of career consequences). I hope changes like this will lead to increasingly prestigious fully open access journals. However, I have found the most practical option to be posting my papers on arXiv.
I have found that the machine learning research community uses twitter pretty effectively. I don't tweet myself, but I do follow a good number of other researchers and often find interesting papers and results through the venue. It's much more effective than scouring the arXiv.
One peculiarity of the academic system (true to some extent in start-ups) is that people transition from researchers to managers without any real training or evaluation of their proclivity for management. While you may be a spectacular researcher, that does not mean that you have the disposition, skill, or compassion to be an effective leader. Of course, some are naturals and many others get ample experience in labs where there's a hierarchical structure. Still, the utter lack of evaluation and reflection on group dynamics, productivity, etc., strikes me as a real weakness for the way that we currently organize academic research.
The MPIs are set up as director-driven research organizations. There is one person, typically a very prominent scientist, that oversees a group of junior faculty (who I believe have only fixed term positions---i.e., no possibility of tenure). Unlike most university settings, where each faculty member controls their research program independently, the Max Planck Institutes add another level to the hierarchy.
I'm a mathematician, but I work on things that are not even remotely related to the Riemann hypothesis. I could not possibly weigh in on the validity of the proof. My colleagues who know this area better are extremely skeptical, in part due to their opinions of Atiyah's work over the last decade. Mathematically, the point of skepticism seems to be the existence of the Todd function with the properties that Atiyah requires in his proof. See [1] for an informed opinion.
Atiyah is one of the most influential living mathematicians. His prior contributions give him a prominent platform from which to announce this result, which has quickly generated a media firestorm. To me, the worst part is that the work doesn't stand alone---a proof is a proof when it is understood by a community of experts. Until it has been carefully evaluated, no one should be criticizing him due to his age or past failures, nor should they be celebrating his achievement.
1. https://motls.blogspot.com/2018/09/nice-try-but-i-am-now-99-...
This write up not only oversimplifies, but it totally neglects one of the more interesting motivations of the original paper [1]. The authors are probing whether or not quantum mechanics is consistent with a single-world interpretation---that is, whether or not there is a unique reality. Formally, the claim is that there is no physical theory that is (1) consistent with QM, (2) consistent with a single-world interpretation, and (3) logically self-consistent.
I only used Apple products for about a decade, but last year I bought a Dell XPS which I run Ubuntu on. If you're looking to run linux, the XPS is a great choice. The hardware quality is extremely high---long battery life, ridiculous screen resolution (and it's a touchscreen), fast processor, and plenty of storage.
My only complaints stem from the fact that, when using linux, there are inevitably things that don't work as smoothly as they do on OSX, e.g., syncing with google drive.
While sci-hub undoubtedly plays a role here, in my field the arXiv is really what has obviated the necessity of journal subscriptions. Essentially all math papers, in both the preprint and post-review form, can be found on the arXiv. I would also emphasize that, with regard to decisions involving journal subscriptions and access, authors generally have very little power. As a researcher, if you have the grant money, you can pay to make your work open access; alternatively, you can restrict yourself to only open access journals (which in some fields, could come at significant opportunity cost).
A lot of the comments here suggest that, because participation is voluntary and because there are other incentives, like spending time outside, reduced sentences, etc, this practice is unquestionably fine. For me, the thing that makes it ethically complex is precisely that there are strong incentives for the prisoners to engage in risky, difficult work. Would you apply the same logic to people engaging voluntarily in affairs with powerful superiors?
The rational provided for these measures is very indirect, and, as such, it almost ensures the interpretation that it is retaliatory. If the concern were truly "integrating" tech workers into the community, perhaps the city would move to provide building permits for large urban housing complexes. Employer owned cafeterias create service industry jobs and are likely more sustainable than catering (with take-out waste). The goal seems more to punish tech workers and compromise a significant benefit to their employment.
There's a lot of work trying to piece apart the "thermodynamics of information". I am not sure that the linked paper is a natural starting point for the curious non-scientist. This popular summary is a much better place to get oriented: https://physicstoday.scitation.org/doi/10.1063/PT.3.2490
No, it's not a joke. The thermodynamics of information is a pretty active area of research (not too distant from the area that I work in). The basic question in this area is related to the Landauer principle, which asserts that irreversible computation requires energy and gives a precise bound for it.
The original article [1] is, in many ways, clearer than the economist's take on it and avoids political diversion.
They summarize Baumol's theory for entrepreneurship as assuming that the total amount of entrepreneurial spirit remains fixed, but, for structural reasons, some of it can get channeled into unproductive rent-seeking behavior (like getting favorable regulation).
They then say that their empirical work (three decades of data so more than one recession in there) shows that there is a decline in new firm formation, "in each state and nearly all metropolitan areas, and in each broad industrial sector, including high tech."
According to Baumol, this should be offset by an increase in "unproductive" behavior. This appears to be harder to measure, but they point to [2], which argues that both labor and capital (i.e., returns to shareholders) have been in decline.
This is all very complicated from the perspective of innovation and entrepreneurial endeavors in tech, where money is often lost despite the fact that the product is useful.
[1] https://hbr.org/2017/06/is-america-encouraging-the-wrong-kin... [2]https://home.uchicago.edu/~barkai/doc/BarkaiDecliningLaborCa...