soVeryTired

He won the breakthrough prize, so he literally has been paid millions.

I don't think Nvidia was a particularly obvious play at the time. GPU bitcoin mining had taken off and GPUs were in high demand, but I don't think many people realised GPU compute would take off the way it did: everyone I knew in ML was still using clusters and most of them were using Matlab.

Even if you took it for granted that GPU-powered deep learning was the next big thing, it wasn't obvious that NVIDIA would be the dominant player. CUDA made GPGPU programming easy at the time, but any other manufacturer could have released something better. It just didn't play out that way.

Alexnet was 2012 by the way. Shortly after Nvidia stock tanked!

I think one interesting way of looking at a company is as a power struggle between

  - the owners
  - the employees
  - the managers
  - the clients
  - (and sometimes) the government / general public

Companies can benefit all of these parties, but a good system should have forces in place to prevent any of them from getting the upper hand. Some of those forces include competition, unionisation, labour mobility, fiduciary duty to shareholders, and the existence of boards of directors.

The view that a company's sole duty is to provide value to shareholders (i.e. to grow) is blinkered and damaging. I'm tempted to heap blame for that way of thinking on the economics establishment and on anti-comunist sentiment from the 20th century but that's probably naive.

Those analysts are incredibly specialised. The rank and file developer ‘at the bleeding edge’ spends their day buried deep in a single codebase. The analyst spends their day carefully picking through industry trends.

You say short sellers are using fundamentals as an ‘excuse’ to short stocks ‘they don’t like’. If not from careful analysis of the business, where should the opinions come from?

GR20 is definitely one of the more famous ones though

So if it's "opinion-size-age-shape-colour-origin-material-purpose noun", why would you say "the big bad wolf", but also "the bad old days"? In the latter, 'bad' is acting as opinion. But what's it doing in the former? Purpose?

I'm not a mathematician and I'm generally unqualified to comment on any of this but I'll have a stab anyway.

The TLDR is that Peter Scholze, one of modern mathematics' most accomplished theory builders, has a concept called a 'condensed set' that he thinks can act as a drop-in replacement for topological spaces in most of mathematics. Topological spaces are perhaps one of the least restrictive models of 'space' is in mathematics. Peter thinks his definition will allow mathematicians to bring algebraic techniques to bear on problems that were previously not amenable to such methods.

The problem is that everything rests on one theorem, whose proof is difficult. Because the proof is so foundational, he'd like to get it checked by computer instead of relying on the opinion of a handful of mathematicians, who may be sloppy in their checking.

I found myself thinking about multiplication tables in the shower today. I think you can get away with knowing the tables for 2, 3, and 7 really well. The rest is just tricks:

4: 2x twice

5: add a zero and divide by 2

6: a 3x followed by a 2x

8: 2x three times

9: 3x twice

Beyond that, it’s repetition of those tricks, plus the distributive property of multiplication over addition.

I think this is a different arm of SoftBank though, not the vision fund.

Also, even capital is sort of irrelevant because of gearing. I can go to the futures market and take the same risk with a $10m fund that you can with a $20m fund in the cash equity market.

Dim your monitor: save the world

>Because you can borrow against it instead of selling and without touching a financial institution, and this gives you liquidity without a tax event

But how do you repay the loan without liquidating the bitcoin?

Interesting point, but I think the academic culture has taken a pragmatic shift: there aren't many 'pure' bayesians or frequentists left anymore, and the boundary between the two viewpoints has softened. Stats has more of a 'shut up and calculate' mentality about it these days.

Anyone have thoughts on the ecosystem around Julia? Suppose you want to hook it into a database, or implement production-grade logging, or hook into e.g. Spark. Are those things do-able these days?

I've heard from some colleagues that Julia as a language is pretty solid these days, but all the machinery that you might need to take a system to production _aside_ from the business logic is a little lacking. I'd be interested to hear if others agree!

Aside from the first paragraph’s claims of inadequate research, I thought GP’s post was actually a pretty good critique of some of game theory - especially when you try to break out of the neat mathematical formulations it espouses and apply it to the real world.

Your post, on the other hand, reads like an argument from authority.

Vegetarian meals don't have to be sad. Indian cuisine is often veggie and exceptionally good. Meera Sodha's book Fresh India [0] is a goldmine for cheap veggie recipies that you wouldn't miss meat from (I say this as someone who likes meat!)

[0] https://www.amazon.co.uk/Fresh-India-Delicious-Vegetarian-Re...

But so often it happens that the working class and the owning class ally to reduce the power of the working class. They rarely seem to ally against members of the owning class.

Arrow's impossibility theorem has always seemed a little fishy to me: I think the axioms are too strong (if you pressed me on this, I'd wonder if the 'independence of irrelevant alternatives' axiom is problematic).

The fact that it holds when there are three or more options but not for two options just doesn't sit right with me.

The notion of comparative advantage that's presented in the article is a little facile though. IMO there are quite a number of real-world practicalities that derail the analysis:

* The infant industry argument. The analysis in TFA assumes comparative advantage is fixed: in reality advantage shifts all the time. Fifty years ago China had a comparative advantage in farming and not much else.

* Measuring comparative advantage: in order to make a good policy recommendation, we don't just know what each country does produce, we need to know how much each country could produce under a huge range of scenarios. Does anyone honestly know how much value a given country could produce if it decided to shift its entire economy in a new direction?

* Mobility of labour: instead of trading with other countries, why not just allow their workers to work in your country and reap the benefits at home? (I recognise that some advantages don't come directly from labour, but we do see this happen in low-skilled jobs). Even Ricardo recognised that labour mobility hurt his argument.

For what it's worth, I'm not saying that trade doesn't usually benefit both parties: I'm saying that _this argument_ isn't the smoking gun that demonstrates that fact.

Nice job uncovering the link to Bannon. In the abstract, aside from the claim of ‘censorship by the scientific community’, I thought this line was a red flag: * In this report, we describe the genomic, structural, medical, and literature evidence...*

If they had decent scientific evidence, they’d be pounding on it. Instead it sounds like they’re trying to cobble together a story from bits of cherry-picked analysis.

Go already required memorizing 'joseki' or standard patterns that crop up again and again (particularly in the opening). The usual mantra is 'don't learn joseki, learn _from_ joseki'. But the truth is - like openings in chess - if you don't know them and you play someone who does, you're often at a disadvantage.

The machinery described in the article is powerful and useful, but you don’t need it to understand dual spaces. Column vectors in R^n form a vector space. Row vectors map them into real numbers via standard matrix multiplication (on the left). Also vice-versa with right multiplication.

So row vectors are the dual of column vectors. Job done!

Edit: I admit you need a bit more for covariance and contravariance though.

But... you can ssh into a container and change state (depending on config options), can't you? I'm not sure I follow this response or the OP's complaint.

I mostly write python code and one very nice pattern I've found is to run a container somewhere (either locally or on a server somewhere) with an open port. You can SSH into it and use the remote interpreter as your main project interpreter. That way your dev environment is 100% reproducible. VS Code an Pycharm Pro both support doing that.

The chart shows a modes uptick which follows a large drop. But neither the drop or the uptick take inflation to a level that’s unusual unusual compared to its recent history. Wake me up if it continues for another six months!

I stumbled on Q.E.D by Feynman at a young age - it had a deep influence on me. I also read parts of "the mathematical experience" by Davis and Hersch, and "Godel, Escher, Bach" by Douglas Hofstadter.

It's not really maths, but Spacetime Physics: Introduction to Special Relativity would have been great for me at that age.

The Princeton Companion is a cool book, but it'd be better suited to a graduate in mathematics.

Oh, I got rid of it years ago. Thanks though. Eventually a big-name mathematician (who some years later won a Fields medal) waded into the discussion and sided with me. I used my real name in that discussion so I won't give more detail than that :)

I nearly based my Master's dissertation on a topic I discovered on a wikipedia math page when I was looking for something to cover. Turns out that subsection was maintained by the 'inventor' of topic and essentially served as a vanity page. The topic itself had no recognition in the community and if I had forged ahead on it, I would have failed my dissertation pretty hard.

That sounds like the beginnings of intuition for exterior algebra. While the gradient is a vector (i.e. it lives in a vector space), it's better thought of as being different to a standard 'displacement' vector.

Look at a topographical map of a landscape and note the contour lines. As you zoom into the contour lines (if they're detailed enough), they'll start to look more and more like parallel lines, densely spaced for a steep slope, or sparsely spaced for a mild slope. These parallel lines are the gradient.

A gradient and a vector 'fit together' to give a real number. The more parallel lines the vector pierces, the bigger the number [0]. So a gradient 'eats' a vector, spitting out a real number, and vice-versa. Just like a row vector 'eats' a column vector and spits out a real number.

I'm saying 'gradient' here, but really what I mean is 'one-form'. Language deliberately imprecise for all y'all mathematicians out there.

[0] https://neutronstars.utk.edu/class/p616_s19/lect3/index.html...

How so?

It reads like someone's playing around with a natural language generator.

This seems to be a decent intro. It's focused on pycharm but the same functionality is available in VS Code and most other IDEs.

https://www.youtube.com/watch?v=QJtWxm12Eo0