Patient0

which makes sense because I believe that polars was written by the same guy that did pandas (hence the name - panda and polar are bears)

The actual paper makes more sense: "the number of galaxies in that field that rotate in the opposite direction relative to the Milky Way galaxy is ∼50 per cent higher than the number of galaxies that rotate in the same direction relative to the Milky Way."

The actual paper makes more sense: "the number of galaxies in that field that rotate in the opposite direction relative to the Milky Way galaxy is ∼50 per cent higher than the number of galaxies that rotate in the same direction relative to the Milky Way."

The actual paper makes more sense: "the number of galaxies in that field that rotate in the opposite direction relative to the Milky Way galaxy is ∼50 per cent higher than the number of galaxies that rotate in the same direction relative to the Milky Way."

Bank of America Global Markets Data Science Apprenticeship | London | Full-time Apprenticeship | Onsite |

If you are UK based, graduating high school, but wish to pursue a career in finance, consider applying to our apprenticeship program (https://careers.bankofamerica.com/en-us/students/job-detail/...)

The Team

Quantitative Strategies and Data Group (QSDG) uses models, data, and analytics to develop and deliver impactful solutions to sales and trading teams across Global Markets. We collaborate across business lines and are guided by the highest standards of governance, ethics and scientific rigor. In your role you will contribute directly to the firm by helping us serve our clients and manage risk. You will be on active projects in the fast-paced environment of the trading floor.

Eligibility

- Hold or predicted to achieve at least 3 A levels/BTEC or equivalent with minimum grades of AAB (136 UCAS Points), with at least an A (or equivalent) in Mathematics

- Have at least a grade A or 8 in Mathematics and at least a grade 6 or B English Language GCSE or equivalent

- Have been a resident of the UK for the past 3 years

- Not in full-time education when starting the apprenticeship

- Have the right to work and remain in the UK indefinitely

- Have not completed a qualification or apprenticeship at the same level or higher which provided you with substantially the same skills or training as you would gain through this apprenticeship with Bank of America

I think a lot of the hostility about over achieving kids is that parents and their fellow students instinctively recognize that a lot of the success that comes from education is "zero sum".

There are only so many places and scholarships available at the elite universities, employers only pick the academic "winners" for their graduate programs etc.

When a child spends all their spare time studying and does better in life as a result, a component of this success is actually at the expense of the kids who may have been as bright, but did not spend as much of their spare time on it.

So now should all the other kids spend all their time studying so that they can keep up and not lose out? Maybe...

I'm not saying that any of this is wrong or even unfair - just pointing it out.

I read an article that stated that Microsoft lost an anti-trust court case against the EU in which the EU mandated that they allow third party competitors to provide this service. Microsoft has its own solution called Windows Defender.

https://www.theregister.com/2024/07/22/windows_crowdstrike_k...

Not to be that guy but this is the sort of thing I could imagine might benefit from an AI also. Train an AI to recognize drowning, and sound an alarm just in case the lifeguard misses it...

"Students tend to understand objects better" -> I actually a disagree.

A problem is mutability. Consider: "Is-a list of triangles a list of shapes?".

Most inexperienced people, using their intuition, would answer "Yes!", and try to design a "ShapeList" base class of "TriangleList", "CircleList", "SquareList"... and then puzzle over where to put the method "void add(Shape s)".

So you end up picking a subset of OO in which you take away inheritance, or mutability ("state"). But take those away and you don't really have OO any more (IMO)...

Even Eiffel got this wrong, fwiw. (https://staffwww.dcs.shef.ac.uk/people/A.Simons/research/pap...)

Bank of America | London, UK | Full-time | ONSITE |

We are seeking a highly skilled and innovative strategist / technologist to join our dynamic cross asset strategy team. As a key member of the team you will be responsible for developing and implementing advanced trading strategies and responding to end-user requests. Our team writes programs in Python that run on the Bank’s strategic platform, Quartz, to answer questions that relate to the entire global markets trading business. We work outside of the traditional IT organisation, based on the trading floor in London.

You will be assisting the Cross Assets Strats team as part of the Strategic Risk and PnL project to re-factor and redesign our market model code. This project seeks to explain the risk and PnL of the Global Markets business in a strategic fashion.

Working as part of a team of 7 people, you will be writing and debugging code in Python, running within Quartz, the in-house bank platform. On-going learning about and debugging existing market model code used for calculating risk and PnL, simplify and optimize to perform more efficiently. Programming ability is most highly prized. Financial knowledge is desirable but not strictly necessary, provided there is a willingness to learn. Please get in touch if:

If you are passionate about programming and are comfortable with a variety of programming languages and paradigms

* You have previously written scripting language, or at least know how you might go about it

* You appreciate functional programming concepts and writing algorithms in a functional style

* You have good mathematical abilities and want to learn more about using mathematical techniques to analyse data

* You have financial and quantitative knowledge, and would like to learn more

Key Requirements

* UK resident - Unfortunately, for this role, we are not able to consider applications from candidates who are not already resident in the UK.

* Bachelors or master’s degree in computer science, Mathematics, Finance or related field

* Experience within quantitative trading, algo trading or related role

* Proficient in Python, C++, C#, Java or Lisp (etc.!)

* Excellent analytical skills

* Good communication skills

Please email paul.hollingsworth@bofa.com if you are interested in applying.

https://careers.bankofamerica.com/en-us/job-detail/24010736/...

I feel like the whole article is an allegory of LLM development (as a critic would write it: devoting a stupendous amount of resources and 'training data' to 'memorize' the solution). I wonder if this was the author's intent?

This is also how virtual methods are implemented by C++ compilers.

yes and perhaps the difficulty of using a frequentist approach on a sample size of 1 is why the parent post thought that you cannot learn anything from a sample size of 1...

I think this is too pessimistic. One always gains some information, even from a single event. If you're catching the bus one and shows up literally the minute that you arrive at the bus stop, yes, you may have been very fortunate and in fact the bus runs only once per day, but you really ought to assign a lower probability to that possibility now, given your recent single new piece of information. There are ways to express this quantitatively too - but your own intuition ought to be enough of a guide here.

This. I remember forking out... $50?... to buy the Commodore 64 Reference Manual from a specialist bookstore (we had to drive there). I was so excited when I brought it home and could finally write programs in assembly (to put sprites on the border using raster interrupts!).

On my phone I cannot read more than the first page. Something wrong with the site?

Bank of America | London, UK | Onsite |

We require a strat/technologist to work in the Cross Asset Strats team. We work outside of the traditional IT organisation and sit on the trading floor in London, responding directly to end-user requests.

Our team writes programs in Python that run on the Bank’s strategic platform, Quartz, to answer questions that relate to the entire global markets trading business.

Please get in touch if any of the following apply to you:

* You love to program and are comfortable with a variety of programming languages and paradigms (e.g. Lisp, Haskell, C++, Python).

* You have written your own scripting language, or at least know how you might go about it

* You appreciate functional programming concepts and writing algorithms in a functional style

* You have good mathematical abilities and want to learn more about using mathematical techniques to analyse data

* You have financial and quantitative knowledge, and would like to learn more

If you are interested, please email your c.v. to emma.brady2@bofa.com

Basically yes - but even now I'd have to admit I have no intuition for why the different "frequency" functions are orthogonal to each other.

The only way I have been able to understand it:

- Functions can be thought of as vectors in an inner product space (https://www.youtube.com/watch?v=TgKwz5Ikpc8)

- the "inner product" operation (integral of the product of the two functions): imagine what would happen as you approximate functions as discrete vector with a very high number of dimensions/co-ordinates and computed the dot-product between those two vectors, but scale the result to be invariant of how many dimensions you used to approximate it => you get the integral formula

- Now, it's just normal linear algebra:

- The "length" of one of these vectors can now be thought of as the square root of the inner product of the function with itself

- The "distance" between two functions can now be thought of by subtracting one function from the other, to get a new "vector/function", and compute its length

- The cosine of the "angle" between two functions is the dot product between two functions scaled to have length 1

- The functions describing a sine or cosine wave are vectors which have a inner-product against themselves of 1, and a dot-product against any other frequencies of 0

- Thus the different frequency functions/vectors form an orthonormal basis

- This means that you can find the co-ordinates of any function by taking the inner product of the function against each fourier basis function

- The "co-ordinates" of your function w.r.t. the orthonormal basis can be computed by taking the inner product against each basis function/vector

- This will be the point that minimizes the distance to your actual function

- These "co-ordinates" are the fourier co-efficients for the fourier series representation of your function

- For non-periodic functions, you can take the limit as your period goes to infinity, that gives you the fourier transform representation.

Or, in short:

1. Functions can be thought of as vectors in an inner product space (https://www.youtube.com/watch?v=TgKwz5Ikpc8)

2. The Fourier series functions form an orthonormal set of basis vectors

3. Now just use normal linear algebra to work out the co-ordinates of your function w.r.t 2

ah and I get it now - the author has associations with the alt-right. For those who had heard of him (I had not), they were already inclined to view his statements uncharitably, even though I think this particular essay does not have any real "alt-right" themes.

right on thanks for making this point - "Also, IIRC, the predictions made by heliocentric model were less accurate than geocentric ones." - that was my recollection also: the epi-cycles had greater predictive power and accuracy.

I thought this was a quite insightful article - and was surprised to find so much criticism in the comments. e.g. those who are attacking his parable on the grounds that "actually the odds aren't actual 50-50 in chess" or "an alien would have to wonder why some people win more than others"... these objections seem like nit-picking to me and miss the opportunity to broaden our perspectives. The example is to help illustrate and illuminate an intriguing idea which does merit consideration: that the scientific method and various aspects of the scientific process are very much a "gradient descent" algorithm which will tend to converge to a local minimum... the large jumps needed to discover the "global minimum" are not sufficiently rewarded... or may not even be within our capabilities!

Here is a more concrete example I have been considering recently: The principal of least action can be derived from Newton's force laws. Alternatively, Newton's force laws can be derived from the principal of least action. Therefore, you could choose to make either "law" the more fundamental one and consider the other to be an emergent phenomena. Might this also apply to, e.g. the principals of thermodynamics? Might it be that all of the laws of physics are actually just an emergent phenomena that can be derived directly from the information theoretic properties of the laws of thermodynamics and entropy?

But then, why even stop there? We choose/discover these examples because our tiny, mediocre brains can only understand the universe from that point of view, whereas the actual universe we live in is unimaginably vast and complex. But might there be a way, for example, of deriving the laws of physics from some narrative principal (e.g. "we live in the best of all possible worlds... therefore by some chain of reasoning: F = ma!"). Maybe that chain of reasoning is only comprehensible to an artificial intelligence which we have yet to invent, involving us inhabiting some element of a vast fractal structure subject to some sort of anthropic principal?

And in general, it is worth considering the concept that not all things which are "true" are objectively measurable - those are just the things that are easiest to prove are true. To give an example of something which you know is true but which is, almost by definition, subjective: the existence of your own consciousness! It is impossible to prove, objectively, that you're not just some simulation that is running in a computer that's about to run out of memory and segfault... that you yourself are "real".

That's not to say that any of the above is actually true... but it might be true. Our recent focus on "provability" and "objectivity" has been used to argue that such things are in fact false, or just not worth considering or thinking about. There is sometimes an arrogance when there should be humility. Just because something might be exponentially difficult to prove, the sheer difficulty in proving it does not mean that it is not true.... There may be things which are true which we will never have the resources to discover. I think it's all worth considering.

Same here. Maybe it does not work correctly on all platforms? I'm using an old Android phone (Samsung G8)

Thanks I was just about to post the same thing

This speech by the SNB chairman describes fractional reserve banking well in my opinion: https://www.snb.ch/en/mmr/speeches/id/ref_20180116_tjn

nitpick: in this context, you would want to use the word "cue" not "queue". "cue" => signal for an actor to go on stage or whatever. "queue" => orderly line of people waiting to buy a movie ticket.

OTOH: https://earthsky.org/space/ligo-gravitational-wave-burst-nea...

He praises XSLT and says that all data should be stored as XML!? This article is surely a troll...

This was a great article - I'd been wanting to understand PCA. Particularly liked the digression on Lagrange multipliers.

" The only technical catch is that you aren't allowed to change the type of the contents in the function. That means if you start with something of type A in the container, you have to get a container containing something of type A at each stage of the pipeline."

I don't think this is correct. What doesn't change is the overall type constructor (e.g. if you start with a list, you'll still end up with a list... If you start with an IO action, you'll still have an IO action). But the type that is wrapped by the container can and usually does change. e.g. if you "map" over a list of A's, you'll get a list of whatever type your map function returns - still a list, but a list of B's, not a list of A's. fmap :: (a->b) -> [a] -> [b].