KF Math Thread - Discuss Math

  • 🐕 I am attempting to get the site runnning as fast as possible. If you are experiencing slow page load times, please report it.
If some of your outputs are more critical, you may want them to contribute more to the loss. That's a simple example, but it's basically another way to shape the training and it's not discussed enough.
Hmmm, Python and R both have existing libraries that handle grid searching for hyperparameters and you can cross-validate so that the test set doesn't get tainted by repeatedly being peaked at. That might help but perhaps what you want is even beyond that.
 
Hmmm, Python and R both have existing libraries that handle grid searching for hyperparameters and you can cross-validate so that the test set doesn't get tainted by repeatedly being peaked at. That might help but perhaps what you want is even beyond that.
Oh I've already kind of gotten a grasp on it, I was just saying it's strange how it's a neglected part of the curriculum
 
Math is the subject of study for real men because you cannot innovate, design, or build complex things without an understanding of math.

People who want to learn how to code miss the forest for the trees. A programming language is just a tool in the same way a hammer is just a tool. You cannot design Gothic cathedral by just knowing how to use a hammer. Likewise, you cannot create your own graphics engine by just knowing Python syntax.

Whether you’re interested in computer audio, machine learning, ray tracing, data compression, computer architecture, or whatever nerdy thing, math is the shibboleth that distinguishes the real niggas from the bitches. And I’m not talking YouTube trivia math that fake nerds spout, like Gödel’s incomplete whatever the fuck.

Without math, you will be a slave to whatever tool you are using. It’s no wonder four of the seven liberal arts are focused on mathematics.
 
Oh I've already kind of gotten a grasp on it, I was just saying it's strange how it's a neglected part of the curriculum
It seems that certain texts (*COUGH* Springer *HACK*) often avoid intensely practical matters like this one
Whether you’re interested in computer audio, machine learning, ray tracing, data compression, computer architecture, or whatever nerdy thing, math is the shibboleth that distinguishes the real niggas from the bitches. And I’m not talking YouTube trivia math that fake nerds spout, like Gödel’s incomplete whatever the fuck.
Outright undecidability / uncomputability rarely seems to be an actual thing that actually crops up in day-to-day computing (with at least one notable exception). It's far, far more common that getting an exact answer involves some determined, finite number of steps but, even if you had every computer on Earth at your command, it would still take about as long as it takes to get out of Hindu / Buddhist hell from having done something really naughty in life. The factorial function is pretty easy to define in most programming languages and gives you a quick idea of how hard solving the Traveling Salesman Problem by brute force is.

(EDIT: factorial(100) for example is 93326215443944152681699238856266700490715968264381621468592963895217599993229915608941463976156518286253697920827223758251185210916864000000000000000000000000, or ~9.33 * 10^157, which is hardly extravagant for an optimization task with factorial time complexity)
 
Last edited:
  • Informative
Reactions: Named User
I was kidding.

It's not actually that hard it just takes fucking forever.
If you learn how to use R there's a lot built-in that expedites things. Python works too but it's more of a general-purpose programming language and R's hideously polluted global namespace at least comes with a few advantages.
 
If you learn how to use R there's a lot built-in that expedites things. Python works too but it's more of a general-purpose programming language and R's hideously polluted global namespace at least comes with a few advantages.
It's the "show your work" part that's the killer. If I could just do it all on the stat program or excel it wouldn't be as annoying.
 
Thanks OP for making this thread! I checked the archives last summer to see and there was a thread on maths from some years ago but it didn't go anywhere really. Hopefully this one blossoms!

Regarding maths and its practice: I would say that maths really is about how to think and how to structure your thought. This was how the concept "mathematics" was understood by the Greeks, a manner of learning and reasoning. While this might not be evident with the early stages of numbers, calculus and algebra, as soon as things become proof-based, it becomes all encompassing.

When reading maths, it's important to interrogate the material. Ask yourself: Why is done this way? Can it be done another way? Is this really the case? And prove/show it to yourself. Furthermore, be experimental and keep a good picture. Making things tangible but poking at it and visualizing what is going on can only help you.

I had an interesting thought while out having a smoke earlier, I was thinking about "quasi-random" algorithms, (algorithms that actually reduce randomness to play toward humans' perception of what random is), like shuffle features on music apps, and I started to wonder about how I could iteratively build an ordered set where at each step, the set is "as random" as it could be.
I would say that what you are getting to harkens to Ramsey theory which studies the question of given some arbitrary structure: past what extent does some sort of order appear.
"Randomness" is not so much the absence of order, or more accurately "regularity" or "pattern", but rather the absence of perceived order. Your ordered set, which one may suppose is a sequence of, say, numbers, is as "random as it could be" if it doesn't commit to any perceivable order. There might be one, you just can't ever see it. In this sense, you are asking the converse of the central question of Ramsey theory: given some arbitrary structure, to what extent does some sort of order not appear.
I will say that in my experience processes which generate randomness tend to have their randomness rooted in something else - say the digits of pi or radio static or some chaotic processes. You see this in random number generators for stuff like games and ciphers.
In this sense, the problem of generating a void of order is offloaded to something where no order is perceived. Of course, in the digits of pi, there is a pattern, we just don't know it.

Outright undecidability / uncomputability rarely seems to be an actual thing that actually crops up in day-to-day computing (with at least one notable exception).
Gödel's theorems are largely asymptotic notions about the extents of logic and the standard models of reasoning in maths. Their reputation is well deserved, with them killing the program that tried to formalize all of maths on logic headed by Hilbert (which subsequently made him never touch logic again), but honestly I find that those who hype the theorems up rarely understand them. For the working logician mathematician (logicians don't work), it's a matter of limits of possibilities, not the usual operation. Sure the totality is impossible, and that's great to know, but what of the partial. That sort of thing is more of interest.

Math is the subject of study for real men because you cannot innovate, design, or build complex things without an understanding of math.
Calculus went wrong when you made it for niggers. That's when it went wrong. It's like how many people, it's like, "Ugh it's epsilon/delta.." Ah, fuck you man. White people don't mind it, ok? The white people are like, "Yeah it's continuity. So what's your point?" The niggers are all like, "Nooo! We don't like limits!" Here's the difference in a white person and a nigger: Do you like real analysis? Ok, you're a nigger. Fuck you. Get the fuck out.

Also, being very geometrically inclined in both research and mentality: straightedge-and-compass geometry or classical Euclidean geometry has always fascinated me and the proofs tend to almost be works of art themselves. So when I get around to it, I'll see about posting their elaborations in this thread. The subject itself is a very good practice, if only mostly for leisure, as apart from architecture and art, I can scant think of its modern uses.
 
I would say that what you are getting to harkens to Ramsey theory which studies the question of given some arbitrary structure: past what extent does some sort of order appear.
"Randomness" is not so much the absence of order, or more accurately "regularity" or "pattern", but rather the absence of perceived order. Your ordered set, which one may suppose is a sequence of, say, numbers, is as "random as it could be" if it doesn't commit to any perceivable order. There might be one, you just can't ever see it. In this sense, you are asking the converse of the central question of Ramsey theory: given some arbitrary structure, to what extent does some sort of order not appear.
I will say that in my experience processes which generate randomness tend to have their randomness rooted in something else - say the digits of pi or radio static or some chaotic processes. You see this in random number generators for stuff like games and ciphers.
In this sense, the problem of generating a void of order is offloaded to something where no order is perceived. Of course, in the digits of pi, there is a pattern, we just don't know it.
My thought was to create some metric for human randomness, and create an algorithm to generate sets of a given length to maximize the metric. The irony being that in objective terms, I'm fixating on an extremely small subset at each iteration, making it the opposite of random
 
My thought was to create some metric for human randomness, and create an algorithm to generate sets of a given length to maximize the metric. The irony being that in objective terms, I'm fixating on an extremely small subset at each iteration, making it the opposite of random
Randomness is fickle in that trying to be more "random" is itself a pattern that makes things less "random". Though definitely think more about though and see if there's any supporting literature! It's interesting to me as a negative of Ramsey's theory in a way.
 
In this sense, the problem of generating a void of order is offloaded to something where no order is perceived. Of course, in the digits of pi, there is a pattern, we just don't know it.
There are the beginnings of patterns sensed in pi but if I had to guess, based on the past, there are still centuries if not millennia to go. Very possibly the search for patterns in pi could go on in perpetuity. And the reason I say that is actually because of number theory. Yes I know pi doesn't exactly fit in this category (though there appears to be some overlap) because it's an irrational real number but number theory is also centuries old and starts with something arguably even more simple from the outset: just regular-ass integers. But if you've ever solved problems on a site like Project Euler, you'll know that even just integers can be riddled with insanely autistic patterns. (Perfect numbers? What the hell is this shit?) And for literal centuries number theory was thought practically useless, but then the 20th century happened and all of a sudden stuff like really big semiprimes started to matter a lot when computing got commercialized and, maybe, you wouldn't want just anyone to read your credit or debit card info. So maybe generating as many digits of pi as possible isn't just a silly academic pissing match. Maybe digging for weird secrets like that will really matter some day.

There's actually a really elegant argument to that effect in the novel The Man Who Counted by the late Brazilian mathematician and author Júlio César de Mello e Souza, who wrote said novel before number theory went from "who cares" material to hot shit. If anyone is really interested I'll try to dig it up.
 
Pi actually holds an even higher distinction, that of a transcendental number.
Remembering these different distinctions among numbers calls to mind when I learned about the distinction between countable and uncountable infinity. If I'm not mistaken, these are called aleph_0 and aleph_1, respectively, and there are even further infinities beyond the two. No wonder Georg Cantor got ridiculed his whole life and ended up in a sanitarium to his dying days.
math-demotivational-cantor.png
Math, not even once kids!
 
  • Like
Reactions: Space Police
Remembering these different distinctions among numbers calls to mind when I learned about the distinction between countable and uncountable infinity. If I'm not mistaken, these are called aleph_0 and aleph_1, respectively, and there are even further infinities beyond the two. No wonder Georg Cantor got ridiculed his whole life and ended up in a sanitarium to his dying days.
View attachment 5852975
Math, not even once kids!
The way I think of the two is that one can be listed out in infinite time, the other cannot
 
The way I think of the two is that one can be listed out in infinite time, the other cannot
Well once you bring "time" into the picture, then we're talking about physics too, and that just really complicates things. I remember finding the diagonalization argument quite convincing but it's been years since then. Same goes for the argument over sqrt(2) that allegedly got a member of the Pythagorean cult killed (though this just might be a sick / amusing story).
 
Well once you bring "time" into the picture, then we're talking about physics too, and that just really complicates things. I remember finding the diagonalization argument quite convincing but it's been years since then. Same goes for the argument over sqrt(2) that allegedly got a member of the Pythagorean cult killed (though this just might be a sick / amusing story).
Oh, basically that's just perturbation theory jewery we use to approximate stuff
 
Back