Started being a Teaching Assistant for Intro to Programming at the uni I study at a while ago and, although it's not entirely my piece of cake, here are some "highlights":

* students were asked to use functions, so someone was ingenious (laughed my ass off for this one):
def all_lines(input):
all_lines =input
return all_lines

* "you need to use functions" part 2
*moves the whole code from main to a function*

* for Math-related coding assignments, someone was always reading the input as a string and parsing it, instead of reading it as numbers, and was incredibly surprised that he can do the latter "I always thought you can't read numbers! Technology has gone so far!"

* for an assignment requiring a class with 3 private variables, someone actually declared each variable needed as a vector and was handling all these 3 vectors as 3D matrices

* because the lecturer specified that the length of the program does not matter, as long as it does its job and is well-written, someone wrote a 100-lines program on one single line

* someone was spamming me with emails to tell me that the grade I gave them was unfair (on the reason that it was directly crashing when run), because it was running on their machine (they included pictures), but was not running on mine, because "my Python version was expired". They sent at least 20 emails in less than 2h

* "But if it works, why do I still have to make it look better and more understandable?"

* "can't we assume the input is always going to be correct? Who'd want to type in garbage?"

* *writes 10 if-statements that could be basically replaced by one for-loop*
"okay, here, you can use a for-loop"
*writes the for loop, includes all the if-statements from before, one for each of the 10 values the for-loop variable gets*

* this picture

N.B.: depending on how many others I remember, I may include them in the comments afterwards

I wrote a Blender plugin that uses vector math, matrices, calculus, trigonometry, and likely other types of math. There's recursion, filesystem access, image processing, interface logic, and on and on.

And worst of all - other people are expected to use it, so there's added pressure to do a good job.

Oh, the hours I spent trying to figure out why the imported geometry looked like an exploded mess. Fumbling around with mathematics I didn't fully understand was exhausting. Finding help was impossible at times because I didn't have the vocabulary to even describe the problems I was having. And getting it to complete an import before the heat death of the universe was not easy.

Every time I made progress and thought I was done, I would discover a bug that other importers didn't have, leaving me to sift through languages that definitely aren't Python to see if I could reverse engineer the logic they used.

I almost gave up a few times, but didn't.

Now I have something that, while not used by many people, works very well, is very efficient, and doubles as a palette cleanser when I need to do something for fun or for a challenge. Plus I learned a lot along the way.

So here I am sitting on my dusty laptop gaming laptop (because supposedly it would offer me better performance in compiling code and working with CUDA according to the people above me) at a research institute where I just started working at. I am told that there are some issues with the code and that it fails to build on Windows with MSVC that ships with Visual Studio 2017 and later.

I poor some hot tea from my insulated bottle I brought from home and start reading.

I look in this header file and what do I see - a custom uint24_t struct. Interesting...

I keep sifting through the code base. I find some functions that check and change Endianess. Ok, but the software is developed, built on and runs only on Win7 and later desktop systems. Never mind...

Further I find a custom "allocator" that is used throughout the whole code base. It has three inline static class member functions: allocate, copy and deallocate plus some private constructors. And these just wrap around the standard new and free calls. Some flavours of this class actually only deallocate (with a comment above them: "This allocator does not allocate. HANDLE WITH CARE!!!", which is btw the only "code documentation" I have managed to find).

But wait! What is this? A custom thread and mutex. Oh, and string, and vector.

Further down the rabbit hole I find a custom math library with a matrix class that does not support multiplication between a matrix and a vector. Perhaps not a use case I guess...

I continue and come across some UI-related calls. Interesting, I wonder what they are using as a framework. Oh, my...We have an extensive GUI custom framework written from scratch (drawing buttons and all).

All of this is to load an OBJ file and render it on the screen on a standard Windows PC in some way.

Very nice... ;_;

I don't like vector math for gamedev. It's not that it's hard. Vector math isn't hard until operations become O(n log n). It's that it's unintuitive, slow to write, and when I finally come to a solution after arduous number crunching, it always looks obvious, boring and kind of ugly. I don't think I could ever write a piece of vector math that I could be proud of.

I have an internship at some research company. My point is making face recognition apps with prog lang I know. This place is awesome. Well, compsci it's not my background, but I met many people. And they are great at math ....
.....

... Like they do 29 gray-scale images as a vector for PCA algorithm with size 64x64 pixel and COUNT A COVARIANCE MATRIXES WITHOUT TOUCHING ANY CALCULATOR OR PEN AND PAPER AND GET THE RIGHT NUMBER!

Man, this is insane. I don't even know 64x64. I love compsci

Dunno why but writing a Vector Math Library is super fun! Especially in Lua!

See now why I understand that in essence given a vector (parameters), you modify weights and biases minimally and these get passed through a set of dropoff style layers like ReLU and that in the end each layer leading to an output will basically sum up to a value that goes through sigmoid and concurrently equals the value desired once trained..... i don't see how this could cover all bases when parts of the math used to calculate the output is trigonemetric and polynomial. I mean not complex math ! Real basic things in my case, but a polar from cartesian coordinate conversion, angle and leg size, etc all going into determining that a target equals a landing zone and if not how to move things to it.

Is there something I'm missing where you kind of model the math because at best sin and cos could be a power series.

Looking to get a good understanding of the fundamental ideology and math behind neural networks and support vector machines. I am well versed with math so I can deal with heavier stuff if needed, I would like to see formulas but an explanation to their conception would be nice. Does anyone have any resources like this? Practical hands on practice exercises would be a plus