/c/DSP. Digital signal processing, i.e. how to transform, filter, and live with digital signals (e.g. audio files, image files, video files, sensor measurements, etc.). It involves a lot of math, so unless we get R*ddit-like numbers I don’t really know how such a community could keep moving.

The exam is tomorrow (today is another)

Ouch. Been there. Good luck on your exams!

Sounds like fun! I’m going to bed soonish but I’m willing to answer questions about multivariable calculus probably when I wake up.

When I took multivariable calculus, the two books that really helped me “get the picture” were Multivariable Calculus with Linear Algebra and Series by Trench and Kolman, and Calculus of Vector Functions by Williamson, Crowell, and Trotter. Both are on LibGen and both are cheap because they’re old books. But their real strength lies in the fact that both books start with basic matrix algebra, and the interplay between calculus and linear algebra is stressed throughout, unlike a lot of the books I looked at (and frankly the class I took) which tried to hide the underlying linear algebra.

I’m autistic too and I had to relearn math as an adult. Now I know calculus and advanced mathematics.

I can go find some book recommendations, but when I was first learning I really got a lot out of watching The Organic Chemistry Tutor.

Linear algebra (ex: multiply the matrices A and B), multivariable calculus (example: find ∇F with F=[xy,yz,xz]^T ), or actual “multidimensional analysis” (example: define the norm of [1m,1m/s,1m/s^2 ] in a way that makes sense)? I can help with all three.

Humans are good enough. It’s the shareholders who suck here.

Gross

It can use ChatGPT I believe, or you could use a local GPT or several other LLM architectures.

GPTs are trained by “trying to fill in the next word”, or more simply could be described as a “spicy autocomplete”, whereas BERTs try to “fill in the blanks”. So it might be worth looking into other LLM architectures if you’re not in the market for an autocomplete.

Personally, I’m going to look into this. Also it would furnish a good excuse to learn about Docker and how SearXNG works.

LLMs are not necessarily evil. This project seems to be free and open source, and it allows you to run everything locally. Obviously this doesn’t solve everything (e.g., the environmental impact of training, systemic bias learned from datasets, usually the weights themselves are derived from questionably collected datasets), but it seems like it’s worth keeping an eye on.

Google using ai, everyone hates it

Because Google has a long history of doing the worst shit imaginable with technology immediately. Google (and other corporations) must be viewed with extra suspicion compared to any other group or individual because they are known to be the worst and most likely people to abuse technology.

Literally if Google does literally anything, it sucks by default and it’s going to take a lot more proof to convince me otherwise for a given Google product. Same goes for Meta, Apple, and any other corporations.

I have a “unique” accent of some kind. Basically, I don’t sound like I’m from “around here”, even though I have lived in the same area for my entire life and so have my parents. Probably because I read a lot more than I actually talk to people.

If that happens then I swear to fucking God I’m done with YouTube forever.

Edit: i.e. if it breaks FreeTube, Invidious, Sponsorblock, etc. (because I’m already done with the main site forever) then I’m out. If the choice is between content vs no ads, I’ll take no ads even if it means no content.

What fucking plans

Last panel should be the entire US Customary System, which is literally just a rescaling of the SI (“metric” system) units. US Customary is derived directly from SI.

Going to bars. I can’t drink with my medications and I don’t like going to loud, social places unless there’s some other reason to be there.

Desperately needs more upvotes

DSP (digital signal processing) is the field of applied mathematics and engineering dedicated to transforming and manipulating digital signals.

Examples of real digital signals include audio files, image files, video files, and digitized recordings of various physical quantities by computers like the configuration of a robot as it moves in time, measurements of the processes in a factory, the trajectory of a spacecraft — almost anything that can be periodically sampled and take on a finite set of values [1] can be seen as a digital signal.

DSP includes using tools like the Discrete Fourier Transform (DFT), the Z-transform, wavelet analysis, probability, statistics, and linear algebra to do things such as filter a signal (example: audio equalizer), predict future values (example: weather forecasting), data compression (example: JPEGs), system identification (example: fit a model of the earth to predict seismic activity), control (example: make a DC motor to respond to position commands), and stabilization (example: keep plane from “wanting” to smash into the ground). Particularly, it requires a careful consideration of the effect of sampling a signal (example: if done carelessly, you can make the sampled system unstable [read: explode]), as well as an interpolation process of some kind if you plan on using that signal outside your computer (example: you want to hear an audio signal stored on your computer).

I got into DSP because I was an audio engineer and musician [2], and I wanted to design my own audio plugins. IMO I think almost everyone would benefit from some knowledge of DSP, but the math is really intense. Personally, I found out late in life that I have a nearly infinite appetite for math, so it’s a good fit for me.

Here’s a playlist about DSP if you’re interested.

[1] Actually, a lot of basic DSP books don’t restrict the signal to be in a finite set because it makes the math easier if the signal could be any real number. However, certain structures that would be exactly equivalent in theory are not equivalent on a real computer because ordinary computer arithmetic is approximate.

[2] I still play music, but not as much as before engineering school.

Digital signal processing

No, turtles never even existed, I’ve never seen one. /s

Infinite-dimensional vector spaces also show up in another context: functional analysis.

From an engineering perspective, functional analysis is the main mathematical framework behind (1) and (2) in my previous comment. Although they didn’t teach functional analysis for real in any of my coursework, I kinda picked up that it was going to be an important topic for what I want to do when I kept seeing textbooks for it cited in PDE and “signals and systems” books. I’ve been learning it on my own since I finished Calc III like four years ago.

Such an incredibly interesting and deep topic IMO.

I actually designed a digital equalizer using an IIR filter this semester, which actually does theoretically work on sequences of numbers, which constitutes an infinite dimensional vector space. The actual math was just algebra and programming, but it was an implementation of a Z-transform transfer function which is a sequence operator (maps input sequence to output sequence).

IMO infinite-dimensional stuff shows up in two types of problems:

1. For some reason, you need to solve the partial differential equation you started with, i.e. you can’t use symmetry or approximations to simplify it into an ordinary differential equation.

2. When you’re dealing with signals that change in time or space, you have to decompose those signals into simpler signals that are easier to analyze.