The article's awkward opening statement proves it wasn't written by AI.
I have been interested in machine-assisted ways to do and teach mathematics from as far back as 1999, when I started coding several applets in Java 1.0, both for my complex analysis and linear algebra courses, to visualize various mathematical objects I was interested in (such as honeycombs or Besicovitch sets).
For the curious, eigenvalues only exist for square matrices. Singular values are like generalized eigenvalues.
Singular values are like the fundamental frequencies of your matrix. You know how you can define any color with RGB? In a (pretty handwavy) way, singular values are like RGB color codes for us math guys.
Optimizers like Muon and Adam play around with weights' first, or second order singular values to train models.
I'm not trying to be reductive but the article's a lot of words for "We're vibecoding our app now and the glorious (almost almighty) Haskell compiler is too slow for the agent to iterate it's mistakes until it gets it right."
Compiling my C code with gcc happens in a single eye blink.
Changing the format to .cpp and compiling with g++ takes 5 eye blinks.
I've never used rust and I can't imagine blinking more than 6 times while waiting for my code to compile.
Is this how they view us? Like yeah. I'm going to spend all night arguing with the Haskell lovers (that's a slur in my books) on Reddit, but it's not my hobby. Rather, responsibility demands I direct these functional programmers away from the devil.
I was looking at matmul algorithms and their hardware implementations (to start a hardware startup) and I saw that the naive O(n^3) version is what everyone uses.
Extreme mental clarity in "Eventually, she told me that it made more sense for me to quit my job while she worked, so that I could spend all of my free time trying to get another tech job".
I could've never imagined long-term-thinking like this from a former addict.
It's an appeal to the attention economy. "All of human cooking compressed into 2 MB" is(mentally) palatable relative to "Navigating the Emergent Geometry of Food Ingredient Embeddings".
This is the final phase of solving discrete logarithm problems where our goal is to convert a big integer into a product of small prime numbers over our factor base.
Information is scant about the process so in the spirit of "name-and-conquer", this phase is also called the Reduction Step, the Descent Phase or Individual Logarithm Collection phase of solving a DLP.
This paper introduces the Galbraith&Ruprai modification to the Gaudry-Schost collision finding algorithm for solving discrete logarithm problems in an interval in (1.36√N) field operations.
OP takes issue with GitHub's constant outages and alludes to agents (and Copilot bloat) as the primary cause.
Lots of big services are like this. Google Colab's 'Connect to Drive' is down as we speak. I'm up right now because I know my Runpod VM in Kentucky is going to die rather abruptly and I'll need to manually get it up.
Everything has its flaws.
Microsoft lets you host your code, websites and media for free and
The junior hiring collapse is all so bizarre. I graduated recently and my career prospects are jarringly limited.
I thought I'd go back for a Masters/PhD but then Trump mercurially defunded lots of STEM grad programs. Ngl, I found myself stuck. Zero job openings, zero PhD program openings. It's all so frustrating.
Odd anecdote. I completed high school in 2017 and my home country demanded us use mathematical tables, not calculators, to find logs and sines for our version of SAT math.
I got my highest-paying numerical programming contract (in the US) just because I knew (from high school math table experience) how to use LUTs to calculate a lot of useful stuff i.e quarter squares.
Modernization is great and all. However, it's disappointing to know lots of new programmers are oblivious of the fundamentals.
This comment put everything into perspective.
I can't name anything beyond Facebook, Instagram or Whatsapp that Meta's created and I've used in the past 10 years.
I've never even (knowingly) used the LLama models tbh.
We introduce the Hidden Number Problem with Chosen Errors (HNP-CE), a variant of the linear hidden number problem where the exact noise terms (βi) are unknown but shared across different relations with known weights(w).
Twitter: https://twitter.com/murage_kibicho
Email: [email protected]