The stacks project works well for its intended purpose, which is NOT learning stacks but collecting precise statements and proofs for people already in the know.
19.02.2026 19:33
👍 5
🔁 1
💬 1
📌 0
The stacks project works well for its intended purpose, which is NOT learning stacks but collecting precise statements and proofs for people already in the know.
exactly!!
"A good mathematical joke is better, and better mathematics, than a dozen mediocre papers." [J E Littlewood]
Degree of symmetry of Riemannian manifolds www2.math.upenn.edu/~wziller/mat...
Still interested?
LeeSM, LeeRM. That's all.
WHAT
I should embroider this and hang it in my office.
👍👏❤️😘🥰😍
same!
i love when the textbook says "give a careful proof of" instead of "prove." oh so you think my proofs are normally careless. because i'm a bad mathematician. you hate me. you want me to hurt myself
Well, there are actually 4 kinds of [goes on about it forever...]
i just finished switching from onenote to obsidian before i saw this post. it was really easy to import everything. i got sick of microsoft having access to all my work. You can also import google keep!
obsidian.md
I disabled the view "title" (filename) option and write the "title" as a H1 header.
The new file template I use automatically adds the "title" in the three places: "title" property, "aliases" property and the H1 header.
Feel free to correct me if any errors are found.
Day 1: Hopfield Network (doesn't look very elegant but that's fine ig)
mood lately
Cauchy problem for inviscid Burgers' equation is not "solvable" (globally) for many initial conditions...but we can depict the graph of the function that "could have been the solution"... (and the characteristic surface of course)
The boundary of the preimage must be z: |p(z)|=1, so pp-bar =1, which (LHS) is a real two-variable polynomial? So we can check if it's an equation for an ellipse or not...
Green is after integration by parts: oscillates less, so we get a better bound for all of green => better bound for its integer values!
This would be so crazy a century ago! Old books literally had a "list of figures" in them...
Exactly! I am using it for almost everything I can think of now. Example: while trying to give bounds for Fourier coefficients of a 1-periodic function on R, we realise integration by parts gives non-trivial bounds. Here, red is the original naive integral, we only care about integer values...
Question: how can it get better over time if we don't give it enough data to copy from? Its good at coding because it is copying from all of github? We should train llms on graduate math books then? Elsavier, Springer vs OpenAI would be fun then (ofc won't ever happen)
It's fruitless to ask it to do any math above first-year undergrad level. It once hallucinated two invariant functions for a flow on the plane (which is "kinda" impossible if the flow is non-trivial, functions are non-cons etc)
However, it does randomly inspire "physics-type" arguments quickly...
Purple mangoes?!
Unfortunate. I had the same experince in commutative algebra without any knowledge of alg geom or NT
A sympy script in #obsidianmd. Powers of Arnold cat map matrix gives you FIbonacci coefficients and thus they grow exponentially...
Red: preimage of unit disk under a complex polynomial, or |p(z)|< 1
Violet, black: Curves defined by real and complex parts of p = 0 resp.
My friends say that about me 🤣
For 2 positive integers, the only possibility is {1,2} it seems math.stackexchange.com/a/3805538/12...
