@feigenbaumconst

Elliptically fibered

Joined August 2026
  • Tweets70
  • Following22
  • Followers13
1Media
this thread has an awareness about the world that is rare in the gen pop of recent ai x math essays
Jack@Jackw1549Oct 10
Replying to @nihilunbounded @xobnu
Justification is the interesting thing here. The drug-addled finance bro deals with immediate intelligibilities: no school is needed to feel the power of money. His whole `career' deals in producing commodities, nothing more. It is a plane of rapidity because understanding is
2
73
- be willing to have scientific and policy discussions with people who aren't tenured professors at Harvard/Stanford. i get that AI/ML academia is very hierarchical and power is concentrated in a small handful of labs, but this is not true for all fields. 5/n
1
41
- release all the problem statements that were tried along with partial solutions if any. no human parallel here, but this would be tremendously helpful for building on, and for getting a better understanding of capabilities. 6/6
10
ball multiplier theorem retweeted
Yiannis Sakellaridis@yiannissakSep 4
If your claim to fame is pushing a button to optimize a breakthrough announced by a junior mathematician a few days ago, it’s not very impressive, to put it politely. Without getting into rumors on the sequence of events…
Axiom
@axiommathaiSep 3
1/ ✨ World record today on bounded gaps between primes, one of number theory’s oldest open conjectures.  We show that infinitely many pairs of prime numbers are separated by 212 or less.
4
8
2
215
20,191
ball multiplier theorem retweeted
Scott Armstrong
@scottnarmstrongAug 29
Tuomo Kuusi and I have lean formalized the main results in our paper on anomalous superdiffusion (joint with Ahmed Bou-Rabee). There we used rigorous renormalization group arguments (based on quantitative stochastic homogenization techniques) to show anomalous diffusion for a drift-diffusion equation. The analysis is notably complicated (the paper is 155 pages) and the heart is a big induction argument with many moving parts.
3
8
1
85
7,340
Ok let's figure out why S^{4n} has no almost-complex structure for each n >= 1. It's a good graduate quals problem.
1
1
130
Write E for the tangent bundle. Assume that E has a complex structure. Then look at total Pontryagin and Chern classes: \sum_i (-1)^i p_i(E) = (\sum_i c_i(E))(\sum_i (-1)^i c_i(E)). This is a sphere so only p_n and c_{2n} are nontrivial. Then (-1)^n p_n(E) = 2c_{2n}(E) = 4.
1
126
This implies E is not stably trivial which is a contradiction.
13
Local transit rider replaces social media browsing with math hillclimbing (heartwarming)
1
89
Lol great case study of the variable outputs of vibe proving. Haven't read closely but it seems like none of the papers bother to make a case for why we care about this conjecture besides it being old and being posed by Gromov. This is a common issue even pre-AI though...
Steve McCormick
@QuasilocalAug 19
A fourth proof this week just landed: arxiv.org/abs/2608.17977 Mentions AI but it's less clear to me how much is AI compared to the other two with AI statements.
97
Fun lemma: X and Y are independent random variables. Then X+Y and X-Y are independent if and only if X and Y are Gaussians.
1
73
Proof: Fourier transform! Use independence to show the characteristic functions solve M(2t) = M(t)^4 and so are exponential.
52
My research process is officially in the weights
prinz
@deredleritt3rAug 10
Keep going! Believe in yourself!
65
Good stuff arxiv.org/abs/2608.00802

Counterexamples to the Landis conjecture in dimensions three and higher

In every dimension three and higher we construct a nonzero function that satisfies the stationary Schrödinger equation with bounded, real-valued potential and decays like $\exp(-c|x|^{4/3})$....

arxiv.org
44
Many math professors are worrying about whether we'll be able to train PhD students anymore. I'm wondering whether the PhD students will need us to train them. (They are very smart)
40
what if someone does spc4 without the m-p strategy what then
41
ball multiplier theorem retweeted
Alexander Johansen
@AlexRoseJoAug 2
The amount of AI bros shitting on $80k/year math post docs careers being “over” is wild
18
11
2
352
21,174
The modern proof of 2D Kakeya is fun > Kakeya set E in R^2. > Thicken to E_\delta which contains a \delta-separated set of 1 x \delta tubes T > Easy to estimate lower bound on overlaps > Cauchy-Schwarz for \delta^2\log(1/\delta) bound
1
59
Davies' original proof is a paragraph but I really don't think it's a paragraph
50
Load more