October 20, 2026
|
|
|---|---|
![]() |
Motes on the recent OpenAI breakthrough refutation of the Landau-Siegel zeroes hypothesis, using the polynomial method! My live lecture video going over this proof is on YouTube.
The Legendre character χ(p) := (q/p) is used only in one Freshman's Dream computation, that's all. My summary of the polynomial method in the context of this proof: I want to force a (non-zero) polynomial F (X, Y, Z) to vanish at every grid point ∈ G. The key trick is to show “sparseness” of residue patterns (of values of polynomials f on G) mod P (for some large P ), use “pigeonhole and subtraction” to get a polynomial F = f − g that takes values divisible by P on G but has small coefficients, and use essentially the “discreteness phenomenon” that the only multiple of P in [0, P) is 0.
Key geometric input to the main result (reminds me of PFR/Marton's conjecture, feels like a really classical additive combinatorics lemma):
Theorem 0.4 (Low sumset complexity forces low subspace complexity). Let 0 ∈ E ⊆ ℂ³ be finite. Suppose a nonzero polynomial F ∈ ℂ[x, y, z] of degree D vanishes on the triple sumset E + E + E. Then there exists a proper ℂ-vector subspace V ⊊ ℂ³ such that E is covered by at most Dcodim(V) translates of V. Equivalently, letting r := codim(V), |(E + V)/V| ≤ Dr or #(E mod V) ≤ Dr. Slogan. “If repeated addition doesn’t increase complexity much, then there must be hidden translation symmetry.” |
March 20, 2026
|
|
![]() |
Notes on the AKS primality testing algorithm ("PRIMES is in P"), using the polynomial method! My live lecture video going over this proof is on YouTube. The proof involves 6 steps. We consider the set of evaluations 𝒢 := {P(ζᵣ) : P ∈ 𝒫 defined in the PDF} ⊆ 𝔽, e.g. (ζᵣ + 10)(ζᵣ + 3)⁵ ∈ 𝒢 ⊆ 𝔽. The goal will be to prove (Steps 4, 5) the lower and upper bound (or “sandwich bound”) The upper bound (Step 5) will use the assumption (from our “f.s.o.c.”) that n is not a power of p. Finally, in Step 6, we will show that “good” r leads to t > (log₂ n)², which contradicts 2ᵗ ≤ n√t The “sandwich bound” on |𝒢| will follow from 2 ingredients: the polynomial method (really just the factor theorem), and some “magical commutativity” (of P(•) and •ᵉ, on ζᵣ), derived in Steps 1, 2, 3: ∀e ∈ ℰ ⊆ ℕ⁺, ∀P ∈ 𝒫 ⊆ ℤ[X], we have equality of evaluations P(ζᵣ)ᵉ = P(ζᵣᵉ) in 𝔽. |
December 14, 2025
|
|
![]() |
As tight a writeup as I could make for Roth's Fields-medal winning work on Diophantine approximation. Heavily indebted to Gaurish2016. |
July 1, 2025
|
|
![]() |
An illustrated write-up of the 2020 breakthrough on the Erdős-Rado sunflower problem, explained using the analogy of throwing pokeballs to catch pokemon. It's a beautiful proof. It's not as short as I originally envisioned (it's not much shorter than Rao's exposition and he does a lot more than what I do here...sad), but I hope that it's intuitive and that the diagram is helpful. |
April 26, 2025
|
|
![]() |
An attempt to introduce scheme theory (explain the sheaf of rings concept), and Hilbert's Nullstellensatz. The technical impetus for (affine) schemes really boils down to the following humble observation: I can always build "fractions" from a commutative unital ring (cr1ng) A (or thing=module acted on by A) --- identical to the construction building the rationals from the integers we are familiar with from childhood, for A an integral domain; and with a slight hiccup ("stabilization at sufficiently high powers") in the presence of zero divisors --- and all those (subrings of) fractions assemble together in a "sheafy" way. Where the crucial sheaf gluing property comes (as in the subject of differential geometry or topology) from a "partition of unity", which one first sees in the commutative algebra setting in the form of Hilbert's Nullstellensatz. See the Nullstellensatz section if you just care about that; it is self-contained.
I tried to make the "main story" accessible to even relatively fresh/green students (especially my "simplest/earliest/most natural substantive examples"), but some examples/sidenotes/tangents may go too far afield. Ignore the examples/remarks for which you don't even recognize the words, but try to understand those for which you do. I also spent way too much time on the model theory stuff. That wasn't even meant to be a focus. And it's probably incomprehensible anyways. Alas... |
June 6, 2024
|
|
![]() |
Notes for a (2hr) seminar talk I (and another graduate student!) gave on primes in arithmetic progressions (APs) and sieve theory, split into 3 parts: Dirichlet's 1837 theorem on the infinitude of primes in APs to introduce Dirichlet characters, especially the real/quadratic ones and the problem of "Siegel poles/zeroes" (i.e. conspiracy of a quadratic Dirichlet character with the function 1, or the Liouville function λ); an introduction to the main ideas/philosophy/"black-boxable tools" of sieve theory and its limitations (the parity problem); and Matomäki&Merikoski&Teräväinen's use of sieve theory in the Siegel pole/zero/neither cases to prove Linnik's theorem (the first prime in an AP appears in "polynomial time"). A preliminary recording is here. I will try to make a complete recording (might be ~3hrs because I want to give the material its due space to breathe, which we sadly didn't get the time to do during the actual seminar) of the material of the talk and link it here sometime soon. |
March 6, 2024
|
|
![]() |
Slides for a talk giving an exposition/introduction/overview/survey of elementary analytic number theory (introducing the Riemann zeta function and the relationship between its analytic properties/properties revealed by analytic methods, and number theoretic questions; (cheap and bona-fide) Mertens' theorems; Mobius inversion and cancellation; divisor sums and taking advantage of averaging/Fubini; and providing what I think to be a very well-motivated approach to the Selberg symmetry formula, Brun-Titchmarsh and its factor of 2 related to the parity problem, etc.).
Also I had a "hidden agenda" of introducing some important "tricks/themes" from analysis, like dyadic decomposition; big-O/little-o notation and the idea that away from the main term we can actually be much looser than what one may be used to in one's standard "intro analysis training" (and the IMHO unexpected phenomenon of "discrete resolution jumps", e.g. a O(1) error term might resolve further to c+O(1/log x), where there's a discrete gap between constant-order resolution and 1/log(x)-order resolution); and the flexibility of flowing between sums/integrals/measures. Culminates in Terry Tao's proof of the prime number theorem (PNT) using the language/machinery of Banach algebras. |
Febuary 12, 2024
|
|
![]() |
Notes on a talk about the lace expansion method in the analysis of self-avoiding random walks (SAW) I (and 2 other graduate students!) gave during a Winter2024 seminar on perturbative methods in probability. |
August 26, 2021
|
|
![]() |
Brief overview of Dirichlet's theorem on primes in arithmetic progressions (truly baffles me to know that Dirichlet gave a completely rigorous proof of this in 1837, more than twenty years before Riemann's foundational work in extending the zeta function to the complex plane in 1859; compare this with Galois's discovery of the Galois group in 1830... just awe-inspiring that in one paper someone can lay the foundation of an entire branch of mathematics). Full of links to resources (notes/articles) to learn more, mixed with some commentary. (Feb. 7, 2022 uploaded newer version; old version found here) |
June 11, 2021
|
|
![]() |
My senior thesis, i.e. the apotheosis of my undergrad career, about providing a motivation-rich exposition to topology. In my opinion, I cover an absurd amount of material considering the length of the document, but topology has a way of drawing in a huge wealth of ideas. We will start with some basic definitions, and quickly move into motivating the development of abstract topological spaces by thinking about ways of defining local and global continuity of functions without the notion of distance. We will follow a narrative that meanders around ideas like partitions of unity, some theorems of Urysohn, Tietze's extension theorem, the countability and separation axioms, metrization, compactness, ordinals (yes, some wacky set theory! ...but perhaps not as surpising considering the connection between compactness and AC via Tychonoff), topological dimension, and manifolds, ultimately culminating in a proof of a version of Whitney's embedding theorem. There's some fun bonus material at the end regarding Brouwer's fixed-point theorem (via a combinatorial result, Sperner's lemma!) and deep results that typically come from algebraic topology, like the invariance of dimension and the Jordan curve theorem. (2024: the section on JCT was presented by me in this video.) |
Aug 10, 2020
|
|
![]() |
Should've posted this earlier, but oh well. This paper is an exposition of the Longest Increasing Subsequence (LIS) problem and the Baik-Deift-Johansson theorem, as well as important results about the Tracy-Widom (TW) distribution and the problems/contexts in which it appears. As an added bonus, I ran some computer programs to try to empirically determine the rate of convergence of the length of the LIS to the TW-distribution, which I found seemed to be on the order of n-1/6. This result/conjecture seems to not have appeared before in the literature. |
Jun 3, 2020
|
|
![]() |
An introduction to measure theory, building up to the Caratheodory extension theorem. I've long thought that the Caratheodory measurability criterion was very unintuitive, and very mysterious, so I spent 20 pages trying to demystify it in the context of the real line and Lebesgue measure. I believe that mathematics should be a journey of discovery, and should be communicated as such; one should play around and tinker before making definitions. I attempt to do this in this paper, where I go over outer measure, inner measure, Lebesgue's definition of measurability, the open-closed/inner-outer definition of measurability, Caratheodory's definition (now intuive in the context of the other definitions), sigma-algebras (and other set structures, like semi-rings), and finally the extension theorem itself. My magnum opus of my life so far. |
![]() |
The entire collection of homeworks I did for MATH 522. This class continually gets me to produce some of my highest quality work, which I'm very happy about, but also kind of stressed about; a blessing and a curse. In fact, my 33X homeworks kind of decreased in quality because (a) there were too many problems to think deeply on and (b) they weren't particularly satisfying or interesting to me, so I will not post them :( |
![]() |
The entire collection of homeworks I did for MATH 523. Only 4 homeworks and a midterm, but the rest of the work I did for this class lies in the Longest Increasing Subsequence paper I have yet to complete/upload. |
Dec 30, 2019
|
|
![]() |
Every wonder where in the world did the formulas for dots and cross products come from? Why didn't we just define vector multiplication like we did vector addition? Or what about determinants -- did you ever want to know how someone came up for the formula for the determinant? Well, now you can find out. Part 1 of "Math on a Desert Isle". |
Dec 6, 2019
|
|
![]() |
This is the entire collection of homeworks I did for MATH 334 (arranged last to first); the first homeworks focused on point-set topology and we slowly moved on to continuity, differentiation, and integration. Overall I'm pretty happy with it (especially the last and first homework sets), but it certainly was a struggle. |
![]() |
Similarly, this is the entire collection of homeworks I did for MATH 521, my first graduate class (though I audited), on a subject I had no prior experience in (probability/measure theory and statistics). The homeworks were really hard but a lot of fun -- I think I learned a lot of cool and beautiful stuff. I only managed to do about a half of the homework every week, but I'm pretty happy with the stuff I was able to do. |
May 25, 2018
|
|
![]() |
This paper is about very unintuitively diverging series, culminating in the proof that the sum of the reciprocals of the primes is infinite, by using the sum of the reciprocals of the natural numbers and the sum of the reciprocals of the squared numbers |

