A computer-assisted proof of Krzy.z's conjecture for the sixth coefficient
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Mathematics. The day in mathematics: new theorems, new proofs, and what they open.
Ada: We're looking at the paper: "A computer-assisted proof of Krzy.z's conjecture for the sixth coefficient".
Emmy: Here's the gist of the paper. Every holomorphic function f(z) = Pakz k from the unit disk to the punctured disk satisfies a6 ≤ 2/e.
Ada: The Core Result The main result settles the full-degree question for n = 6.
Emmy: This leads to the conclusion that Krzy˙z’s conjecture holds for n = 6, meaning M6 = 2/e and the extremal functions are exactly ζf6(εz), ζ = ε = 1.
Ada: Analytic Reduction and Extremal Structure The proof combines the recent solution of the Agler–McCarthy entropy conjecture with a finite computation that excludes extremal functions having fewer than six atoms in their Herglotz representation.
Title and authors: Ada: So, we’re looking at this paper now, "A computer-assisted proof of Krzy.z's conjecture for the sixth coefficient." It seems like a pretty big statement on how well we understand these functions from the unit disk to the punctured disk.
Emmy: Yeah, and it involves Frank Wikstrom doing a computer-assisted proof for this thing. It’s about bounding that sixth coefficient of these functions, which is always less than or equal to two over e.
Ada: Exactly. The main implication here is that they’re not just proving a bound; they're showing how to rigorously find the functions that actually hit that limit, which they call the extremal functions.
Kolya: So what does this mean for us probabilists? If we can pin down these extremal structures, it gives us concrete bounds on error terms in certain statistical models.
Ada: Right. It’s about taking a general conjecture and turning it into a verifiable search problem that we can solve computationally. It moves it from just theory to something you can check with a computer algorithm.
Emmy: And the authors are using techniques from things like the Agler and McCarthy entropy reduction to make sure they only have to check configurations that actually matter for finding that equality case, which is key.
The paper's summary: Ada: So, basically, they’re taking the general setup—where a function is represented by a Herglotz class—and breaking it down into cases based on how many atoms that function has in its representation.
Emmy: They show that for any degree of functions, if you can find an extremal function with fewer than six atoms, you can rule out the possibility of an extremal function with six or more atoms being the one causing the equality.
Ada: It’s this reduction—if we prove we can't get equality from functions with fewer than six atoms, then any remaining extremals must have exactly six atoms.
Kolya: From a probabilistic view, that means if you are looking at a random variable in a certain class, the probability of hitting the absolute maximum bound is governed by these specific configurations involving those six atoms.
Emmy: And they use this reduction to handle cases like degree three, four, and five independently. They show that even though there are general analyses for any degree n, they don't need to do all that case-by-case work because this method handles it automatically.
Ada: It really simplifies the whole landscape by providing a single computational framework that works across different degrees of these functions.
The paper's improvements: Ada: Now, they suggest ways to make this computation better. They point out that a direct search over all six-atom configurations would be too big, but they use the entropy theorem to narrow that down significantly before they even start the heavy search.
Emmy: And they also mention how near those exact equality functions, the gap between what you actually find and two over e gets really tiny, about zero point zero three five <ref:2610.15206#pg2>. So you need very tight bounds to distinguish an extremal function from something just slightly below the limit.
Ada: That margin is where the computer-assisted branch-and-bound comes in. They use explicit first-order conditions, which are basically how you check if a configuration is truly an extremum, not just a point on the boundary.
Kolya: And they also mention this convexity argument for the Poisson integral. That gives them another way to get a bound on the total mass involved, which helps keep their search space from getting too wild.
Emmy: So it’s not just one trick; they combine these different mathematical tools—the entropy reduction, the first-order conditions, and that convexity argument—to build a robust exclusion method.
Conclusion: Ada: So, we’re wrapping up this discussion on "A computer-assisted proof of Krzy.z's conjecture for the sixth coefficient." The main thing is that they’ve used a computer to systematically exclude all configurations that could be extremal unless they have exactly six atoms.
Emmy: It confirms that the conjecture holds for the sixth coefficient, meaning you can rigorously identify those specific functions where the bound two over e is actually reached.
Kolya: From my side, it means we have a clearer picture of the tail behavior governed by these specific six-atom structures in related probability problems.
Ada: Yeah, and they also showed this method works independently for degrees three, four, and five as well, which is a nice piece of structural confirmation.
Emmy: It’s a solid computational framework that proves that the structure of these extremal functions is limited to six atoms for the sixth coefficient problem.
Lund University
math.CV
Submitted: 2026-10-10
Updated: 2026-11-04
Comments: 18 pages
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Importance score: 92/100
The gist: Every holomorphic function f(z) = Pakz k from the unit disk to the punctured disk satisfies a6 ≤ 2/e.
Key concepts
- Herglotz representation
- This describes how a specific type of holomorphic function can be broken down into simpler building blocks, specifically functions with non-negative real parts. This decomposition is crucial for analyzing the structure of the function and applying mathematical tools to bound its coefficients.
- Branch-and-Bound
- This is a computational search method used to find an optimal solution by systematically exploring a large space. The process involves dividing the search area into smaller boxes and testing them against specific conditions. It helps in efficiently excluding configurations that cannot be the extremal ones.
- First-order conditions
- These are mathematical inequalities derived from classical variational principles and theorems like Fejér–Riesz. They provide necessary requirements that any function must satisfy to be considered an extremal function, helping to narrow down the search space significantly.
- Total mass (t)
- This is a quantity calculated from the coefficients of a function, representing a measure of its complexity or size within the context of the problem. The search uses bounds on this total mass to exclude configurations that violate the required coefficient bound.
Terminology
Summary
Every holomorphic function f(z) = Pakz k from the unit disk to the punctured disk satisfies a6 ≤ 2/e. p. 0
The Core Result
The main result settles the full-degree question for n = 6. This leads to the conclusion that Krzy˙z’s conjecture holds for n = 6, meaning M6 = 2/e and the extremal functions are exactly ζf6(εz), ζ = ε = 1. p. 1
Analytic Reduction and Extremal Structure
The proof combines the recent solution of the Agler–McCarthy entropy conjecture with a finite computation that excludes extremal functions having fewer than six atoms in their Herglotz representation. Every f ∈ B0 can be written f = e−g with g in the Herglotz class H of holomorphic functions with Re g ≥ 0. p. 1
Exclusion via Branch-and-Bound
The search excludes configurations that could be extremal by seeking either a strict coefficient bound below 2/e or a contradiction to one of these conditions. Two ingredients make these exclusions effective: first, Theorem 4. p. 22 gives an explicit form of the first-order conditions for every N ≤ n, derived from the classical variational inequalities [5, 15] and the Fej´er–Riesz theorem. p. 2 Second, a convexity argument for the Poisson integral bounds the total mass t = P j λj: Proposition 5. p. 22 gives a6 < 2/e for t ≥ 6. p. 2989.
Computational Verification
The computation proves that no Mn-extremal function has exactly N atoms for N < n by showing that every box of a cover of RN is closed by one of the tests T, S, P, A, V.
Outlook
The runner-up for n = 7 has six atoms and a gap of 0.0298. Degree-three Taylor models, and the relations (5) for N ≤ n − 2 with the σi as extra unknowns, could reduce the tree. p. 0
References
[1] J. Agler and J. p. 0 E. McCarthy, The Krzy˙z conjecture and an entropy conjecture, J. p. 0 Anal. Math. 144 (2021), 207–226
[5] J. A. Hummel, S. Scheinberg and L. p. 1 Zalcman, A coefficient problem for bounded nonvanishing functions, J. p. 0 Anal. Math. 31 (1977), 169–190
[7] S. L. Krushkal, Hyperbolic distances, nonvanishing holomorphic functions and Krzy˙z’s conjecture, arXiv:0908. p. 02587 (2009)
[12] J. Lei and T. Zhang, Proof of the Agler–McCarthy entropy conjecture, arXiv:2605. p. 003949 (2026)
[13] S. F. MacDonald, Structural aspects of extremal functions in the Krzy˙z conjecture, arXiv:2605. p. 008411 (2026)
[18] N. Samaris, A proof of Krzy˙z’s conjecture for the fifth coefficient, Complex Var. p. 0 Theory Appl. 48 (2003), 753–766
[20] F. Wikstr¨om, Implementation of the branch-and-bound for the sixth Krzy˙z coefficient, supplementary material, in [22]
[21] F. Wikstr¨om, Rigorous enclosures in the branch-and-bound for the sixth Krzy˙z coefficient, supplementary material, in [22]
[22] F. Wikstr¨om, Software and logs for “A computer-assisted proof of Krzy˙z’s conjecture for the sixth coefficient”, Zenodo (2026), doi:10. p. 05281/zenodo.23288285
[6] F. Johansson, Arb: efficient arbitrary-precision midpoint-radius interval arithmetic, IEEE Trans. p. 8 Comput. 66 (2017), 1281–1292
Pruning redundant reasoning tokens
The search uses a box-wise version of Proposition 5.2 to exclude configurations where the coefficient bound is not met. p. 2 p. 2
How the training signal is built
The search is parallel: the work is split into jobs, boxes that are explored depth first by one thread and split further when other threads are idle. The leaves within each job are not logged individually: their coverage rests on the depth-first search, about 30 lines of C that push both halves of every bisected box. p. 2
Where the gains come from
The first-order conditions cannot do this alone: numerical exploration finds five-atom configurations that satisfy them with a6 < 2/e. The search combines them with the requirement a6 ≥ 2/e (tests P and V). p. 9
The training signal is built
The function of w is inner, so bk(λ) ≤ 1.
The phase gives (6), while the modulus gives u(0) 2 = e−t/ Q jσi. Since v(e iφ) = u(e−iφ) = u(0) Q j1 − σie−iφ, this is (4). p. 6
Bounds for the total mass
The search uses a box-wise version of Proposition 5.2 to exclude configurations where the coefficient bound is not met.
The phase gives (6), while the modulus gives u(0) 2 = e−t/ Q jσi. Since v(e iφ) = u(e−iφ) = u(0) Q j1 − σie−iφ>. p. 6
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Precision in extremal function identification: The system can now rigorously exclude configurations that could be extremal by checking both |a6| >= 2/e and KKT conditions, as described in Section 7.3 of the paper.
-
Automated verification of full-degree property: The AI system can confirm the full-degree property for known cases n = 3, 4, and 5 independently of case analyses by running the same computation used to prove it for n = 6 across all these degrees.
-
Enhanced search efficiency via symmetry exploitation: The system can utilize Section 2.5's symmetries to reduce the parameter space searched, specifically by fixing alpha1 = 1 and imposing constraints like theta2 + thetaN <= 2pi, thereby avoiding redundant computations in the branch-and-bound process.
-
Robust handling of numerical enclosures: The improved system can rely on explicit error bounds from [21] for floating-point enclosures and certified constants from ball arithmetic scripts to ensure the validity of its exclusion tests rather than relying solely on agreement with sampled values.
-
Detailed analysis of local maxima: The system can perform a Hessian check modulo rotation to verify if a found configuration is a local maximum, as described in Section 7.3, ensuring that the search is not misled by non-maximal points.
Sources
- Hyperbolic distances, nonvanishing holomorphic functions and Krzyz's conjecture
- Hyperbolic geodesics, Krzyz's conjecture and beyond
- A new look at Krzyz's conjecture
- Proof of the Agler--McCarthy entropy conjecture
- Structural aspects of extremal functions in the Krzy\.z conjecture
Related papers
- Resonant saddles, tangencies with small spheres, and the characteristic flow
- Local Characterizations of Oka Manifolds
- A new strong rigidity phenomenon for the Bergman metric
- Exponential Green energies, quantization, and minimizers on compact Riemann surfaces
- Multiplicity-Weighted Moments of a Two-Monomial Harmonic Map Beyond the Caustic
- Kobayashi visibility of strictly convex domains