Semi-stable models, local heights and quadratic Chabauty for X 0(N)*
summary
The gist
Quadratic Chabauty computations for X 0(N)* are complicated by local height contributions at primes of bad reduction…
In short
The paper constructs a global p-adic height for $X_0(N)^*$ for squarefree levels $N > 714$ by managing local height contributions at bad primes. It achieves this by analyzing the minimal regular model of the curve and using linear conditions derived from its components. This allows for explicit quadratic Chabauty computations to find rational points on $X_0(N)^*$.
Key concepts
- p-adic height
- A function defined over a number field that measures the arithmetic complexity of points on an algebraic variety, specifically related to the prime $p$. The paper constructs a global version where local contributions from primes of bad reduction are controlled, which is essential for applying Chabauty methods.
- Minimal Regular Model
- This is the simplest algebraic structure (a model) that describes the curve $X_0(N)^*$ at primes where it has 'bad' behavior. Understanding this model helps in controlling the local height contributions, allowing researchers to enforce vanishing contributions for squarefree levels.
- Quadratic Chabauty Computation
- A method used to find rational points on an algebraic curve by using $p$-adic analysis, specifically by bounding the number of points that can exist based on their height. The paper uses the constructed p-adic height to perform these computations explicitly for $X_0(N)^*$.
- Dual Graph $\mathcal{G}_0(N)^*$ and Loop Counts
- This graph represents the structure of the curve's points related to Hecke operators. Analyzing its properties, such as the number of 'long loops,' helps in bounding the genus and controlling other arithmetic invariants necessary for establishing conditions where p-adic heights are guaranteed to exist.
Terminology used across episodes
This episode discusses
- Semi-stable models, local heights and quadratic Chabauty for X 0(N)* · Paper Radio
- Hyperelliptic Atkin-Lehner quotients of Shimura curves
- The local theory of unipotent Kummer maps and refined Selmer schemes
- Rational points on X 0(N)* when N is non-squarefree
The paper
Semi-stable models, local heights and quadratic Chabauty for X 0(N)* · Read on arXiv
University of Zagreb · Croatian Science Foundation · European Union
Quadratic Chabauty computations for X 0(N)* are complicated by local height contributions at primes of bad reduction. For squarefree N, we explain a strategy for constructing a global p-adic height for which all the contributions at the primes of bad reduction vanish. As our first main result, we show that such a p-adic height exists for all squarefree levels N>714. In order to do this, we first give an explicit description of the minimal regular model of X 0(N)* at primes of bad reduction, and show this model is semi-stable. Each component of this model gives rise to a linear condition which ensures that the contribution at that component vanishes. Using embeddings of quadratic orders into quaternion algebras, we obtain an upper bound for the number of these conditions; when the genus of X 0(N)* is greater than one more than this bound, there is enough freedom to choose a suitable correspondence, simplifying quadratic Chabauty computations. As an application, we determine the rational points on several curves X 0(N)* for which this was not done before.
Transcript
Introduction to the show: ident: Mathematics. The day in mathematics: new theorems, new proofs, and what they open.
Ada: I'm Ada.
Emmy: And I'm Emmy, here with Kolya.
Ada: We're looking at the paper: "Semi-stable models, local heights and quadratic Chabauty for X 0(N)*".
Emmy: Here's the gist of the paper. Quadratic Chabauty computations for X 0(N)* are complicated by local height contributions at primes of bad reduction, necessitating a strategy to construct a global p-adic height where these contributions vanish for squarefree levels N.
Ada: The primary result establishes the existence of such a p-adic height for all squarefree levels N > 714, achieved by describing the minimal regular model of X 0(N)* at primes of bad reduction, demonstrating its semi-stable nature, and employing linear conditions derived from its components to enforce vanishing contributions.
Emmy: An explicit, model-free algorithm (Proposition 1.5) is provided for constructing these heights when their existence is guaranteed, which directly enables explicit quadratic Chabauty computations using QCMod for determining rational points on X 0(N)*.
Ada: The existence of p-adic heights supported only at a specific prime p is fundamentally linked to bounding the genus g(X 0(N)*) and the number of long loops, L(N), in the dual graph G 0(N)*.
Title and authors: Ada: So we're looking at this paper, "Semi-stable models, local heights and quadratic Chabauty for X zero(N)*", which tackles the headache of local height contributions when doing quadratic Chabauty computations on X zero(N)* <ref:2609.40207#pg2>. It points out that those contributions usually complicate things at primes of bad reduction.
Emmy: Exactly. The paper explains a strategy to build a global p-adic height where those local contributions just disappear, especially for squarefree levels N. It achieves this by first describing the minimal regular model of X zero(N)* at those bad primes and showing that this model is semi-stable <ref:2609.40207#pg2>.
Kolya: So it’s about making the local stuff vanish so we can do a global computation. But what's the catch? The paper says they only get this guarantee for squarefree levels N greater than seven hundred fourteen which is a pretty specific starting point for applying this strategy <ref:2609.40207#pg1>.
Ada: That's right. The first main result they show is that such a p-adic height exists for all squarefree levels N bigger than seven hundred fourteen and they do this by looking at how the minimal regular model behaves at primes of bad reduction <ref:2609.40207#pg1,that such a p-adic height exists for all squarefree levels $N>.
Emmy: And the mechanism behind it involves each component of that model giving rise to a linear condition. These conditions are what ensure that the contribution from each bad prime vanishes when you choose your correspondence correctly.
Kolya: So, if we look at the structure, this means that for every bad prime, there's a condition we have to satisfy, and they get an upper bound on how many of these conditions there can be.
Ada: That upper bound comes from looking at embeddings of quadratic orders into quaternion algebras. And then they tie the existence of enough freedom to choose a correspondence to the genus of X zero(N)* <ref:2609.40207#pg2>.
Emmy: Specifically, when the genus of X zero(N)* is greater than one more than that upper bound, they have enough choices to pick a suitable correspondence that makes all those local height contributions vanish <ref:2609.40207#pg2>.
Kolya: That sounds like a big deal for computation because it moves us away from having to calculate those messy local terms directly. But how do we know when the genus is actually big enough?
Title and authors: Ada: Well, they establish some bounds on the number of long loops in the dual graph of X zero(N)* <ref:2609.40207#pg2>. They link these loop counts to fixed point counts, and they show a lower bound for the genus.
Emmy: Specifically, Corollary six point one three gives us a genus lower bound based on the sum of fixed point counts nu(N, d). This formula involves counting how many supersingular points there are in characteristic d.
Kolya: So the idea is that if this genus lower bound is high enough relative to those loop counts, we can guarantee that there's a suitable correspondence available. But what about the cases where it's not big enough?
Ada: They have a key criterion for guaranteed vanishing contributions. It’s when the genus of X zero(N)* is greater than one more than the bound derived from those loop counts <ref:2609.40207#pg2>. This is related to a quantity phi(N).
Emmy: Lemma seven point three gives us a quantitative relationship here for squarefree N bigger than six, showing that if you set specific constants, the difference between the genus lower bound and that bound is at least some function phi(N) <ref:2609.40207#pg2>.
Kolya: So they are saying that if this phi(N) is positive, then the genus will definitely be greater than one more than B(N), which is related to those loop counts. That seems like a solid way to ensure the conditions are met.
Ada: The paper proves a stronger version of this idea in Theorem seven point five for levels N bigger than N zero which is two million five hundred eighty-nine thousand five hundred ten defined as the largest squarefree level where phi(N) is less than or equal to zero.
Emmy: When they say N is greater than N zero it means that for all those levels, phi(N) must be positive because N zero was chosen as the largest one where it wasn't <ref:2609.40207#pg2>. That forces the genus condition to hold for all squarefree levels above that threshold.
Kolya: So they’ve effectively found a hard cutoff, N zero above which this method is guaranteed to work without worrying about the genus bound failing <ref:2609.40207#pg2>. But what about smaller squarefree N ?
Ada: For those smaller cases, like the three specific examples of one hundred eighty-seven two hundred forty-seven and three hundred nineteen mentioned in Proposition one point six, they still managed to determine the rational points using their techniques for local heights <ref:2609.40207#pg3,to determine the rational points>.
Title and authors: Emmy: That shows that even when the genus is not necessarily large enough to guarantee a general correspondence, you can still find non-trivial correspondences with supported heights on those curves. It’s about finding specific points rather than a universal construction.
Kolya: I wonder if this means we can actually use this for computing points on curves where the genus is low, even if we can't automate the whole process yet, like they mentioned with QCMod and genus eight curves <ref:2609.40207#pg2>.
Ada: That's the practical application they are pointing towards. They suggest that in principle, using QCMod to determine rational points on X zero(N)* should be feasible for hundreds of squarefree levels N where the genus is less than or equal to eight, even if it isn't automated yet <ref:2609.40207#pg2>.
Emmy: So the core idea is that semi-stable models give us the necessary structure to relate local heights to graph properties, and then those graph properties allow us to enforce linear conditions that simplify the quadratic Chabauty problem.
Kolya: The paper also gives explicit formulae for counting supersingular points in characteristic using Eichler's trace formula. That helps with the input data for all these genus bounds we discussed earlier.
Ada: And finally, they show that the number of these heights is at least the genus minus some error term that grows slower than any fixed positive power of N. This means as N gets huge, you are guaranteed to find enough independent heights to solve the problem.
Emmy: So in summary, this paper on "Semi-stable models, local heights and quadratic Chabauty for X zero(N)*" provides a construction method for p-adic heights that works well for squarefree levels N above a certain point, using semi-stable models to control bad prime contributions <ref:2609.40207#pg2>.
Kolya: It seems like the main win here is having an explicit algorithm, Proposition one point five, for constructing these height polynomials when existence is guaranteed, which directly lets us use QCMod to find rational points on X zero(N)* <ref:2609.40207#pg2>.
Ada: And that's the connection. The paper shows how to bridge the gap between complex local geometry and concrete point counting on these modular curves. It’s a lot of machinery, but it’s all pointed toward making explicit computations possible where they weren't before.
The paper's summary: Ada: So, to wrap up what we just heard about the paper "Semi-stable models, local heights and quadratic Chabauty for X zero(N)*," it’s really about finding a way to do those hard quadratic Chabauty calculations without getting bogged down by those messy local height contributions at primes of bad reduction.
Emmy: Exactly. They showed that you can construct a global p-adic height where all those local problems just disappear, provided you stick to squarefree levels N greater than seven hundred fourteen.
Kolya: So what’s the catch on that guarantee? It turns out they build this existence proof by looking at the minimal regular model of X zero(N)* at those bad primes and showing it has a semi-stable structure.
Ada: Right, and that semi-stable structure is key because it lets them derive linear conditions from its components, and those conditions are what force the local height contributions to vanish when you pick the right correspondence.
Emmy: It’s like they map out the geometry of these curves so you know exactly where to look for those vanishing points in a global sense. They use these conditions to build a set of polynomials that define those heights.
Kolya: But how do they know when it actually works? They connect the existence of enough choices for your correspondence to the genus of X zero(N)* and how many long loops are in its dual graph structure.
Ada: They establish a genus lower bound based on counting fixed points, and if that lower bound is high enough relative to the loop counts, they have guaranteed enough freedom to find those vanishing contributions.
Emmy: And they found a specific threshold, N zero which is two million five hundred eighty-nine thousand five hundred ten. For any squarefree level N bigger than that number, this method guarantees the existence of the height you need.
Kolya: So for the listener who just wants to know what this means practically, it’s that for large enough levels, they have a solid algorithm to find rational points on these curves using something called QCMod.
Ada: Precisely. They give you an explicit model-free algorithm that lets you construct a sequence of polynomials which yield those linearly independent p-adic heights supported only at one specific prime p.
Emmy: That means for the listener, the implication is that they can actually use these tools to explicitly determine rational points on X zero(N)* for several previously unsolved values of N, even when the genus isn't huge.
Kolya: And they showed that by counting those fixed points and using certain lemmas, you get a lower bound on the number of these heights that grows slower than any fixed positive power of N.
Ada: Which means as N gets really large, you are guaranteed to find enough independent heights to solve the problem without running into infinite local problems.
Emmy: So for anyone interested in how this connects the topology of a curve to number theory computations, it’s about using semi-stable geometry to tame complicated local behavior and turn it into a solvable problem for point counting.
Kolya: This method also gives them tools to count those supersingular points in characteristic l using Eichler's trace formula, which feeds directly into those genus bounds we discussed earlier.
Ada: And the structure of these semi-stable models is what connects everything—from the dual graph properties to the linear conditions needed for Algorithm four hundred thirteen. It’s a very interconnected argument.
Emmy: So, while it's quite technical, the core idea is that by understanding how these curves behave locally at bad primes and relating that to global graph structure, you can build a robust machinery for finding points on them.
Kolya: This work sets up a clear path forward for quadratic Chabauty computations on modular curves when the input levels are squarefree and large enough.
Ada: And it opens the door for using these explicit constructions with QCMod to find rational points where they were previously stuck because of those nasty local height terms.
The paper's improvements: Ada: We just talked about how they built that existence proof for p-adic heights on X zero(N)*, and now we're going to look at what the authors suggest next to make that method even better.
Emmy: They’ve already laid out a solid framework, but they aren't stopping there; they are suggesting several improvements to strengthen the construction of these height polynomials.
Kolya: What kind of improvements are we talking about? Are they just tweaking the constants or changing how the genus bound relates to those loop counts?
Ada: They’re focusing on making sure that when you use that algorithm from Proposition one point five, you can automate the construction of a sequence of polynomials f one through fn.
Emmy: So instead of just proving existence, they want an actual recipe for generating linearly independent p-adic heights supported at prime p.
Kolya: That’s interesting for computation because it moves us from a theoretical guarantee to a functional tool. How does that sequence of polynomials scale? Does it stay manageable?
Ada: They state that this automated construction is possible using the same model-free approach they used before, but now applied specifically to generating those heights.
Emmy: It’s about turning the structural information from the semi-stable model directly into a practical computational sequence for point determination. It makes the whole process more explicit.
Kolya: If you can automate that part, it helps with the question of whether these constants they found survive when we actually apply them to curves where N is even larger.
Ada: Exactly. The hope is that by having this automated sequence, we can move past just knowing a height exists and start actually finding those rational points on X zero(N)* for specific values of N.
Emmy: It’s an effort to bridge the gap between the geometric structure they discovered and the actual computation of algebraic points.
Kolya: So, if this automated sequence works, it should give us a concrete way to test their bounds on those loop counts and fixed points against real data.
Ada: Right. And by making the construction more explicit, they’re also setting up a better way to check the sufficiency of that genus condition we talked about earlier.
Emmy: It suggests that this paper isn't just a statement about existence, but a blueprint for how to actually compute these points using modern computational tools.
Kolya: So the next step is moving from proving it can be done to providing the actual code or algorithm to do it.
Conclusion: Tom: So we’re wrapping up this discussion on "Semi-stable models, local heights and quadratic Chabauty for X zero(N)*," which was all about building a robust method to find rational points on these modular curves by controlling those tricky local height contributions.
Ada: Exactly. The authors showed that by analyzing the semi-stable model of X zero(N)* at bad primes, you can build a global p-adic height where those local issues just vanish for squarefree levels N bigger than seven hundred fourteen.
Emmy: It’s about using the geometry of the curve—specifically its semi-stable structure—to derive linear conditions that enforce that vanishing behavior when you choose your correspondence correctly.
Kolya: So, what does this mean for us in terms of actual computation? Does it really simplify the quadratic Chabauty problem?
Ada: It does because they provide an explicit, model-free algorithm for building a sequence of polynomials f one through fn that yield those linearly independent p-adic heights supported at a specific prime.
Emmy: That means we’re not just guessing; we have a systematic way to generate the necessary height information when the existence conditions are met.
Kolya: If you can automate that construction, it helps with the tail problem—the question of whether those constants actually survive when we apply them to curves with much larger N.
Ada: They’ve established bounds showing that for levels N bigger than two million five hundred eighty-nine thousand five hundred ten the genus is always large enough to guarantee that we have this necessary freedom.
Emmy: This work really connects the structure of algebraic geometry, like semi-stable models, directly to number theory problems involving point counting on these curves.
Kolya: I just want to ask about the limitations; what does this method not cover? Where does it stop working?
Ada: The main caveat is that while they guarantee existence for large N, the methods they use are most effective when N is squarefree, and their specific bounds rely on certain structural properties of the model.
Emmy: And looking ahead, the improvements they suggest point toward a fully automated routine for finding those heights, which would be a huge step toward using this method routinely.
Kolya: I think if they nail that automation part, it could give us a much better way to probe the distribution of rational points on these curves.
Ada: So we’ve seen how they use the geometry of semi-stable models to tame local contributions and build tools for explicit computations with X zero(N)*.
Emmy: It’s a solid piece of work showing how deep structural understanding can lead to practical results in algebraic number theory.
Kolya: Moving forward, we'll be watching how they use this new automated construction to test those bounds on the tail probability for these height distributions.
More episodes
- 2508.18266-On models of affine arithmetic
- 2508.20749-Rates of convergence for extremal spacings in Kakutani's random interval-splitting process
- 2509.05860-Concentration Inequalities for Branching Random Walks
- 2509.19309-Small Prime Primitive Roots in Arithmetic Progressions
- 2509.21830-Noncollapsing for Curvature Flows with Inhomogeneous Speeds
- 2510.08346-Separability and Submetrizability in Locally Convex Spaces
- 2510.27290-Borel Combinatorics of Schreier Graphs of Z-actions
- 2511.13500-Sampling Density for Gabor Phase Retrieval
- 2511.19161-Linear dynamics of random products of weighted shifts
- 2511.21548-Metastability of diffusion processes in narrow tubes