Division polynomials of Weierstrass curves #
This file defines certain polynomials associated to division polynomials of Weierstrass curves.
These are defined in terms of the auxiliary sequences for normalised elliptic divisibility sequences
(EDS) as defined in Mathlib.NumberTheory.EllipticDivisibilitySequence
.
Mathematical background #
Let $W$ be a Weierstrass curve over a commutative ring $R$. The sequence of $n$-division polynomials $\psi_n \in R[X, Y]$ of $W$ is the normalised EDS with initial values
- $\psi_0 := 0$,
- $\psi_1 := 1$,
- $\psi_2 := 2Y + a_1X + a_3$,
- $\psi_3 := 3X^4 + b_2X^3 + 3b_4X ^ 2 + 3b_6X + b_8$, and
- $\psi_4 := \psi_2 \cdot (2X^6 + b_2X^5 + 5b_4X^4 + 10b_6X^3 + 10b_8X^2 + (b_2b_8 - b_4b_6)X + (b_4b_8 - b_6^2))$.
Furthermore, define the associated sequences $\phi_n, \omega_n \in R[X, Y]$ by
- $\phi_n := X\psi_n^2 - \psi_{n + 1} \cdot \psi_{n - 1}$, and
- $\omega_n := (\psi_{2n} / \psi_n - \psi_n \cdot (a_1\phi_n + a_3\psi_n^2)) / 2$.
Note that $\omega_n$ is always well-defined as a polynomial in $R[X, Y]$. As a start, it can be shown by induction that $\psi_n$ always divides $\psi_{2n}$ in $R[X, Y]$, so that $\psi_{2n} / \psi_n$ is always well-defined as a polynomial, while division by 2 is well-defined when $R$ has characteristic different from 2. In general, it can be shown that 2 always divides the polynomial $\psi_{2n} / \psi_n - \psi_n \cdot (a_1\phi_n + a_3\psi_n^2)$ in the characteristic zero universal ring $\mathcal{R}[X, Y] := \mathbb{Z}[A_1, A_2, A_3, A_4, A_6][X, Y]$ of $W$, where the $A_i$ are indeterminates. Then $\omega_n$ can be equivalently defined as the image of this division under the associated universal morphism $\mathcal{R}[X, Y] \to R[X, Y]$ mapping $A_i$ to $a_i$.
Now, in the coordinate ring $R[W]$, note that $\psi_2^2$ is congruent to the polynomial $\Psi_2^{[2]} := 4X^3 + b_2X^2 + 2b_4X + b_6 \in R[X]$. As such, in $R[W]$, the recurrences associated to a normalised EDS show that $\psi_n / \psi_2$ are congruent to certain polynomials in $R[X]$. In particular, define $\tilde{\Psi}_n \in R[X]$ as the auxiliary sequence for a normalised EDS with extra parameter $(\Psi_2^{[2]})^2$ and initial values
- $\tilde{\Psi}_0 := 0$,
- $\tilde{\Psi}_1 := 1$,
- $\tilde{\Psi}_2 := 1$,
- $\tilde{\Psi}_3 := \psi_3$, and
- $\tilde{\Psi}_4 := \psi_4 / \psi_2$.
The corresponding normalised EDS $\Psi_n \in R[X, Y]$ is then given by
- $\Psi_n := \tilde{\Psi}_n \cdot \psi_2$ if $n$ is even, and
- $\Psi_n := \tilde{\Psi}_n$ if $n$ is odd.
Furthermore, define the associated sequences $\Psi_n^{[2]}, \Phi_n \in R[X]$ by
- $\Psi_n^{[2]} := \tilde{\Psi}_n^2 \cdot \Psi_2^{[2]}$ if $n$ is even,
- $\Psi_n^{[2]} := \tilde{\Psi}_n^2$ if $n$ is odd,
- $\Phi_n := X\Psi_n^{[2]} - \tilde{\Psi}_{n + 1} \cdot \tilde{\Psi}_{n - 1}$ if $n$ is even, and
- $\Phi_n := X\Psi_n^{[2]} - \tilde{\Psi}_{n + 1} \cdot \tilde{\Psi}_{n - 1} \cdot \Psi_2^{[2]}$ if $n$ is odd.
With these definitions in mind, $\psi_n \in R[X, Y]$ and $\phi_n \in R[X, Y]$ are congruent in $R[W]$ to $\Psi_n \in R[X, Y]$ and $\Phi_n \in R[X]$ respectively, which are defined in terms of $\Psi_2^{[2]} \in R[X]$ and $\tilde{\Psi}_n \in R[X]$.
Main definitions #
WeierstrassCurve.preΨ
: the univariate polynomials $\tilde{\Psi}_n$.WeierstrassCurve.ΨSq
: the univariate polynomials $\Psi_n^{[2]}$.WeierstrassCurve.Ψ
: the bivariate polynomials $\Psi_n$.WeierstrassCurve.Φ
: the univariate polynomials $\Phi_n$.WeierstrassCurve.ψ
: the bivariate $n$-division polynomials $\psi_n$.WeierstrassCurve.φ
: the bivariate polynomials $\phi_n$.- TODO: the bivariate polynomials $\omega_n$.
Implementation notes #
Analogously to Mathlib.NumberTheory.EllipticDivisibilitySequence
, the bivariate polynomials
$\Psi_n$ are defined in terms of the univariate polynomials $\tilde{\Psi}_n$. This is done partially
to avoid ring division, but more crucially to allow the definition of $\Psi_n^{[2]}$ and $\Phi_n$ as
univariate polynomials without needing to work under the coordinate ring, and to allow the
computation of their leading terms without ambiguity. Furthermore, evaluating these polynomials at a
rational point on $W$ recovers their original definition up to linear combinations of the
Weierstrass equation of $W$, hence also avoiding the need to work in the coordinate ring.
TODO: implementation notes for the definition of $\omega_n$.
References #
[J Silverman, The Arithmetic of Elliptic Curves][silverman2009]
Tags #
elliptic curve, division polynomial, torsion point
The univariate polynomial $\Psi_2^{[2]}$ #
The $2$-division polynomial $\psi_2 = \Psi_2$.
Equations
- W.ψ₂ = W.toAffine.polynomialY
Instances For
The univariate polynomial $\Psi_2^{[2]}$ congruent to $\psi_2^2$.
Equations
Instances For
The univariate polynomials $\tilde{\Psi}_n$ for $n \in \mathbb{N}$ #
The $3$-division polynomial $\psi_3 = \Psi_3$.
Equations
Instances For
The univariate polynomial $\tilde{\Psi}_4$, which is auxiliary to the $4$-division polynomial $\psi_4 = \Psi_4 = \tilde{\Psi}_4\psi_2$.
Equations
- One or more equations did not get rendered due to their size.
Instances For
The univariate polynomials $\tilde{\Psi}_n$ for $n \in \mathbb{N}$, which are auxiliary to the bivariate polynomials $\Psi_n$ congruent to the bivariate $n$-division polynomials $\psi_n$.
Equations
- W.preΨ' n = preNormEDS' (W.Ψ₂Sq ^ 2) W.Ψ₃ W.preΨ₄ n
Instances For
The univariate polynomials $\tilde{\Psi}_n$ for $n \in \mathbb{Z}$ #
The univariate polynomials $\tilde{\Psi}_n$ for $n \in \mathbb{Z}$, which are auxiliary to the bivariate polynomials $\Psi_n$ congruent to the bivariate $n$-division polynomials $\psi_n$.
Equations
- W.preΨ n = preNormEDS (W.Ψ₂Sq ^ 2) W.Ψ₃ W.preΨ₄ n
Instances For
The univariate polynomials $\Psi_n^{[2]}$ #
The univariate polynomials $\Psi_n^{[2]}$ congruent to $\psi_n^2$.
Instances For
The bivariate polynomials $\Psi_n$ #
The bivariate polynomials $\Psi_n$ congruent to the $n$-division polynomials $\psi_n$.
Instances For
The univariate polynomials $\Phi_n$ #
The univariate polynomials $\Phi_n$ congruent to $\phi_n$.
Equations
Instances For
The bivariate polynomials $\psi_n$ #
The bivariate $n$-division polynomials $\psi_n$.
Instances For
The bivariate polynomials $\phi_n$ #
The bivariate polynomials $\phi_n$.