Documentation

TauCeti.AlgebraicGeometry.EllipticCurve.DivisionPolynomial.Torsion.Roots

The roots of ΨSqₙ are the abscissae of the nonzero n-torsion #

ΨSqₙ is the square of the n-division polynomial, pushed down to a polynomial in x alone. Its roots are exactly the x-coordinates of the affine points killed by n: one direction holds over any field, the other needs the base field algebraically closed, so that the y completing a root to a point exists.

This is the dictionary the n-torsion is counted through — the kernel of [n] maps to the roots of ΨSqₙ two-to-one away from the 2-torsion, which is what matches #ker [n] = n ² against deg preΨₙ.

Main results #

References #

Over an algebraically closed field the roots of ΨSqₙ are exactly the abscissae of the n-torsion points. Solving the Weierstrass equation for y gives a point over a root, and ΨSqₙ(x) = 0 makes ψₙ vanish there, which annihilates it; the converse needs no closure, since the y is supplied.

Over an algebraically closed field the n-torsion is trivial exactly when ΨSqₙ has no root: a root is the abscissa of a nonzero affine n-torsion point, and the point at infinity is the only point with no abscissa.

Characteristic two and three #

In characteristic 2 the 2-torsion of an elliptic curve over an algebraically closed field is trivial exactly when a₁ = 0: then ΨSq₂ = a₃² is a nonzero constant, and otherwise x = a₃ / a₁ is a root of ΨSq₂ = a₁² x² + a₃².

In characteristic 3 the 3-torsion of an elliptic curve over an algebraically closed field is trivial exactly when b₂ = 0: there ψ₃ = b₂ x³ + b₈, whose constant term b₈ cannot vanish together with b₂, and which has a root as soon as b₂ ≠ 0.