K3 surfaces, numerically #
A K3 surface is a smooth projective surface with trivial canonical bundle and no first cohomology. Numerically that is exactly two conditions on the Todd class:
td₁(X) = −K_X/2 = 0;∫_X td₂(X) = χ(O_X) = 2.
IsK3 asserts those two and nothing else, so every consequence below is a consequence of
those two facts about the Todd class, not of anything else one knows about K3 surfaces.
Main results #
chi_eq—χ(E) = 2·rank(E) + ∫_X ch₂(E).mukaiSelfPairing_eq—⟨v(E), v(E)⟩ = ∫_X Δ(E) − 2·rank(E)², relating the Mukai self-pairing to the Bogomolov–Gieseker discriminant.
The second identity is what makes the Mukai lattice and the discriminant two views of one
object: v² ≥ −2 (the condition for a moduli space to be nonempty) and ∫Δ ≥ 0
(Bogomolov) differ by exactly 2r².
Not proved here #
That the axioms have a K3 model: a concrete NumericalVarietyData 2 A N with
A = ℚ[t]/(t³), ∫t² = 2d, satisfying IsK3. Only the dimension-zero witness in
Numerical/Models/DimensionZero/Point.lean exists so far, so the statements below are
conditional on a
NumericalVarietyData 2 A N existing at all. Building that model is the next task; it needs the
internality of the grading, i.e. linear independence of 1, t, t².
The numerical signature of a K3 surface: trivial canonical class and χ(O_X) = 2.
td₁(X) = −K_X/2 = 0.∫_X td₂(X) = χ(O_X) = 2.
Instances For
Riemann–Roch on a K3 surface: χ(E) = 2·rank(E) + ∫_X ch₂(E).
The c₁-term of the surface formula drops out because td₁ = 0; this is the whole of what
triviality of the canonical class buys.
The third component of the Mukai vector v(E) = (r, c₁, s), integrated:
s = rank(E) + ∫_X ch₂(E).
For a K3, v(E) = ch(E)·√td(X) with √td(X) = 1 + [pt], so the top component of v is
ch₂(E) + rank(E)·[pt]; mukaiS is its degree.
Instances For
Riemann–Roch in Mukai coordinates: χ(E) = r + s.
The Mukai self-pairing ⟨v(E), v(E)⟩ = c₁² − 2rs.
Equations
Instances For
The Mukai pairing and the Bogomolov–Gieseker discriminant differ by 2r²:
⟨v(E), v(E)⟩ = ∫_X Δ(E) − 2·rank(E)².
So v² ≥ −2 and ∫Δ ≥ 0 are statements about the same quantity, shifted.
IsK3 is omitted deliberately: this identity is arithmetic relating two definitions and
holds on any surface. Only the reading of (r, c₁, s) as a Mukai vector is K3-specific.
A rank-zero object has ⟨v, v⟩ = ∫_X c₁². Also independent of IsK3.