The derivative ideal under one blow-up: the cusp x² + y³ #
For X = 𝔸², I = (x² + y³), m = 2, the blow-up of the origin and j = 1 (a plane-curve
instance in the spirit of [Kol07, Example 11]): D(I) = (x, y²); in the chart x = x₁ y₁,
y = y₁ of the blow-up, I₁ = π_*^{-1}(I, 2) = (x₁² + y₁), J₁ = π_*^{-1}(D(I), 1) = (x₁, y₁),
and D(I₁) = (x₁² + y₁, x₁, 1) = (1); so J₁ ⊆ D(I₁), which is [Kol07, Theorem 76]
(Π_*^{-1}(D^j(I, m)) ⊆ D^j Π_*^{-1}(I, m)) for one blow-up and j = 1, and the inclusion is
strict.
The computation is done on the local model of the blow-up at the origin: R a regular local
ℚ-algebra with a two-dimensional coordinate system c (x₀ = x, x₁ = y), the center
P = chartCenter c.x 1 = (x₀, x₁), the chart ring R' = R[x₀/x₁] of the coordinate x₁
(y₀ = x₀/x₁ is Kollár's x₁, and x₁ his y₁), the coordinate derivative c.D on R and
the derivative c.chartD 1 of the chart with respect to the transformed derivations ∂'ⱼ. The
transform π_*^{-1}(x₀² + x₁³, 2) = y₀² + x₁ is transformElem_sq_add_cube.
∂ᵢ(x₀² + x₁³) ∈ (x₀, x₁²) for both coordinates.
D(x² + y³) = (x, y²) (characteristic zero: 2 and 3 are units).
x₀ ∈ P and x₁² ∈ P.
π_*^{-1}(x₀, 1) = y₀.
π_*^{-1}(x₁², 1) = x₁.
J₁ = π_*^{-1}((x, y²), 1) = (x₁, y₁); in the model, (y₀, x₁).
I₁ = π_*^{-1}((x² + y³), 2) = (x₁² + y₁); in the model, (y₀² + x₁).
∂'₁(y₀² + x₁) = 1 in the chart ring.
D'(I₁) = (x₁² + y₁, x₁, 1) = (1).
J₁ = (y₀, x₁) is a proper ideal: it lies in the origin of the chart.