ETH Zurich, Algebraic Geometry, Spring 2024 Lecturer: Johannes Schmitt

Exercise Sheet 8

Exercise 1. Let A e 3 be the blow-up of A3 at the line L = V (x1 , x2 ) ∼

= A1 . Show that
1 1
its exceptional set is isomorphic to A × P . When do the strict transforms of two lines
in A3 through L intersect in the blow-up? What is therefore the geometric meaning of
the points in the exceptional set (analogously to the blow-up of a point, in which case the
points of the exceptional set correspond to the directions through the blown-up point)?

e 3 = Blx ,x A3 is contained in
Solution. The ideal of the line L is generated by x1 , x2 , so A 1 2

e 3 ⊆ Y = V (x1 y2 − x2 y1 ) ⊆ A3 × P1 .
A
As in [Gathmann, Example 9.15] we see that Y has one open patch
{((x1 , x2 , x3 ), (y1 : y2 )) ∈ Y : y1 ̸= 0} ∼
= {((x1 , x1 y2 , x3 ), (1 : y2 )) : (x1 , y2 , x3 ) ∈ A3 } ∼
= A3
and similarly another patch A3 for y2 ̸= 0. Thus Y is irreducible of dimension 3 and thus
for dimension reasons, the containment A e 3 ⊆ Y must be an equality.
To get the exceptional divisor, i.e. the preimage of L = V (x1 , x2 ) we set x1 , x2 to zero
and obtain
E = V (x1 , x2 , x1 y2 − x2 y1 ) = V (x1 , x2 ) = {(0, 0)} × A1 × P1 ⊆ A3 × P1 .

Let L1 , L2 ⊆ A3 be lines through L. If their strict transforms Le1 , L

e2 meet, then also L1 , L2
have to meet (since the map A e 3 → A3 sends L ei to Li . So assume that L1 , L2 go through
a point in L. By translation along the third coordinate, let’s assume they go through the
origin (0, 0, 0) ∈ L, and so they are given as
Li = {si (ai,1 , ai,2 , ai,3 ) : si ∈ A1 } .
Since none of the two lines are equal to L (because in this case their strict transform is
empty), we have (ai,1 , ai,2 ) ̸= 0. Then for si ̸= 0 the corresponding point of Li \ {0} ⊆ Y
is given by
((si ai,1 , si ai,2 , si ai,3 ), (ai,1 : ai,2 )) ∈ Y ⊆ A3 × P1 .
Indeed, the equation x1 y2 −x2 y1 exactly forces (y1 : y2 ) = (si ai,1 : si ai,2 ) = (ai,1 : ai,2 ) ∈ P1 .
Taking the limit si = 0 we obtain the point ((0, 0, 0) : (ai,1 : ai,2 )) ∈ L ei . Thus the two
1
strict transforms meet if and only if (a1,1 : a1,2 ) = (a2,1 : a2,2 ) ∈ P .
As a geometric interpretation: the lines on the exceptional set parameterize the choice
of a point on L together with a normal direction in K 3 /{(0, 0)} × K ∼ = K 2 up to scaling.
This explains why E ∼ = A1 × P1 = L × P1 .
Exercise 2. Show that any irreducible quadric hypersurface Q ⊆ Pn over a field of
characteristic not equal to 2 is birational to Pn−1 . Can you give an example of some Q
which is not isomorphic to Pn−1 ?

Page 1
ETH Zurich, Algebraic Geometry, Spring 2024 Lecturer: Johannes Schmitt

Solution. The basic idea to get the birational map Q 99K Pn−1 is to project from a point
p0 ∈ Q to some hyperplane H ∼ = Pn−1 ⊆ Pn .
To make our life easier, we can use a projective automorphism of Pn to move some
point of Q to p0 = (1 : 0 : . . . : 0), so without loss of generality we can assume that the
above p0 is contained in Q. Then we project to the hyperplane H = V (x0 ). The resulting
morphism
f : Q \ {p0 } → Pn−1 , (x0 : x1 : . . . , xn ) 7→ (x1 : . . . , xn )
is defined away from p0 . To write down the inverse, let

F = a0 x20 + a1 x0 + a2 ∈ K[x0 , x1 , . . . , xn ]2

be the quadratic equation cutting out Q = V (F ). Here

a0 ∈ K, a1 ∈ K[x1 , . . . , xn ]1 and a2 ∈ K[x1 , . . . , xn ]2

are the coefficients, seeing F as a polynomial in x0 over K[x1 , . . . , xn ]. The containment

p0 ∈ Q forces F (1, 0, . . . , 0) = 0 which implies a0 = 0. Then we distinguish two cases:
Case 1: a1 = 0
Then the equation of F does not depend on x0 at all. By induction on n we know that
V (a2 ) ⊆ Pn−1 is birational to Pn−2 , say by some rational map g : Pn−2 99K V (a2 ). Then
one can check that

P1 × Pn−2 99K Q, ((s : t), (y1 : y2 : . . . : yn )) 7→ (s0 : ty1 : . . . : tyn )

is a birational map. Since P1 × Pn−2 is birational to Pn−1 , this finishes the proof.
Case 2: a1 ̸= 0
Then on the locus U = Pn−1 \ V (a1 ) ⊆ Pn−1 the inverse of the projection f is given by
 
a2 (x)
U → Q, (x1 , . . . , xn ) 7→ − : x1 : . . . : xn = (−a2 (x) : x1 a1 (x) : . . . : xn a1 (x)) .
a1 (x)

All the components of this map are homogeneous polynomials of degree 2, which don’t
vanish simultaneously (since (x1 , . . . , xn ) ̸= 0 and a1 (x) ̸= 0 on its domain). Hence by
[Gathmann, Lemma 7.4] this indeed defines a morphism, and by a short calculation it is
the inverse of f . Hence f is birational, as claimed.
For an example of Q which is not isomorphic to Pn−1 take Q = V (x0 x3 − x1 x2 ) ⊆ P3 .
Then we have seen in [Gathmann, Example 7.11] that Q ∼ = P1 × P1 , and on Sheet 6,
Exercise 1, we proved that P1 × P1 ∼ ̸ = P2 .
Exercise 3. Let X ⊆ An be an affine variety, and let Y1 , Y2 ⊆ X be irreducible, closed
subsets, none contained in the other. Moreover, let X e be the blow-up of X at the ideal
I(Y1 ) + I(Y2 ). Show that the strict transforms of Y1 and Y2 in X
e are disjoint.
Solution. Let I(Y1 ) = ⟨f1 , . . . , fr ⟩ and I(Y2 ) = ⟨fr+1 , . . . , fr+s ⟩, then of course

I(Y1 ) + I(Y2 ) = ⟨f1 , . . . , fr+s ⟩ ⊴ K[x1 , . . . , xn ] .

Thus we can calculate X

e = Blf1 ,...,fr+s X as the blow-up at the union of these generator
sets. Then we find
e ⊆ Z = {((x1 , . . . , xn ), (y1 , . . . , yr+s )) ∈ An × Pr+s−1 : yi fj (x) = yj fi (x)∀i, j} .
X

Page 2
ETH Zurich, Algebraic Geometry, Spring 2024 Lecturer: Johannes Schmitt

Claim: Ye1 ⊆ V (y1 , . . . , yr ) ⊆ Z and Ye2 ⊆ V (yr+1 , . . . , yr+s ) ⊆ Z.

Assuming the claim, we have

Ye1 ∩ Ye2 ⊆ V (y1 , . . . , yr+s ) = ∅ ,

since not all coordinates of the point y ∈ Pr+s−1 can vanish simultaneously. Thus Ye1 and
Ye2 are indeed disjoint.
Proof of claim: We prove the statement for Ye1 , with Ye2 working similarly. By the
assumption that Y1 is not contained in Y2 , we have that U = Y1 \ Y2 ⊆ Y1 is open, and
thus dense since Y1 is irreducible. Let i = 1, . . . , r, then we claim that yi vanishes at any
point x ∈ U (and thus also on the closure Ye1 of U in Ye1 , finishing the proof of the claim).
Since x ∈ U is disjoint from Y2 = V (fr+1 , . . . fr+s ) we find an index j ∈ {r + 1, . . . , r + s}
such that fj (x) ̸= 0. But then the equation

yi fj (x) = yj fi (x)
| {z } | {z }
̸=0 =0 as x∈Y1

implies yi = 0 as claimed.
Exercise 4. Let J ⊴ K[x1 , . . . , xn ] be an ideal, and assume that the corresponding affine
variety X = V (J) ⊆ An contains the origin. Consider the blow-up X e ⊆A e n ⊆ An × Pn−1
at x1 , . . . , xn , and denote the homogeneous coordinates of Pn−1 by y1 , . . . , yn .
e n can be covered by affine spaces, with one coordinate patch
a) We know already that A
being

i : U = An → A e n ⊆ An × Pn−1 ,
(x1 , y2 , . . . , yn ) 7→ ((x1 , x1 y2 , . . . , x1 yn ), (1 : y2 : · · · : yn )).

Prove that on this coordinate patch the blow-up X e is given as the zero locus of the
polynomials
f (x1 , x1 y2 , . . . , x1 yn )
xmin
1
deg f

for all non-zero f ∈ J, where min deg f denotes the smallest degree of a monomial
in f .
Hint: You can use without proof the following variant of [Gathmann, Exercise 2.23]:

For I, J ⊴ K[x1 , . . . , xn ] one has V (I) \ V (J) = V (I : J ∞ ) where

(I : J ∞ ) = {f ∈ K[x1 , . . . , xn ] : ∃m ∈ N, g ∈ J m with f g ∈ I} .

b) Show that the exceptional set of the blow-up X

e is

Vp f in (y) : f ∈ J ⊆ Pn−1 ∼
= {0} × Pn−1 ,

where f in is the initial term of f , i.e. the sum of all monomials in f of smallest
degree. Consequently, the tangent cone of X at the origin is

C0 X = Va (f in : f ∈ J) ⊆ An .

Solution.

Page 3
ETH Zurich, Algebraic Geometry, Spring 2024 Lecturer: Johannes Schmitt

a) By definition, the blow-up X e is obtained by taking the closure of X \ {0} inside

An × Pn−1 , which is automatically contained in A e n as seen in the lecture. Taking
this closure and intersecting with the open patch U = An ⊆ A e n mentioned above,
is equivalent (by basic topology) to first intersecting with U and then taking the
closure.
By the map i : U = An → A e n , we have

i−1 V (f (x1 , . . . , xn )) = V (f (x1 , x1 y2 , . . . , x1 yn )),

so

i−1 (X \ {0}) = i−1 (V (f : 0 ̸= f ∈ J) \ V (x1 , . . . , xn ))

= V (f (x1 , x1 y2 , . . . , x1 yn ) : 0 ̸= f ∈ J) \ V (x1 , x1 y2 , . . . , x1 yn ) .
| {z }
=V (x1 )

To take the Zariski closure i−1 (X \ {0}), we just apply the hint given above, and
we see that this closure is cut out by the ideal

f (x1 , x1 y2 , . . . , x1 yn )
(⟨f (x1 , x1 y2 , . . . , x1 yn ) : 0 ̸= f ∈ J⟩ : ⟨x1 ⟩∞ ) = ⟨ : 0 ̸= f ∈ J⟩ .
xmin
1
deg f

b) We check the equality on the open subset U above. To get the exceptional set, we
impose the additional condition x1 = 0. Given 0 ̸= f ∈ J with minimal degree d and
total degree e we write its homogeneous decomposition as f = f in + fd+1 + . . . + fe .
Then we have

f (x1 , x1 y2 , . . . , x1 yn ) = xd1 f in (1, y2 , . . . , yn )+xd+1 e

1 fd+1 (1, y2 , . . . , yn )+. . .+x1 fe (1, y2 , . . . , yn ) .

Dividing by xmin
1
deg f
= xd1 and setting x1 = 0, all the terms except the first vanish,
and we have
f (x1 , x1 y2 , . . . , x1 yn )
min deg f
|x1 =0 = f in (1, y2 , . . . , yn ) .
x1
This is exactly the initial term of f in the affine coordinates U0 ⊆ Pn−1 , which
proves the first statement.
The second is then just an application of this result to the definition of the tangent
cone at the origin.

Page 4