1. Affine Space and affine algebraic sets
Let A = k[X1, · · · , Xn] be the ring of polynomials in variables X1, · · · , Xn over an alge- braically closed field k. For any index set I = (i1, · · · , in), we denote XI = X1i1· · · Xnin. Elements of A are of the form f (X) = P
IcIXI, where cI ∈ k. The affine n-space over k denoted by An(k) is the set of n-tuples: {(a1, · · · , an) : a1, · · · , an ∈ k}. For any P (a1, · · · , an) ∈ An(k), we set
f (P ) =X
I
cIai11· · · ainn.
Then f : An(k) → k defines a k-valued function on An(k). The zero set V (f ) of f is the set V (f ) = {P ∈ An(k) : f (P ) = 0}.
We also denote An(k) \ V (f ) by D(f ). When f is nonconstant, V (f ) is called a hypersurface in An(k). A hypersurface in A2(k) is called an affine plane curve. When deg f = 1, V (f ) is called a hyperplane in An(k) and called a line if n = 2.
If S is a family of polynomials in A, the common zero set of S is defined to be V (S) = \
f ∈S
V (f ) = {P ∈ An(k) : f (P ) = 0, ∀f ∈ S}.
A subset Y of An(k) is called an affine algebraic set if Y = V (S) for some family of polynomials S in A. A subset U of An(k) is said to be Zariski open if An(k) \ U is an affine algebraic set.
Proposition 1.1. The Zariski open sets of An(k) forms a topology on An(k). The topology defined by the Zariski open sets is called the Zariski topology of An(k).
Proof. It is equivalent to prove the following.
(1) ∅, An(k) are affine algebraic.
(2) Any intersection of affine algebraic sets is affine algebraic.
(3) Any finite union of affine algebraic sets is affine algebraic.
(1) follows from the fact that V (0) = An(k) and V (1) = ∅, i.e. An(k) and ∅ are affine algebraic sets and their complement ∅ and An(k) in An(k) are Zariski open by definition.
(2) follows from the fact that T
iV (Ti) = V (S
iTi). Here Ti are sets of polynomials for all i.
Let T1, T2 be two sets of polynomials and T be the set of polynomials {f g : f ∈ T1, g ∈ T2}. Then V (T1) ∪ V (T2) = V (T ). (3) can be proved by induction.
From now on, we only consider the Zariski topology on An(k).
A topological space X is reducible if it is a union of two proper closed subsets, i.e.
X = X1∪ X2 where X1, X2 are closed and X1, X2 6= X. A topological space is irreducible if it is not reducible. A nonempty subset Y of a topological space is irreducible if Y with the subspace topology induced from X is an irreducible space.
Definition 1.1. An affine variety is an irreducible closed subset of An(k). A quasi-affine variety is an open subset of an affine variety.
Lemma 1.1. If T1 ⊂ T2 are subsets of A, then V (T1) ⊃ V (T2).
Proof. If P ∈ V (T2), then f (P ) = 0 for all f ∈ T2. Since T1 ⊂ T2, we see that f (P ) = 0 for all f ∈ T1. This implies that P ∈ V (T1). Hence V (T2) ⊂ V (T1).
1
2
Corollary 1.1. Let a be the ideal of A generated by a set of polynomials T. Then V (a) = V (T ).
Proof. We know that a is the smallest ideal containing T. By the previous lemma, V (a) ⊂ V (T ). Let P ∈ V (T ). We find f (P ) = 0 for all f ∈ T. If h ∈ a, there exist a1, · · · , ar ∈ A and f1, · · · , fr ∈ T so that h = a1f1+ · · · + arfr. Since fi ∈ T for 1 ≤ i ≤ r, fi(P ) = 0 for 1 ≤ i ≤ r. Hence h(P ) = 0. We see that P ∈ V (a). We conclude that V (a) = V (T ). Corollary 1.2. Every affine algebraic subset of A1(k) is either finite or A1(k) itself.
Proof. Let Y be an affine algebraic subset of A1(k) so that Y 6= A1(k). Then there is a proper ideal I of k[X] so that Y = V (I). Since every polynomial in k[X] only has finite number of zeros, V (f ) is a finite subset of A1 for all f (X) ∈ k[X]. Since V (I) = ∩f ∈IV (f ), V (I) is the intersection of finite subsets of A1(k). This implies that V (I) is either the empty
set or a finite subset of A1(k).
Since k is an algebraically closed field, A1(k) has infinite number of elements. Since every proper closed subsets of A1(k) is finite, A1(k) can not be a union of two proper closed subsets. Hence A1(k) is irreducible.
2. The Ideal of a Set of Points Let P be any point of An(k), we denote
IP = {f ∈ A : f (P ) = 0}.
Then IP is a maximal ideal of A. This is because IP is the kernel of the ring epimorphism
P : A → k defined by f 7→ f (P ) and k[X]/IP ∼= k is a field. If P = (a1, · · · , an), then IP is the ideal hX1− a1, · · · , Xn− ani.
For any subset Y of An(k), we define
I(Y ) = \
P ∈Y
IP.
Since the intersection of ideals of a ring is again an ideal of that ring, I(Y ) is an ideal of A.
Moreover I(Y ) is the set of all f in A so that f (P ) = 0 for all P ∈ Y.
Lemma 2.1. If Y1 ⊂ Y2 are subsets of An(k), then I(Y1) ⊃ I(Y2).
Proof. Let f ∈ I(Y2). Then f (P ) = 0 for all P ∈ Y2. Since Y1 is contained in Y2, f (P ) = 0 for all P ∈ Y1. Therefore f ∈ I(Y1). We find I(Y2) ⊂ I(Y1). Lemma 2.2. For any two subsets Y1, Y2 of An(k), I(Y1∪ Y2) = I(Y1) ∩ I(Y2).
Proof. Let f ∈ I(Y1 ∪ Y2). Then f (P ) = 0 for all P ∈ Y1∪ Y2. Hence f (P ) = 0 for all P ∈ Y1 and all P ∈ Y2. Then f ∈ I(Y1) and f ∈ I(Y2). We see f ∈ I(Y1) ∩ I(Y2). Then I(Y1∪ Y2) = I(Y1) ∩ I(Y2).
Conversely, let f ∈ I(Y1) ∩ I(Y2). Then f (P ) = 0 for all P ∈ Y1 and all P ∈ Y2. Hence f (P ) = 0 for all P ∈ Y1 ∪ Y2. This implies that f ∈ I(Y1 ∪ Y2). Hence I(Y1) ∩ I(Y2) ⊂ I(Y1∪ Y2).
Let R be any commutative ring with identity and I be an ideal of R. The radical of I denoted by √
I is the set of all r so that there exists n > 0 so that rn∈ I. We leave to the reader to verify that√
I is an ideal of R containing I. When I = (0) is the trivial ideal of R, p(0) is also denoted by Nil(R) called the nilradical of R.
3
Theorem 2.1. (Hilbert Nullstellensatz) Let k be an algebraically closed field and I be an ideal of A. Let f ∈ A be a polynomial so that f (P ) = 0 for all P ∈ V (I). Then fr∈ I for some r > 0.
Proposition 2.1. For any ideal a of A, I(V (a)) =√ a.
Proof. Suppose f ∈ I(V (a)). Then f (P ) = 0 for all P ∈ V (a). By the Hilbert-Nullstellensatz, fr∈ a for some r ≥ 0. In other words, f ∈√
a. Conversely, if f ∈√
a. Then fr ∈ a for some r ≥ 0. Hence for all P ∈ V (a), (f (P ))r = 0. Since k is a field, f (P ) = 0. This shows that
f ∈ I(V (a)). We prove the assertion.
Proposition 2.2. Let Y be any subset of An(k). Then V (I(Y )) = Y .
Proof. For P ∈ Y, for all f ∈ I(Y ), we have f (P ) = 0 by definition. Hence P ∈ V (I(Y )).
In other words, Y ⊂ V (I(Y )). Since V (I(Y )) is closed containing Y and Y is the smallest closed subset of An(k) containing Y, V (I(Y )) contains Y .
Since Y is closed in An(k), Y = V (T ) for some set T of polynomials. Then we know that for all P ∈ Y, and all f ∈ T, f (P ) = 0. Hence T ⊂ I(Y ). By Lemma, V (T ) ⊃ V (I(Y )), i.e.
Y ⊃ V (I(Y )). This proves our assertion.
Proposition 2.3. An algebraic set is irreducible if and only if its ideal is a prime ideal.
Proof. Let us assume Y is an affine variety. Then we want to show that I(Y ) is a prime ideal in A. Let f g ∈ I(Y ). Then V (f g) ⊃ Y. Since Y is closed, Y ∩ V (f g) is closed. Since V (f g) = V (f ) ∪ V (g),
Y ∩ V (f g) = (Y ∩ V (f )) ∪ (Y ∩ V (g)).
We know that Y ∩ V (f ) and Y ∩ V (g) are closed subsets of Y. By the irreducibility, either V (f ) = Y or V (g) = Y. Assume that V (f ) = Y. Then I(Y ) = I(V (f )) = p(f). Since f ∈p(f), f ∈ I(Y ). We prove that I(Y ) is a prime ideal of A. Corollary 2.1. The affine n-space An(k) is irreducible.
Proof. The ideal defining An(k) is (0). Since A = k[X1, · · · , Xn] is an integral domain, the zero ideal (0) is a prime ideal of A. Proposition 2.3 implies that An(k) is irreducible. Corollary 2.2. There is a one-to-one inclusion reversing correspondence between algebraic sets in An(k) and radical ideals in A.
Proof. The correspondence is given by Y 7→ I(Y ).
Definition 2.1. The coordinate ring of an affine algebraic set Y is the quotient ring A(Y ) = A/I(Y ).
Proposition 2.4. Let B be an algebra over k which is also a domain. B is finitely generated if and only if it is isomorphic to the coordinate ring of an affine variety.
Proof. Since A is generated by {x1, · · · , xn} as an algebra over k, A is a finitely generated complex algebra. If B = A(Y ) for some affine variety in An(k), then B is generated by {π(x1), · · · , π(xn)}, where π : A → B is the quotient map. Since Y is an affine variety, I(Y ) is a prime ideal. Then A(Y ) = A/I(Y ) is an integral domain.
Conversely, assume that B is a finitely generated l-algebra which is also a domain. Let {b1, · · · , bn} be a set of generators of B as a C-algebra. Define a ring homomorphism
ϕ : k[X1, · · · , Xn] → B
4
sending Xi to bi. Then ϕ is surjective. By the first isomorphism theorem, B ∼= k[X1, · · · , Xn]/I,
where I = ker ϕ. Since B is a domain, I is a prime ideal. Let Y = V (I). Then Y is an algebraic set in An. Moreover I(Y ) = I(V (I)) = √
I. Let f ∈ I(Y ). Then f ∈ √
I. There is r > 0 so that fr ∈ I. Since I is prime, f ∈ I. We find I(Y ) ⊂ I. Since I ⊂ √
I and
√
I ⊂ I(Y ), I ⊂ I(Y ). We conclude that I = I(Y ). Since I = I(Y ) is prime, Y is an affine
variety. This prove our assertion.
Proposition 2.5. An(k) with the Zariski topology is a Noetherian space.
Proof. Let {Yi} be a descending chain of closed subsets in An(k) : Yi ⊃ Yi+1· · · . Then we obtain an ascending chain of ideals (ai) : ai ⊂ ai+1, where ai = I(Yi). Since A is a Noetherian ring, there exists r > 0 so that ai= ar for all i ≥ r. In this case, Yi = Yr for all
i ≥ r. We prove that An(k) is a Noetherian space.
Since the subspace of any Noetherian space is also Noetherian, an affine algebraic set is a Noetherian space with the subspace topology of An(k). Using the property of Noetherian space, we can decompose any algebraic set in An(k) uniquely into a finite union of its irreducible components. Hence we conclude that:
Corollary 2.3. Every algebraic set in An(k) can be expressed uniquely as a finite union of varieties, no one containing another.