This document provides solutions to several problems involving field extensions and Galois theory. Some key points:
- It proves formulas for the degree of composite field extensions and gives examples where the degree is not necessarily the product of the degrees.
- Quadratic field extensions of Q are classified up to isomorphism based on whether a square root of one element is in the other field.
- Irreducibility of polynomials is shown to be equivalent when viewed over different rational function fields.
- Several specific field extensions are constructed and their degrees and bases determined.
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0 ratings0% found this document useful (0 votes)
29 views2 pages
Soln8 1
This document provides solutions to several problems involving field extensions and Galois theory. Some key points:
- It proves formulas for the degree of composite field extensions and gives examples where the degree is not necessarily the product of the degrees.
- Quadratic field extensions of Q are classified up to isomorphism based on whether a square root of one element is in the other field.
- Irreducibility of polynomials is shown to be equivalent when viewed over different rational function fields.
- Several specific field extensions are constructed and their degrees and bases determined.
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2
Solutions to Homework 8
1. Let L1 and L2 be two finite extensions of k inside of an extension L/k.
(i) Prove [L1 L2 : k] ≤ [L1 : k][L2 : k]. (ii) Assume that [L1 : k] and [L2 : k] are relatively prime. Show that [L1 L2 : k] = [L1 : k][L2 : k]. If we instead assume that L1 ∩ L2 = k, then does the equality necessarily hold? (iii) As an application of (ii), if α ∈ L is algebraic over k with [k(α) : k] relatively prime to [L1 : k], prove that [L1 (α) : L1 ] = [k(α) : k], and conclude that minimal polynomial for α over k is irreducible over L1 and hence serves as the minimal polynomial for α over L1 . (i) Since [L1 L2 : k] = [L1 L2 : L1 ][L2 : k], it suffices to show [L1 L2 : L1 ] ≤ [L2 : k]. Let {ei } be a k-basis of L2 . Since the Lj ’s are finite algebraic over k, L1 L2 consists of polynomials in elements of L2 with coefficients in L1 , so L1 L2 is spanned by the {ei } over L1 (but possibly with linear dependence relations). (ii) Since [Li : k] divides [L1 L2 : k], we see that [L1 : k][L2 : k] divides [L1 L2 : k], so [L1 : k][L2 : k] ≤ [L1 L2 : k]. By Exercise 4.2 below, we have the reverse inequality. If we consider Li = Q(ai ) with a1 and a2 two distinct cube roots of 2 in C, so a1 = a2 ω with ω a non-trivial cube root of 1, we see that L1 L2 = Q(a1 , ω) is a splitting field of X 3 − 2, and this has degree 6 over Q, by the first part. Since Q(a1 ) 6= Q(a2 ) (why?), we conclude that Q(a1 ) ∩ Q(a2 ) = Q by degree considerations, so this furnishes a counterexample to the last part. (iii) By previous parts, [L1 (α) : k] = [k(α) : k][L1 : k], so [L1 (α) : L] = [k(α) : k]. Thus, the minimal polynomial f for α over K is a monic in K[X] ⊆ L[X] which has the same degree as the minimal polynomial for α over L, and vanishes at α. Thus, f is also the minimal polynomial for α over L. 2. (i) Let L1 and L2 be two quadratic extensions of k. Assume k does not have characteristic 2, so Li = k(ai ) with a2i = bi ∈ k × . Show that L1 ' L2 as extensions of k if and only if b1 /b2 is a square in k × . Use this to give a complete list (without repetitions) of all quadratic extensions of Q, up to isomorphism. (ii) Consider the identification ι between isomorphism classes of quadratic extensions of k and the group 2 k × /k × , as explained in (i). If L1 and L2 are two quadratic distinct extensions of k inside of an extension L/k, show that the composite L1 L2 is a degree 4 extension of k and the non-trivial subextensions over k are L1 , L2 , and the field corresponding to the ‘product’ of L1 and L2 under ι. √ (i) The ‘if’ is clear, since b1 = (b1 /b2 )b2 . As for ‘only if’, suppose b2 is a square in k( b1 ), so for some x, y ∈ k, (x + ya1 )2 = b2 . This says that b2 = x2 + b1 y 2 and 2xya1 = 0. Since 2 ∈ k × , xy = 0. Because 2 2 b2 ∈/ k × , we can’t have y = 0, so x = 0 and b1 /b2 = y 2 ∈ k × . √ √ √ (ii) Say Li = k( bi ). Then L1 L2 = k( b1 , b2 ) is of degree 4 over k since the degree divides 4 and is at least 2, so if the composite degree weren’t √ 4, we would have L1 = L1 L2 = L2 , a contradiction. The composite contains the quadratic subfield k( b1 b2 ), which we see is distinct from L1 and L2 since L1 6= L2 and each Li 6= k. This corresponds to the ‘product’ under ι. √ It remains to check that L1 L2 contains √ no other quadratic subfields over k. Say k( b) lies in L1 L2 , × √ with√ b ∈ k a non-square. √ Since L1 ( b) lies between L1 = √ k( b1 ) and the quadratic extension L1 L2 , √ √ L1 ( b) = L1 or L1 ( b) = L1 L2 = L1 ( √b2 ). In the first case, k(√ b) = k( b1 ). Consider the second case, so b/b2 is a √ square in L1 , say b/b2√= (x + y b1 )2 = x2 + y 2 b1 + 2xy b1 . This forces 2xy = 0, so x or y vanishes. √ √ Thus, k( b) = k( b1 b2 ) or k( b) = k( b2 ), respectively. 3. Let f ∈ k[T, X], with f not divisible by any non-constants in k[T ] or k[X]. Show that f is irreducible when viewed in k(X)[T ] if and only if it is irreducible when viewed in k(T )[X]. By Gauss’ Lemma, both cases are equivalent to f being irreducible in k[T, X] (since f has positive degree both in T and X, and no non-constant elements of k[T, X] which are units in k[T ] or k[X] can divide f , by hypothesis). 4. For each of the following extensions L/k, determine [L : k] and find a basis for L as a k-vector space: k = Q, L = Q(a, b) with a2 = 6, b3 = 2 k = C(T ), L is the splitting field of X n − T over k 1 2
k = Fp (T ), L is the splitting field of X p − T over k, with p a prime (same p in both places!).
For the first one, the degree is 6 (by Exercise 1) and a basis is the set of ai bj with i = 0, 1 and j = 0, 1, 2 (by the proof of the multiplicativity formula). In the second case, once we adjoin a single solution x to X n = T , then we already have all of the others, since X n − 1 splits completely over C (and hence over C(T )). Since X n − T is irreducible over C(T ) by Exercise 2, [L : k] = n and a basis consists of xi with 0 ≤ i < n. A similar argument applies to any field in which X n − 1 splits completely. This applies in particular to n = p and the base field Fp (T ), settling the final case. 5. Let L/K be a field extension, and α ∈ K be algebraic over L. Consider the multiplication map mα : K(α) → K(α) on the finite-dimensional K-vector space K(α) = K[α]. Using a matrix for this relative to a suitable basis, prove that the characteristic polynomial of this linear map is the minimal polynomial of α over K. In terms of this minimal polynomial, what are the trace and determinant of this map? Let f = X d + ad−1 X d−1 + · · · + a0 be the minimal polynomial of α over K. Since K(α) ' K[X]/f over K, {αi } with 0 ≤ i < d is an ordered K-basis of K(α). With respect to this ordered basis, the matrix for Tα is 0 0 ... 0 −a0 1 0 . . . 0 −a1 M = 0 1 . . . 0 −a2 .. .. .. .. . . · · · . . 0 0 . . . 1 −ad−1 In order to show that det(λI − M ) = f (λ), we expand along the right column. Be careful about powers of −1. The trace is −ad−1 and the determinant is (−1)d times the constant term, which is to say (−1)d a0 . 6. Prove that f (X) = X 3 + 3X + 1 is irreducible over Q. If we let α denote a root of f in some extension, use the fact that f vanishes at α and f (X − 1) vanishes at α + 1 to express 1/α and 1/(α + 1) as quadratic polynomials in α with Q coefficients. We can apply the rational root theorem (or apply Eisenstein to f (X − 1) = X 3 − 3X 2 + 6X − 3 with p = 3) to prove irreducibility over Q. Using the relation f (α) = 0, we see that α 6= 0 and −(α2 + 3) = α−1 . Since f (X − 1) vanishes on β = α + 1 6= 0, we can express β −1 as a rational polynomial in β. Using β = α + 1 then yields β −1 = (1/3)α2 − (1/3)α + 4/3. 7. Prove that X 4 − 5X 2 + 6 and X 4 + 5X 2 + 6 are reducible over Q with splitting fields of degree 4 which you should describe concretely (give a basis and express in the form Q(α) for suitable α). Prove also that X 4 − 5 is irreducible over Q but with splitting field of degree 8 over Q which you should describe in terms of some field generators and a basis. Since X 4 −5X 2 +6 = (X 2 −3)(X 2 −2) and 2/3 is not a square in Q (by a variety of arguments), by Exercise 1 we see that the splitting field is Q(a, b) with a2 = 3 and b2 = 2, and Q(a) 6= Q(b), so the extension√has degree 4. Since (a+b)2 = 5+2ab, if Q(a+b) is not the entire splitting field, then Q(a+b) = Q(ab) = Q( 6). But then a + b is a Q-linear combination of 1 and ab, contradicting the fact that 1, a, b, and ab are a basis of the splitting field over Q. The case of X 4 + 5X 2 + 6 is done similarly. Now consider X 4 − 5. This is irreducible over Q by Eisenstein, so if a is a root in a splitting field, then Q(a)/Q has degree 4. From our knowledge of C, this polynomial has four distinct roots in a splitting field, so the roots are of the form aζ, with ζ a set of 4 distinct roots of X 4 = 1. Taking ratios of roots of X 4 − 5, we see that a splitting field is a composite of subfields Q(a) and Q(ζ) with ζ 4 = 1 and ζ 6= 1, −1 — that is, ζ 2 + 1 = 0. Now ζ ∈ / Q(a), since Q(a) admits a real embedding, yet X 2 + 1 has no roots in R. So the splitting field Q(a, ζ) is quadratic over Q(a) and therefore has degree 8 over Q. If you are interested, you might suspect a + ζ could be a primitive generator for the splitting field over Q. To prove this without Galois theory, one way to is to somehow verify that (a + ζ)i for 0 ≤ i ≤ 7 are linearly independent over Q. This can be done by expanding these all in terms of the basis an ζ m (0 ≤ n ≤ 3, 0 ≤ m ≤ 1) of the splitting field over Q and then computing the relevant 8 by 8 determinant is non-zero. Quite painful.