You are on page 1of 3

Algebra Homework Set 7

Hung Tran.

8.3.11
(⇒) Suppose R is PID then R is a UFD by theorem 8.3.14. Furthermore, since
every ideal of R is principal; in particular, any ideal generated by two elements is
principal. So R is a Bezout domain.
(⇐) Suppose R is a UFD and also a Bezout domain. Let I be an arbitrary ideal of
R. If I contains only zero, it is obviously principal. Otherwise, it contains nonzero
elements. Since R is UFD, each nonzero element of I can be uniquely factorized
as a product of irreducibles. Let a be one element with the minimal number of ir-
reducibles in its factorization (that exists since for each element UFD-ness implies
that the number of irreducibles is a finite integer). We claim that I = (a). Let b
be an arbitrary element in I. Since R is a Bezout domain, (a, b) = (d) for some d
in R.
Since d ∈ (a, b) there exists x, yinR such that d = xa + yb. Since a, b ∈ I, so is xa
and yb. Thus d is in I. But a ∈ (d) implies that a = rd for some r in R. Since R is
a UFD, each r and d has its unique factorizations. Then, because the factorization
of a is also unique, it must be the combination of the factorizations of r and d. If
r is not a unit, then d is the element in I that has a smaller number of irreducibles
then a, a contradition. So r is a unit, and so (d) = (a); therefore, b ∈ (a). As b is
arbitrary, I = (a). As I is arbitrary, R is a PID. QED

9.3.4.
a.
R is certainly a commutative ring with identity. It remains to show it does not
have any zero-divisor. Suppose (a + xH(x))(b + xG(x)) = 0 in R then the equality
certainly holds in Q[x]. But Q[x] is a Euclidean domain by theorem 9.2.3, so either
a + xH(x) = 0 or b + xG(x) = 0 in Q[x]. Then either is 0 in R also, making R an
integral domain
(a + xH(x))(b + xG(x)) = 1 then ab = 1 and bH(x) + aG(x) + xH(x)G(x) = 0.
Since each G(x),H(x) is a polynomial, the latter equality only holds iff both are of
degree 0 and then their constant terms must also be zero to make xG(x)H(x) be
of degree 0. So it comes down to ab = 1 for a, b ∈ Z. So a, b must be 1 or −1. Thus
the units are 1 and -1.
b.
a + xH(x) is a irreducible iff it can not be represented as a product of two non-unit
elements. If |a| 6= 1 and 0 and H(x) 6= 0 then a + xH(x) = a(1 + xH(x)/a).
H(x)/a ∈ Q(x) and by part a, a and 1 + xH(x)/a are not units. So a + xH(x) is
reducible if |a| =
6 1and0 and H(x) 6= 0.
If a = 0 then xH(x) = x.H(x) is reducible if H(x) 6= 0.
If |a| = 1 then 1+xH(x) is reducible in R if it is reducible in Q[x] since the constant
term for each factor of it in Q[x] must have absolute value equal 1 and, hence, that
comes as a legitimate factorization in R. Moreover, if it is irreducible in Q[x] it cer-
tainly must be irreducible in R as R ⊂ Q[x]. Now if F = a + xH(x) is irreducbile
and F divides MN in R, then certainly F divides M,N in Q[x]. Since Q[x] is a ED
and in particular, a UFD, and F is also irreducible in Q[x] F is a prime in Q[x]. So
either F divides M or N in Q[x]; as F has a constant term 1, F also divides M or N
1
2

in R(as the result of dividing M or N by F in Q[x] has the constant term in Z that
comes from the constant term in M or N). Thus F is also a prime in R (1)
If H(x) 6= 0 then it is obvious that a is irreducible in R as a is irreducible in Z. So
a must be a prime or its associate. If a divides (b + xH(x)(c + xG(x)) in R then a
divides bc since a divides H(x) and G(x) in Q[x]. So either a divides b or c; thus,
a divides either b + xH(x) or c + xH(x); then a is a prime(2)
(1) and (2) prove part (b).
c.
Since the only irreducibles in R are primes and their associates in Z and irreducibles
in Q[x] with constant term equal 1 or -1, by checking the degree, the only way x
can possibly be factorized into irreducibles is to have one polynomial of degree 1
and other constant polynomials. So x = (ax + 1)b with b a product of primes or
associates in Z. It follows immediately that b = 0 and x = 0, a contradiction. So x
can not be factorized into a product of irreducibles in R.
d.
We have x = (x/2).2 and both x/2 and 2 are not in the ideal generated by x (as
(a + xH(x))x = ax + x2 H(x) a polynomial with the coefficient of the term of degree
1 is an integer and no constant term(*)). So x is not a prime.
By (*), in R/(x), a + x(b + H(x)) ∼ c + x(d + G(x)) iff b ≡ d(mod1) and a = c. So
R/[x] is isomorphic to the ring of polynomials with degree less than 2, the coeffi-
cient of the term of degree 1 is in Q/Z while the constant term is in Z. Informally,
the ring can be written as (Q/Z)x + Z. QED

Additional problems
1
As F is a field, F [x] is a ED by section 9.2. So there is a division algorithm and if a
is a root of f in F, then f = (x − a)g(x) + r for r is a constant. As f (a) = 0, r = 0.
So f = (x − a)g(x) in which both (x − a) and g are in F [x]. By Gauss’s lemma,
f = a(x)h(x) with a(x) and h(x) in R[x] and a(x) is of degree 1. Now since f is
monic, a(x) must also be monic (by checking the leading coefficient of both sides
of the equation). Thus a(x) = ux + b as u is a unit in R. Hence, f has a root in R,
namely bu−1 . QED

2
As F is a field, F × contains all elements except for 0.
Let K be a set of primes in R uniquely up to associates, that is, each is chosen from
a class of associates.
For each ps in K, denote Zs = {pzs ; z ∈ Z} with the multiplication law inherited
from F. Obviously, Zs is a group isomorphic to Z (the group Zs is written multi-
plicatively while Z is written additively).
By the construction of the fieldQof fractions for an UFD, any element of F can be
written uniquely in the form u s∈S pβs s with S a finite set, ps in K, βs ∈ Z, u is a
unit in R.
Define φ : F × → pt ∈K Zt by φ(u s∈S pβs s ) = h such that h(s) = pβs s in Zs for s
L Q

in S while h(t) = 1 for t ∈/ S while pt is a prime in R.


Since S is finite, the function is well-defined. It is also obvious that the function is
a homomorphism of groups.
Since the direct sum only admits a finite number of non-trivial components, the
3

map is certainly surjective.


Finally, ker(φ) = R× so F × /R× ≡ × ×
L
pt ∈K Zt . Hence, F /R is a free abelian
group. QED

3.
We prove by induction on n. For n = 1, the statement follows immediately since x
is irreducible in Z[x].
Suppose the statement is true for n < k and consider X a k × k matrix. det(X) is a
polynomial of k 2 variables over Z and also a polynomial of x11 over T = Z[x12 ...xkk ].
So det(X) is irreducible in R iff it is irreducible in T [x11 ].
As Z is a ED, T is a UFD. Let F be its field of fraction. By the minor formula
of determinant, det(X) is a polynomial of degree 1 in F [x11 ]. As any constant
polynomial is a unit in F [x11 ], det(X) is irreducible in F [x11 ]. By corollary 9.3.6
following Gauss’s lemma, det(X) is irreducible in T [x11 ] if the greatest common
divisor of coefficient of det(X) is 1.
det(X) = Ax11 + B with A the determinant of the matrix Y = (xij 2 ≤ i, j ≤ k. By
the induction hypothesis, A is irreducible in H = Z[(xij )] with 2 ≤ i, j ≤ k. So A
is also irreducible in T (if not, then it has a nontrivial factorization that involves at
least one variable not in H; writing each factor as a polynomial of that variable and
checking the degree and the property of no zero divisor lead to a contradiction).
So if the greatest common divisor of A and B are not 1, A must divide B. But if
xij = 1 if i = j + 1 for 1 ≤ j ≤ k − 1 and i = 1, j = k and other components are
zero (i refers to row while j refers to column) then A = 0, |B| = 1, a contradiction
(in other words, we consider a ring homomorphism that sends A to 0, B to 1; as A
divides B, B must be in the ideal generated by A, but 1 is not in the ideal generated
by 0 in R). Thus, the greatest common divisor of A and B is 1 and by corollary
9.3.6, det(X) is irreducible in T [x11 ], and so is in R. QED

4.
a.
Suppose I1 ⊂ I2 ⊂ ... is an ascending chain of principal ideals with Ii = (ai ). Since
the ring R of consideration is an UFD, any element has a unique factorization and
a unique number of irreducibles in its factorization. For r in R, denote this number
c(r).
Ik ⊂ Ik+1 implies ak+1 divides ak . If they are associates then Ik = Ik+1 . Otherwise,
c(ak+1 ) < c(ak ). Since c( a1 ) is finite, the chain can not be properly ascending. So
it must stablize after a finite number of ideals. So R satisfies ACC on principal
ideals.
b.
We have (x/2) ⊂ (x/4)... ⊂ (x/2k ).... is certainly an ascending chain. It remains
to show that the inclusion at each point on the chain is proper. Suppose not, then
there exist k such that (x/2k ) = (x/2k+1 ), so x/2k+1 = F x/2k for some F in R.
So 1 = 2F ; that is F = 1/2 not in R (recall that R contains elements of the form
a + xH(x) for H(x) in Q[x] and a ∈ Z), that leads to a contradiction. So (x/2k ) is
properly contained in (x/2k+1 ). QED

You might also like