Some Hints and Solutions to Homework Problems

Homework 1.
1. Prove Euclid I.6 : if two angles in a tiangle are equal to each other then the
triangle is isosceles.
Euclid argues this way. Let a triangle ABC has the angle BAC equal the angle
BCA but |AB| > |BC|. Take D on AC such that |AD| = |BC|. Then the triangles
ADC and CBA have equal areas by the side-angle-side proposition, which is absurd
since ADC is strictly inside CBA.
2. Suppose that the triangles ABC and DEF have |AB| = |DE|, |AC| = |DF |
and the angle ABC equal the angle DEF . Does this imply that the triangles are
congruent?
No. Fix the angle ABC and the length |AB|. Circles centered at A may intersect
the ray BC at two different points C and C 0 . Then the triangles ABC and ABC 0
are not congruent since one of them is strictly inside the other.
1.10. Ta ◦ R = R ◦ T−a . Respectively,
(Ta ◦ R) ◦ (Tb ◦ R) = (Ta ◦ R) ◦ (R ◦ T−b ) = Ta ◦ (R ◦ R) ◦ T−b = Ta ◦ T−b = Ta−b .

1.18. g · g = e is equivalent to g −1 = g. If this is true for any g ∈ G, then for

any a, b ∈ G we have ab = (ab)−1 = b−1 a−1 = ba.
2.14. If a linear transfromation A is an isometry then At A = I.
The formula (u + v)2 = u2 + 2uv + v 2 applied to the dot-product of vectors shows
how to express the dot-product in terms of sums and lengths of vectors:
1
hu, vi = (hu + v, u + vi − hu, ui − hv, vi) for any u, v.
2
Since A preserves lengths of all vectors and A(u + v) = Au + Av (due to lineariry
of A), we conclude that A preserves dot-products: hAu, Avi = hu, vi for any u and
v. Therefore for any u, v
0 = hAu, Avi − hu, vi = hAt Au, vi − hIu, vi = h(At A − I)u, vi.
Thus for any u the vector (At A − I)u = 0 since it is orthogonal to any v. This
means that the matrix At A − I = 0.

1
 Homework 2.
1.15. g = ag implies e = gg = (ag)g −1 = a(gg −1 ) = ae = a.
−1

1.16. gh = gk implies g −1 (gh) = g −1 (gk) implies (g −1 g)h = (g −1 g)k implies

eh = ek implies h = k.
1.17. e = f g implies g −1 = eg −1 = (f g)g −1 = f (gg −1 ) = f e = f .
2.20a. For any vector x we have Tu (x) = x + u. Thus
ρ(Tu (x)) = ρ(x + u) = ρ(x) + ρ(u) = Tρ(u) (ρ(x)).
2.26b. Hint: Any isometry on the plane is the composition of a translation with
a rotation or reflection. A translation can be written as the composition of two
reflections in two parallel lines perpendicular to the direction of the translation.
Thus compositions of a translation with a reflection can be represented as the
composition of three reflections. The composition of a translation Tu with a rotation
ρ is in fact the rotation through the same angle about the fixed point defined by
the equation ρ(x) + u = x.
Homework 3.
1. Prove that a group homomorphism f from a group G to a group H maps
the identity element of G to the identity element of H and maps inverse elements
to inverse elements.
Indeed, f (a)eH = f (a) = f (aeG ) = f (a)f (eG) implies eH = f (eG ). Similarly,
eH = f (eG ) = f (aa−1 ) = f (a)f (a−1 ) implies f (a)−1 = f (a−1 ).
3.7. The subgroup C4 of rotations in the group D4 of symmetries of the square
is a cyclic group of order 4. It is not isomorphic to V4 since any cyclic subgroup in
V4 has at most two elements.
3.15a. The cyclic group generated by the rotation through an angle α is infinite
cyclic when α is incommensurable with 2π and is finite of order n = 1, 2, 3, ... when
α = 2π/n.
3.15b. SO(2) is not isomorphic to Zn because SO(2) is infinite, and it is not
isomorphic to Z because Z contains no finite cyclic subgroups of order > 1.
3.18. A regular n-gon has n equal angles with the total sum π(n−2). Thus each
angle is π(n − 2)/n which is 60◦ , 90◦ , 108◦ , 120◦ , ... for n = 3, 4, 5, 6, ... respectively.

2
The sum of face’s angles at a vertex of a convex regular polyhedron is < 360◦ .
Therefore the only possible configurations of such faces near a vertex are: three
regular triangles, squares or pentagons, or four or five regular triangles.

Homework 4.
4.5bc. The “stereographic” projection of the circle x2 + y 2 = 1 yields the
description of integer Pythagorean triples a2 + b2 = c2 in the form a = r 2 − s2 ,
b = 2rs, c = r 2 + s2 where r > s are integers. Thus an integer c is the hypotenuse
of a right triangle with integer legs if and only if c can be written as the sum of
squares of two distinct positive integers. For c ≤ 13, the available squares are
1, 4, 9, and the required values of c are 5 = 1 + 4, 10 = 1 + 9 and 13 = 4 + 9.
5.5+5.6. p(x) := x3 + 3x − 4 is increasing since p0 (x) = 3x2 + 3 > 0. Therefore
p(x) = 0 has at most one real solution,√which obviously is x = 1. Thus the real
solution x = y − 1/y where y := (2 + 5)1/3 is in fact equal to 1. Therefore √ for
2
y > 0 we have the quadratic
√ 1/3 √ equation y − y − 1 = 0 and find y = (1 + 5)/2.
Thus (2 + 5) = (1 + 5)/2.
5.9. x3 + 3x − 4 has a root x = 1 and is therefore divisible by x − 1. Explicitly
we have x3 + 3x − 4 = (x − 1)(x2 + x√+ 4). Thus the other two roots are solutions
to x2 + x + 4 = 0. These are (−1 ± i 15)/2.

3
 Homework 5.
5.22. By definition, the point at ∞ plays the role of 0 in the group structure of
the elliptic curve E : y 2 + y = x3 − x. The symmetry (x, y) 7→ (x, −1 − y) about
the line y = −1/2 (found by completing the square y 2 + y = (y + 1/2)2 − 1/4) plays
the role of taking the opposite R 7→ −R in the group E. The sum is defined by
requiring that P + Q + R = 0 whenever P, Q, R ∈ E are on the same line.
The line x = t, y = 0 passes through P = (0, 0) and Q = (1, 0) at t = 0, 1 and
intersects E when 0 = t3 − t which has solutions t = 0, 1, −1. Thus P + Q + R = 0
for R = (−1, 0) and S := P + Q = −R = (x, −1 − y) = (−1, −1).
Furthermore, the tangent line to E at S = (x0 , y0) = (−1, −1) is given by the
equation (2y0 + 1)(y − y0) = (3x20 − 1)(x − x0 ) i.e. y = −2x − 3 or x = t, y = −2t − 3.
It intersects E when (2t+3)2 −(2t+3) = t3 −t or, equivalently, t3 −4t2 −11t−6 = 0.
It has the double root at t = −1 (due to the tangency at S) and therefore has one
more root t = 6. Thus the other root is x = 6, and 2S = −T = (t, −1+2t+3)|t=6 =
(6, 14).
Similarly, the line x = −t, y = −t passes through P, S at t = 0, 1 and meets E
when t2 − t = −t3 + t, at t = 0, 1 and therefore −2. Thus S + P = (x, −1 − y) =
(2, −3). The line x = t + 1, y = −3t passes through Q, S + P at t = 0, 1 and meets
E when 9t2 − 3t = t3 + 3t2 + 3t + 1 − t − 1. The latter equation has solutions t = 0, 1
and therefore 5. Thus (S + P ) + Q = (x, −1 − y) = (5 + 1, −1 + 15) = (6, 14) = 2S
(as expected!)
Finally, the line x = 1 − 2t, y = −t passes through Q, S at t = 0, 1 and intersects
E when t2 − t = −8t3 + 12t2 − 6t + 1 − 1 + 2t or, equivalently, 8t3 − 11t2 + 3t =
t(t−1)(8t−3) = 0 which has solutions t = 0, 1 and 3/8. Thus S +Q = (x, −1−y) =
(1/4, −5/8). The line x = 2t, y = −5t passes through P, S + Q at t = 0, 1/8 and
meets E when 25t2 − 5t = 8t3 − 2t, i.e. at t = 0, 1/8 and therefore 3. Thus
P + (S + Q) = (x, −1 − y) = (6, 14) = 2S — as expected!

4
 5.25. x4 + 4x − 1 = 0 is equivalent to (x2 + y)2 = 2yx2 − 4x + y 2 + 1. The R.H.S.
is a complete square if 16 = 8y(y 2 + 1) or, equivalently, y 3 + y −√ 2 = 0. This is
2 2 2 2
satisfied for y = 1. We have (x + 1) = 2(x − 1) , or x + 1 = ± 2(x − 1). Thus
the roots are:
√ p √ √ p √
2 ± i 10 + 8 2 − 2 ± 8 2 − 10
x1,2 = , x3,4 = .
2 2

6.22. If α5 = 1 then |α| = 1, hence ᾱ = 1/α and therefore T r(α) = α + α−1 .

If α 6= 1 then α4 + α3 + α2 + α + 1 = 0 or α2 + α + 1 + α−1 + α−2 = 0. Putting
t := α+α−1 we find√t2 = α2 +2+α−2 and conclude that 0 = 1+t+t2 −2 = t2 +t−1.
Thus t± = (−1 ± 5)/2, and we have two quadratic equations α2 − t± α + 1 = 0
for the four roots α 6= 1 of x5 − 1. Thus
√ √
5 + 1 5−1
x5 − 1 = (x − 1) (x2 + x + 1) (x2 − x + 1).
2 2
√ √
6.24b. x2 − ix + 1 = 0 has solutions x = (i ± i2 − 4)/2 = (1 ± 5)i/2.
6.27. The norms N = a2 + b2 < 14 of Gaussian integers a + bi and the corre-
sponding Gaussian integers are:
N = 0 = 02 + 02 , a + bi = 0
N = 1 = 12 + 02 , a + bi = ±1, ±i,
N = 2 = 12 + 12 , a + bi = ±1 ± i,
N = 4 = 22 + 02 , a + bi = ±2, ±2i,
N = 5 = 21 + 12 , a + bi = ±2 ± i, ±1 + ±2i,
N = 8 = 22 + 22 , a + bi = ±2 ± 2i,
N = 9 = 32 + 02 , a + bi = ±3, ±3i,
N = 10 = 32 + 12 , a + bi = ±3 ± i, ±1 ± 3i,
N = 13 = 32 + 22 , a + bi = ±3 ± 2i, ±2 ± 3i.

5
 Homework 6.
2
1. On the elliptic curve y = (x − 1)(x − 2)(x − 3), find all points P which satisfy
the condition P + P = 0 with respect to the group law + on the curve. Show that
these points form a subgroup isomorphic to the Klein group V4 .
These are points fixed under the symmetry P 7→ −P defined as (x, y) 7→ (x, −y)
and thus correspond to y = 0. There are four such points: the zero element
at infinity and the three finite symmeric points P1, P2 , P3 defined by (x, y) =
(1, 0), (2, 0), (3, 0). They are on the same line y = 0 and thus P1 + P2 + P3 = 0.
This P1 + P2 = P3 , etc. so that the four points form a subgroup. An isomorphism
with the symmetry group V4 of a rectangle is established by sending P1 and P2 to
the reflections of the rectangle about its symmetry lines and sending P 3 to their
product — the central symmetry.
2. Use the algorithm from the proof of the Newton-Waring theorem to express
r + s4 + t4 + u4 as a polynomial in the elementary symmetric functions of the four
4

variables r, s, t, u.

r 4 + s4 + t4 + u4 = s41 − 4(r 3 s + ...) − 6(r 2 s2 + ...) − 12(r 2 st + ...) − 24rstu

= s41 − 4s21s2 + (8 − 6)(r 2 s2 + ...) + (20 − 12)(r 2 st + ...) + (48 − 24)rstu
= s41 − 4s21 s2 + 2s22 + (20 − 12 − 4)(r 2 st + ...) + (48 − 24 − 12)rstu
= s41 − 4s21 s2 + 2s22 + 4s1 s3 + (48 − 24 − 12 − 16)rstu
= s41 − 4s21 s2 + 2s22 + 4s1 s3 − 4s4 .
7ab. By the fundamental theorem of algebra a complex polynomial factors as
(z − z1 )...(z − zn ) where zi are complex roots. A polynomial with real coefficients is
invariant under the complex conjugation and thus has an invariant set of roots and
their multiplicities. Thus each root is either real or comes in a complex conjugate
pair. In the latter case the product of the factors (z −a)(z −ā) = z 2 −T r(a)z +N (a)
has real coeffcieients. In particular, irreducible polynomials in R[x] are either linear
or quadratic with negative discriminant.
7c. x2 − 2 and x3 − 2 factor in R[x] but are irreducible in Q[x]. Indeed, if
reducible, they would have a linear factor and thus a root in Q, which they don’t.

6
 Homework 7.
7.8. Permitations of 1, 2, 3 also permute the pairs (1, 2), (1, 3), (2, 3) and thus
permute the factors (ri − rj ) in the product δ (but possibly reverse their signs).
Thus σ(δ) = ±δ.
7.9. One way to check that δ 2 (r1 , r2 , r3 ) = 4p3 − 27q 2 , where ri are the roots
of the polynomial x3 − px + q, is to substitute δ = (r1 − r2 )(r1 − r3 )(r2 − r3 ) =
(r1 − r2 )(2r1 + r2 )(r1 + 2r2 ) on the left and p = −(r1 + r2 )r3 − r1 r2 = r12 + r1 r2 + r22
and q = −r1 r2 r3 = r12 r2 + r1 r22 on the right.
7.12. If f¯ =
P
h∈H h(f ) then for any σ ∈ H

σ(f¯) = g(f ) = f¯.

X X X
(σh)f = g(f ) =
h∈H σ−1 g∈H g∈H

7.13. To prove Mathieu’s theorem, put p(r1 , ..., rn ) = r1n−1 r2n−2 ...rn−1
1
rn0 . Then
n−1 0
for all σ ∈ Sn the monomials σ(p) = rσ(1) ...rσ(n) are distinct (and hence linearly
P
independent). For a given subgroup H ∈ Sn put f = p̄ = h∈H h(p). If σ ∈ / H
then σ(f ) 6= f since it contains the monomial σ(p) which is not present in f . If
σ ∈ H then σ(f ) = f by 7.12. Thus (Sn )f = H.
7.10c is a special case of 7.13 with n = 3.

Homework 8.
8.1. Cycle structures of permutations from Sn correspond to partitions n (i.e.
4 = 4 = 3 + 1 = 2 + 2 = 2 + 1 + 1 = 1 + 1 + 1 + 1). Permutation with a given cycle
structure correcponding to a partition P with m1 ones, m2 twos, m2 threes, etc. are
obtained by fitting 1, 2, 3, ..., n (where n = m1 + 2m2 + 3m3 + ...) in any order into
the format of m1 cycles of length 1, m2 cycles of length 2, etc. Yet two fittings yield
the same permutation if they are obtained from each other by cyclic shifts within
the cycles and/or by permuting the mk cycles of the same length k = 1, 2, 3, ....
Thus the total number of permutations of 1, 2, ..., n with a given cycle structure is
equal to
n!
#(P ) = m1
1 (m1 )! 2m2 (m2 )! 3m3 (m3 )! · · ·

7
In the example of n = 4 we have:

P = 4, # = 24/4 : (1234), (1324), (2134), (2314), (3124), (3214)

P = 3 + 1, # = 24/3 : (123), (213), (124), (214), (134), (314), (234), (324)

P = 2 + 2, # = 24/8 : (12)(34), (13)(24), (14)(23)
P = 2 + 1 + 1, # = 24/4 : (12), (13), (14), (23), (24), (34)
P = 1 + 1 + 1 + 1, # = 24/24 : the identity permutation.
8.7. There are 3 ways to partition {1, 2, 3, 4} into two pairs: {12}{34}, {13}{24},
{14}{23}. Respectively, the S4 -orbit of G = (r1 + r2)(r3 + r4) consists of 3 elements:

G, G0 = (r1 + r3 )(r2 + r4 ), G00 = (r1 + r4 )(r2 + r3 ).

Thus (X − G)(X − G0 )(X − G00 ) = X 3 + AX 2 + BX + C, where A, B, C are

polynomials of elementary symmetric functions in r1 , r2 , r3 , r4 . Assuming that ri ’s
are roots of x4 + cx2 + dx + e (and therefore r1 + r2 + r3 + r4 = 0), we find
G = −(r1 +r2 )2 and similarly G0 = −(r1 +r3 )2 , G00 = −(r1 +r4 )2. Thus the equation
x4 +cx2 +dx+e = 0 can be solved as follows. First, compute A, B, C as polynomials
in the elementary symmetric functions c, −d, e. Next find G, G0 , G00 as roots of the
cubic equation X 3 + AX 2 + BX + C = 0. Then compute r1 + r2 , r1 + r3 , r1 + r4 as
square roots of −G, −G0 , −G00 . Finally solve these linear equations (together with
r1 + r2 + r3 + r4 = 0) for r1 , r2 , r3 , r4 .
8.6a. Similarly to the case of G, the orbit of F consists of 6 elements

±F = ±(r1 − r2 )(r3 − r4 ), ±F 0 = ±(r1 − r3 )(r2 − r4 ), ±F 00 = ±(r1 − r4 )(r2 − r3 ).

8.10. Let a group G acts by permutations on a set S, and let H = Ga be

the stabiliser of a ∈ S. For any b in the orbit O of a, the elements of the group
G mapping a to b have the form gh where g is any particular element with this
property, and h is any element from the stabilizer of a. Thus the number of the
elements of G mapping a to b does not depend on b ∈ O and is equal to the number
of elements |H|. Thus we have |G| = |H| × |O|.

8
 Homework 9.
9.8 cde. The characteristic polynomial det(λI − T ) = λ3 + aλ2 + bλ + c has odd
degree and thus has a real root λ0 . Thus there exists x0 ∈ R3 , |x0 | = 1, such that
T x0 = λ0 x0 . Since |T x0 | = |x0 |, we have λ0 = ±1. If λ0 = −1, then the other two
roots have to be 1 and −1 since the product of all 3 roots equals det(T ) = 1. (If the
other two roots were complex conjugate, their product would’ve been positive, and
the product of all 3 roots - negative). Thus in either case there is a unit eigenvector
v with the eigenvalue 1. Since T preserves inner products and preserves the line
spanned by v, T preserves the plane orthogonal to v and acts on it as an orthogonal
transfromation with the determinant 1. Thus picking any orthonormal basis in this
plane in the role of u and w, we find that T has in the basis v, u, w the required
matrix.
9.9. Axis of rotation of a (non-identity) rotation S is the eigen-space corre-
sponding to the eigenvalue 1. To show that RS = SR implies that R fixes the
rotation axis of S, we prove a more general statement: any eigen-space of a lin-
ear operator S is invariant under any operator R commuting with S. Indeed, if
Sx = λx then y := Rx also satisfies Sy = SRx = RSx = λRx = λy.
9.15. The rotation group G of the cube acts on the set vertices of the cube and
acts transitively, i.e. all the 8 vertices form one orbit. The stabilizer G v of one
vertex v consists of the 3 rotations (through the angles 0 and ±2πi/3) about the
diagonal containing v. Thus |G| = 8 × 3 = 24. Next, there is a homomorphism
φ : G → S4 defined by the action of G on the set of the 4 diagonals by permutations.
To prove that φ is an isomorphism, observe that each of the 3 types of non-indetity
rotations in G permutes some diagonals in a non-trivial way. Thus the kernel
ker(φ) := φ−1(e) ⊂ G contains only the identity transformation. This implies that
φ is injective, and since |S4| = 24, it is surjective as well.

9
 Homework 10.
1. Show that the rotation group of the dodecahedron is isomorphic to the group
A5 of all even permutations of 5 symbols (it is called the alternating subgroup in
S5 ). Hint: the rotations permute 5 cubes formed by the diagonals in the pentagonal
faces of the dodecahedron (each cube having one edge in each of the 12 faces of the
dodecahedron).
Outline. The hint describes the construction of a homomorphism f : G → S5
from the rotation group of the dodecahedron to the permutation group. We have
|G| = (# of faces) × (# of rotations preserving a face) = 12 × 5 = 60 and |A5 | =
|S5|/2 = 120/2 = 60. It suffices therefore to prove that f −1 (e) = {e} (injectivity)
and that f (G) ⊂ A5 . The group G consists of the identity element and rotations
through the angles 2π/5 (about the axes passing through the centers of faces),
2π/3 (about the axes passing through vertices) and 2π/2 (about the axes passing
through the midpoints of edges). Since G acts transitively on the sets of faces,
vertices and edges, it suffices to consider only one rotation of each kind and check
that it induces a non-trivial even permutation of the 5 cubes.
2. Find the greatest common divisor of 50339 and 128243.
The Euclidean algorithm shows in a few steps that the numbers are relatively
prime, while an attempt to use prime factorization will require trying all prime
factors up to 71 since 128243 = 257 × 499 and 50339 = 71 × 709 (in fact it was
meant to be 50399 = 101 × 499 which would make the answer more interesting).
3. Referring to the picture on p. 105 of the book, find the ratio of the areas of
the star ACEBD and the small pentagon FGHIJ.
The key point is to notice similarity of the triangles DAC and CDH (due to the
equality of angles ∠DAC = π/5 = ∠CDB). Taking |DC| for the unit of length
and denoting |CH|
√ = x we obtain the “golden ratio” proportion x/1 = 1/(1 + x)
and find x = ( 5 − 1)/2. Then, taking the triangle IDJ for the (new!) unit of
area, we find the area of J IH equal x, the area of J AH equal (1 + x)2 and the area
of J F GH equal
√ (1 + x)2 − 1 = 2x + x2 . Thus in these units the area
√ of the pentagon
2 2
is 3x + x = 5 and the area √ of the star is 5 + 3x + x = 5 + 5. Therefore the
ratio in question is equal to 5 + 1.

10
 10.13+10.14. If m is not divisible by prime p then (p, m) = 1 and by Euclid’s
Lemma 1 = kp + lm for some k and l. Then n = kpn + lmn is divisible by p
whenever mn is divisible by p. Moreover, taking this for the base of induction
on r, and assuming that p divides the product a1...ar with r > 2, but does not
divide a1 , ..., ar−1 we conclude, by the induction hypothesis, that p does not divide
m = a1...ar−1 and therefore divides n = ar by the previous argument.
Homework 11.
11.7d. In Z[x], the divisors of x are x and 1, and the divisors of 2 are 2 and
1. Thus the only common divisor of x and 2 is 1, but it cannot be written in the
form d(x) = 2r(x) + xs(x) with r, s ∈ Z[x]. Indeed, d(0) = 2r(0) ∈ 2Z is even, and
therefore d 6= 1 which is odd.
11.16. Applying Euclid’s lemma to f, g ∈ F [x], we find their greatest common
divisor h(x) = r(x)f (x) + s(x)g(x) ∈ F [x] − {0}. Put x = α ∈ E ⊃ F . Since
f (α) = g(α) = 0, we find h(α) = 0. Thus we the greatest common divisor of f and
g is non-constant.
11.19. An elementary step in the “long division” of xm − 1 by xn − 1 for m ≥ n
consists in replacing xm − 1 with the remainder (xm − 1) − xm−n (xn − 1) = xm−n − 1.
In other words, it is equivalent to replacing the pair m, n with m − n, n which is
an elementary step in the “long division” of m by n. Thus the ultimate remainder
upon division of xm − 1 by xn − 1 is xr − 1 where r is the remainder upon division
of m by n. In particular, the remainder xr − 1 is zero if and only if r = 0.
11.20ab. Here are factorizations of xn − 1 (with Φn (x) in bold) for 1 ≤ n ≤ 12:
(x − 1), (x − 1)(x + 1), (x − 1)(x2 + x + 1), (x − 1)(x + 1)(x2 + 1),
(x − 1)(x4 + x3 + x2 + x + 1), (x − 1)(x + 1)(x2 + x + 1)(x2 − x + 1),
(x − 1)(x6 + x5 + x4 + x3 + x2 + x + 1),
(x − 1)(x + 1)(x2 + 1)(x4 + 1), (x − 1)(x2 + x + 1)(x6 + x3 + 1),
(x − 1)(x + 1)(x4 + x3 + x2 + x + 1)(x4 − x3 + x2 − x + 1),
(x − 1)(x10 + x9 + x8 + x7 + x6 + x5 + x4 + x3 + x2 + x + 1),
(x − 1)(x + 1)(x2 + 1)(x2 + x + 1)(x2 − x + 1)(x4 − x2 + 1).
This table suggests that xn − 1 = d|n Φd (x), and that deg Φn = φ(n) (so that in
Q
P
particular n = d|n φ(d)).

11
 Homework 12.
12.6. a ≡ b ( mod n), c ≡ d ( mod n) means a = b + kn, c = d + ln for some
k, l ∈ Z. Then a + c = b + d + (k + l)n and ac = bd + (bl + ck + kl)n and hence
a + c ≡ b + d ( mod n) and ac ≡ bd ( mod n).
12.10. φ(1) = 1 (by definition), φ(2) = 1, φ(3) = 2, φ(4) = 2, φ(5) = 4,
φ(6) = 2 , φ(7) = 6, φ(8) = 4, φ(9) = 6, φ(10) = 4, φ(11) = 10, φ(12) = 4,
φ(13) = 12, φ(14) = 6, φ(15) = 8, φ(16) = 8, φ(17) = 16, φ(18) = 6, φ(19) = 18,
φ(20) = 8, φ(21) = 12, φ(22) = 10, φ(23) = 22, φ(24) = 12.
12.11. When p is prime φ(pn ) = pn − pn−1 since among the pn remainders
0, 1, 2, ..., pn −1 those and only those pn−1 which are multiples of p are not relatively
prime to p.
12.17b. The Euclidean algorithm for 1 = (147, 17) yields the remainders: 11 =
147 − 8 · 17, 6 = 17 − 11, 5 = 11 − 6, 1 = 6 − 5. Thus

1 = 2 × 6 − 11 = 2 · 17 − 3 · 11 = 26 · 17 − 3 · 147.

Thus 26 · 17 ≡ 1 ( mod 147).

12.24. U5 := (Z/5Z)× is cyclic of order 4 generated by 2:

20 = 1, 21 = 2, 22 = 4, 23 ≡ 3, 24 ≡ 1 ( mod 5).

The group U8 is isomorphic to the Cartesian product of two cyclic groups of order
2 since squares of all elements in U8 equal 1:

12 = 1, 32 = 9 ≡ 1, 52 = 25 ≡ 1, 72 = 49 ≡ 1 ( mod 8).

12
 Homework 13.
1. Prove that the group of units of any finite field is cyclic.
Let G = F × be the group of units of a finite field F of cardinality q. Let r be
the exponent of the group G, i.e. the minimal positive integer such that a r = 1
for all a ∈ G. Since G is abelian, it contains a cyclic subgroups of cardinality r
(as we proved in the class by constructing an element of order L.C.M.(m, n) out of
two elements of orders m and n). Thus it suffices to show that ar = 1 cannot hold
true for all a ∈ G if r < q − 1. But this follows from the fact that the polynomial
xr − 1 ∈ F [x] cannot have q − 1 = |G| distinct roots when its degree r < q − 1.
13.13. Any non-identity element in a group G generates a non-trivial cyclic
subgroup whose cardinality divides that of G (by Lagrange’s theorem). When the
latter cardinality is prime, this cyclic subgroup coincides therefore with the whole
group.
12.30. The cyclotomic polynomial Φn (x) is defined as the monic polynomial
whose roots are primitive complex n-th roots of 1. All n-th roots of 1 form in C ×
a cyclic subgroup Cn of order n. Each n-th root ξ of 1 is a primitive root of order
d equal to the cardinality of the cyclic subgroup generated by ξ. In particular, d is
a divisor of n (by Lagrange’s theorem). Vice versa, when d|n, primitive d-th roots
of 1 are n-th roots of 1 as well. Thus
Y Y
Φd (x) = (x − ξ) = xn − 1,
d|n ξ∈C: ξ n=1

and deg Φn = φ(n), the number of generators in Cn . Since Φ1 = x − 1 has integer

coefficients, it follows by induction on n that Φn has integer coefficients
Q too. Indeed,
by the induction hypothesis we may assume that Pn := d|n,d<n Φd is a monic
polynomial with integer coefficients. The Φn = (xn − 1)/Pn has integer coeffcients
too since the long division algorithm by a monic polynomial produces no fractional
coefficients.

13
 Homework 14.
2 2
14.18. Let n = a + b be diviible by a prime p of the form 4k − 1. Then p
remains prime in Z[i], n factors as (a + ib)(a − ib) in Z[i] and therefore p divides
at least one of a ± ib by the definition of prime elements. Thus both a and b are
divisible by p and n is divisible by p2 and hence is not square-free. In other words,
a square-free product n = p1 ...pr of prime numbers containing a prime of the form
4k − 1 is not the sum of two squres. Vice versa, if none of the prime factors is of
the form 4k − 1, then each of them can be written as the sum of two squares (by
Fermat’s result) and thus n = a2 + b2 for some natural a and b too. Indeed, when
x = (α + iβ)(α − iβ) and y = (γ + iδ)(γ − iδ) then xy = (a + ib)(a − ib) with
a + ib = (α + iβ)(γ + iδ).
16.5. To divide the circle into n = 2, 3, 4, 6, 8, 12 equal parts using straightedge
and compass, for n = 2 — produce a diameter, for n = 4 — produce the diameter
perpendicular to the first one; for n = 6 — use the arcs subtended by the chords
equal to the radius of the circle, for n = 3 — use every 2nd vertex of the 6-gon, for
n = 8, 12 — bisect the arcs constructed respectively for n = 4, 6.
16.15. If α is a real root of a cubic polynomial irredicuble in Q[x] then the
extension field F = Q(α) has dimension 3 as a vector space over Q. If E ⊃ F is
a field of dimension m over Q then m = 3 dimF E is divisible by 3. On the other
hand, if α is constructible, then it belongs to an extension field E ⊃ Q obtained
from Q by consecutive adjoinig roots of quadratic polynomials with previously
constructed coefficients. Thus dimQ E is a power of 2 and is not divisible by 3.
16.17. We have 1 = 2 × 3 − 1 × 5 and hence 1/15 = 2/5 − 1/3. The angle 2π/15
can therefore be constructed by duplicating the central angle 2π/5 of the regular
pentagon and subtracting the central angle 2π/3 of the regular triangle.
18.14. Let t = ξ + ξ −1 where ξ = cos(2π/7) + i sin(2π/7) is the primitive 7-th
root of 1. Then t2 = ξ 2 + 2 + ξ −2, t3 = ξ 3 + 3ξ + 3ξ −1 + ξ −3 and therefore
t3 + t2 − 3t − 1 = ξ 3 + ξ 2 + ξ + 1 + ξ −1 + ξ −2 + ξ −3 = 0.
The polynomial x3 + x2 − 3x − 1 has no rational roots (since none of x = ±1 is a
root) and is therefore irreducible in Q[x]. Thus t is a root of an irreducible over Q
cubic polynomial and is not constructible by 16.5.

14