Scribd
Upload
43 views

Math 461 F Spring 2011 Impossible Constructions Drew Armstrong

(1) The document discusses classical construction problems in geometry that were proven impossible using straightedge and compass, including doubling the cube, trisecting an angle, and constructing a regular heptagon. These problems were shown to be governed by cubic polynomials with rational coefficients but no rational roots, implying the roots are not constructible. (2) A key lemma is presented, showing that if a cubic polynomial with rational coefficients has a root in a quadratic field extension, it must also have a rational root. This lemma is used to prove a theorem stating that if a cubic polynomial with rational coefficients has no rational roots, then its roots are not constructible with straightedge and compass. (3) Each of the

Uploaded by

Vinay
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
43 views

Math 461 F Spring 2011 Impossible Constructions Drew Armstrong

(1) The document discusses classical construction problems in geometry that were proven impossible using straightedge and compass, including doubling the cube, trisecting an angle, and constructing a regular heptagon. These problems were shown to be governed by cubic polynomials with rational coefficients but no rational roots, implying the roots are not constructible. (2) A key lemma is presented, showing that if a cubic polynomial with rational coefficients has a root in a quadratic field extension, it must also have a rational root. This lemma is used to prove a theorem stating that if a cubic polynomial with rational coefficients has no rational roots, then its roots are not constructible with straightedge and compass. (3) Each of the

Uploaded by

Vinay
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Math 461 F Spring 2011

Impossible Constructions Drew Armstrong


√
In class we proved that 2 is not a rational number (the edge and diagonal of a square
are incommensurable). This “crisis of incommensurables” forced the Greeks to base their
mathematics on the concept of “length” instead of “number”. In particular, the Euclidean
system was based on constructions with straightedge-and-compass, which are restricted to
• drawing the line containing two given points,
• drawing the circle with a given center and radius,
• drawing the intersection points of lines and circles.
(Why did they do this? Well, you can’t prove anything if you don’t have some rules.)
However, this system quickly ran into trouble. Classical mathematics was unable to solve
the following problems, and it was not for lack of trying:
(1) to draw a square with area equal to a given circle,
(2) to draw the edge of a cube with volume double that of a given cube,
(3) to draw a line that trisects the angle between two given lines,
(4) to draw the regular heptagon.
Many people suspected that these problems were impossible, but no one could prove it
until Fermat and Descartes (La Géométrie, 1637) finally healed the rift between geometry
and algebra by introducing coordinate geometry. In this note, I will use coordinate
geometry to prove that problems (2),(3),(4) are impossible; Lindemann (1882) proved
that (1) is impossible, but his proof involves too many integrals for this class.
The main tool we will use is the following lemma on quadratic field extensions.
√
Lemma. Consider a quadratic field extension√F ⊆ F [ c] and a cubic polynomial f (x)
with coefficients in F . If f (x) has a root in F [ c], then it also has a root in F .
√
Proof. Suppose we have f (α) = 0 with α ∈ F [ c]. If α ∈ F we are done. Otherwise, let
√ √
a + b c = a − b c be the conjugation map of the extension and note that
f (α) = f (α) = 0 = 0.
Hence α 6= α is another root of f (x). Using the Factor Theorem we can write
f (x) = (x − α)(x − α)(x − β),
√
where β is in F [ c]. We claim in fact that β ∈ F . If not, then β is another solution to
f (x) = 0, not equal to α, α, β, which is impossible. 
Note: This is related to the fact that every cubic polynomial with real coefficients has
a real root. We can now prove the main result.
Theorem. Let f (x) ∈ Q[x] be a cubic polynomial with rational coefficients, but no ratio-
nal roots. Then the roots of f (x) are not constructible with straightedge-and-compass.
Proof. Suppose, for contradiction, that f (α) = 0 for some constructible α. Then there
exists a chain of quadratic extensions
Q = F0 ⊆ F1 ⊆ F2 ⊆ · · · ⊆ Fk ,
such that α ∈ Fk . Since the coefficients of f (x) are in Q ⊆ Fk−1 , the lemma implies that
f (x) has a root in Fk−1 . By repeatedly applying the lemma, we conclude that f (x) has a
root in F0 = Q, which is a contradiction. 
But how does this theorem apply to the classical construction problems? It turns out
that each (except (1)) is governed by a certain cubic polynomial with rational coefficients.

(2) Doubling the Cube. If any√cube can be doubled, then the √ unit cube can be doubled,
which implies that the number 3 2 ∈ R is constructible. But 3 2 is a root of the cubic
polynomial x3 − 2 ∈ Q, and I claim that this polynomial has no rational root. Indeed,
suppose that x = a/b is a root with a, b ∈ Z and a, b coprime. Then we have
a3
− 2 = 0,
b3
a3 − 2b3 = 0,
a3 = 2b3 .
Since a divides 2b3 and a is coprime to b we conclude that a divides 2. Also, since b divides
a3 , we conclude that b = ±1. Thus the only possibilities for a/b are
a ±1, 2
= = ±1, ±2.
b ±1
We can plug in the numbers ±1, ±2 to see that none of them is√a root of x3 − 2. Hence this
polynomial has no rational root, and the theorem implies that 3 2 ∈ R is not constructible.
Hence it is impossible to “double the cube”.

(3) Trisecting the Angle. If an arbitrary angle can be trisected, then the angle π/3 can
be trisected, in which case the number cos(π/9) ∈ R is constructible. But putting θ = π/3
into the identity cos(3θ) = 4 cos3 θ − 3 cos θ yields
π  π  π  1
4 cos3 − 3 cos = cos = ,
9 9 3 2
and then setting u = 2 cos(π/9) gives us a cubic polynomial with rational coefficients:
u3 u 1
4 − 3 − = 0,
8 2 2
u3 − 3u − 1 = 0.
Using the above method, we see that the only possible rational roots are u = a/b with
a/b = ±1, and neither of these is a root. Hence the polynomial u3 − 3u − 1 ∈ Q[u]
has no rational root and the theorem implies that 2 cos(π/9), and hence cos(π/9), is not
constructible. We conclude that it is impossible to “trisect the angle”.

(4) Constructing the Regular Heptagon. If the regular 7-gon can be constructed, then
the regular 7-gon centered at the origin and containing the point (1, 0) can be constructed,
which implies that the point (cos(2π/7), sin(2π/7)) can be constructed. But we saw in
class that u = 2 cos(2π/7) is a root of the cubic polynomial
u3 + u2 − 2u − 1 ∈ Q[u].
You showed on Homework 4 that this polynomial has no rational root, so the theorem
implies that 2 cos(2π/7), and hence cos(2π/7), is not constructible. Hence it is impossible
to construct the regular heptagon.

You might also like