What is a Square Root?

We opened our exploration in real analysis with the notion of \sqrt 2. We could technically define it in the following manner:

\sqrt{2} := \sup\{x \in \mathbb Q : x^2 < 2\}.

However, it would take a ton of work to show that we do get the property (\sqrt 2)^2 = 2, and isn’t easily generalisable to other kinds of roots. Instead, we will take advantage of the intermediate value theorem, and define roots as inverses. The intuition is that

y = \sqrt[n]{x} \quad \iff \quad y^n = x.

Setting n = 2 and x = 2 gives us a definition for \sqrt 2, where \sqrt[2]{\cdot} \equiv \sqrt{\cdot} for brevity. But first, a warm-up exercise.

Theorem 1. For any n \in \mathbb N, the function f_n : \mathbb R \to \mathbb R defined by f_n(x) = x^n is continuous.

Proof. It is obvious that f_1 is continuous. Since products of continuous functions are continuous, f_n = f_1^n is continuous by induction.

How do we know these functions have inverses? When n is odd, life is nice.

Lemma 1. Fix i \in \mathbb N. Consider the sequence \{f_i(n)\}. Then f_i(n) \to \infty as n \to \infty. Furthermore, for odd i, f_i(-n) \to -\infty as n \to \infty. Finally, for even i, f_i(-n) \to \infty as n \to \infty.

Proof. By definition, f_i(n) = n^i. Fix M > 0. Use the Archimedean property to find N \in \mathbb N such that N > \max\{1, M\}. Then

f_i(N) = N^i > \max\{1, M^i\} \geq M.

Hence, for n > N, f_i(n) > f_i(N) > M.

Furthermore, if i is odd, then f_i(-n) = (-n)^i = -n^i = -f_i(n). It suffices to prove a more general result: if x_n \to \infty, then -x_n \to -\infty.

To that end, assume x_n \to \infty. Fix M > 0. Since x_n \to \infty, find N \in \mathbb N such that n > N implies x_n > M \iff -x_n < -M, which proves -x_n \to -\infty as required.

Finally, for even i, f_i(-n) = (-n)^i = n^i = f_i(n) \to \infty.

Theorem 2. Let n \in \mathbb N be odd. Then f_n : \mathbb R \to \mathbb R is bijective and f_n^{-1} := \sqrt[n]{\cdot} exists.

Proof. We first prove injectivity. Fix x,y \in \mathbb R with x < y.

  • If x \geq 0, then f_n(x) = x^n < y^n = f_n(y).
  • If x < 0 and y > 0, then f_n(x) = x^n < 0 < y^n = f_n(y).
  • If x < y < 0, then -x > -y > 0 implies -x^n = (-x)^n > (-y)^n = -y^n, which yields f_n(x) = x^n < y^n = f_n(y).

Thus f_n is strictly increasing, and thus injective. For surjectivity, fix y \in \mathbb R. Since f_n(\pm k) \to \pm \infty as k \to \infty, find M \in \mathbb N such that

f_n(-M)  < y <  f_n(M).

By the intermediate value theorem, there exists c \in (-M,M) such that y = f_n(c) = c^n.

A similar argument shows that for even n, if we restrict the domain of f_n to \mathbb R_{\geq 0}, we still get inverses.

Theorem 3. Let n \in \mathbb N be even. Then f_n : \mathbb R_{\geq 0} \to \mathbb R is bijective and f_n^{-1} := \sqrt[n]{\cdot} exists.

This will help us define rational exponents in a future post.

—Joel Kindiak, 19 Dec 24, 2138H

,

Published by


Responses

  1. Demystifying the Square Root – KindiakMath

    […] it is possible to define the function by , so that is bijective. This result, however, requires more effort to thoroughly […]

    Like

  2. Creating Euclidean Geometry – KindiakMath

    […] working model for Euclidean geometry, defining quadrances and spreads in general fields, then take square roots in the case and even employing trigonometric functions to obtain our classical notions of […]

    Like

Leave a comment