Mathematical-Reasoning ยท Unit 25 ยท Video 2 ยท Interactive Practice

Image and Preimage: Why Squaring Splits an Interval

IKey Formulas

NameDefinitionMembership test
Image of AโІSA \subseteq Sf(A)={f(x)โˆฃxโˆˆA}โІTf(A) = \{ f(x) \mid x \in A \} \subseteq Tyโˆˆf(A)y \in f(A) exactly when y=f(x)y = f(x) for at least one xโˆˆAx \in A
Preimage of CโІTC \subseteq Tfโˆ’1(C)={xโˆˆSโˆฃf(x)โˆˆC}โІSf^{-1}(C) = \{ x \in S \mid f(x) \in C \} \subseteq Sxโˆˆfโˆ’1(C)x \in f^{-1}(C) exactly when f(x)โˆˆCf(x) \in C
For g(x)=x2g(x) = x^2g([1,2])=[1,4]g([1, 2]) = [1, 4]gโˆ’1([1,4])=[โˆ’2,โˆ’1]โˆช[1,2]g^{-1}([1, 4]) = [-2, -1] \cup [1, 2]

Key Insight: The preimage fโˆ’1(C)f^{-1}(C) is defined for every function. It uses ff alone, so no inverse function is needed: g(x)=x2g(x) = x^2 has none, yet gโˆ’1([1,4])g^{-1}([1, 4]) is a well-defined set.

IIReading Images and Preimages off the Graph

An interval of inputs maps to one interval of heights, but a band of heights can pull back to two pieces.

IIIBuilding a Preimage One Input at a Time

A preimage needs no inverse: test each input, and keep it when its output lands in CC.

IVQuiz Questions

Problem 1 ยท An Image That Crosses Zero

Given: g:Rโ†’Rg: \mathbb{R} \to \mathbb{R}, g(x)=x2g(x) = x^2. Find the image g([โˆ’1,3])g([-1, 3]).

โœ… Correct! The interval contains 00, so the lowest height is g(0)=0g(0) = 0; the highest is g(3)=9g(3) = 9.
โŒ Not quite. Squaring the end points misses the bottom of the parabola: 0โˆˆ[โˆ’1,3]0 \in [-1, 3] and g(0)=0g(0) = 0.
โŒ Not quite. Every image under gg consists of squares, and no square is negative.
โŒ Not quite. That is the set of inputs whose squares lie in [0,9][0, 9], a preimage. The image is a set of outputs.
Show solution

On [โˆ’1,0][-1, 0] the graph falls from g(โˆ’1)=1g(-1) = 1 to g(0)=0g(0) = 0; on [0,3][0, 3] it rises from 00 to g(3)=9g(3) = 9. Every height between 00 and 99 is reached, and no other height.

g([โˆ’1,3])=[0,9].g([-1, 3]) = [0, 9].

Problem 2 ยท The Preimage of a Band

Given: g(x)=x2g(x) = x^2 on R\mathbb{R}. Find the preimage gโˆ’1([4,9])g^{-1}([4, 9]).

โœ… Correct! 4โ‰คx2โ‰ค94 \le x^2 \le 9 holds for 2โ‰คxโ‰ค32 \le x \le 3 and, symmetrically, for โˆ’3โ‰คxโ‰คโˆ’2-3 \le x \le -2.
โŒ Close, butโ€ฆ taking square roots forgets the negative inputs: (โˆ’2.5)2=6.25โˆˆ[4,9](-2.5)^2 = 6.25 \in [4, 9].
โŒ Not quite. Inputs near 00 are too small: g(1)=1โˆ‰[4,9]g(1) = 1 \notin [4, 9], so the stretch (โˆ’2,2)(-2, 2) is left out.
โŒ Not quite. That is the image g([4,9])g([4, 9]). The preimage collects inputs whose outputs lie in [4,9][4, 9].
Show solution

We need every real xx with 4โ‰คx2โ‰ค94 \le x^2 \le 9. The condition x2โ‰ค9x^2 \le 9 means โˆ’3โ‰คxโ‰ค3-3 \le x \le 3, and x2โ‰ฅ4x^2 \ge 4 means xโ‰คโˆ’2x \le -2 or xโ‰ฅ2x \ge 2. Both hold on two separate stretches:

gโˆ’1([4,9])=[โˆ’3,โˆ’2]โˆช[2,3].g^{-1}([4, 9]) = [-3, -2] \cup [2, 3].

On the graph, these are the xx values where the parabola lies between the horizontal lines y=4y = 4 and y=9y = 9.

Problem 3 ยท A Shifted Parabola

Given: h:Rโ†’Rh: \mathbb{R} \to \mathbb{R}, h(x)=x2โˆ’2xh(x) = x^2 - 2x, so h(x)=(xโˆ’1)2โˆ’1h(x) = (x - 1)^2 - 1. Find hโˆ’1([0,3])h^{-1}([0, 3]).

โœ… Correct! 0โ‰ค(xโˆ’1)2โˆ’1โ‰ค30 \le (x-1)^2 - 1 \le 3 means 1โ‰ค(xโˆ’1)2โ‰ค41 \le (x - 1)^2 \le 4, so xโˆ’1โˆˆ[โˆ’2,โˆ’1]โˆช[1,2]x - 1 \in [-2, -1] \cup [1, 2].
โŒ Close, butโ€ฆ that is only the right-hand piece. h(โˆ’0.5)=1.25โˆˆ[0,3]h(-0.5) = 1.25 \in [0, 3], so inputs left of the vertex qualify too.
โŒ Not quite. The lower bound matters here: h(1)=โˆ’1โˆ‰[0,3]h(1) = -1 \notin [0, 3], so inputs near the vertex are left out.
โŒ Not quite. A preimage is defined for every function; it never needs an inverse function. Test the inputs: h(3)=3โˆˆ[0,3]h(3) = 3 \in [0, 3].
Show solution

xโˆˆhโˆ’1([0,3])x \in h^{-1}([0, 3]) exactly when 0โ‰ค(xโˆ’1)2โˆ’1โ‰ค30 \le (x-1)^2 - 1 \le 3, that is, 1โ‰ค(xโˆ’1)2โ‰ค41 \le (x - 1)^2 \le 4.

Let u=xโˆ’1u = x - 1. Then 1โ‰คu2โ‰ค41 \le u^2 \le 4 holds exactly for uโˆˆ[โˆ’2,โˆ’1]โˆช[1,2]u \in [-2, -1] \cup [1, 2], as in the video's gโˆ’1([1,4])g^{-1}([1, 4]). Adding 11 gives

hโˆ’1([0,3])=[โˆ’1,0]โˆช[2,3].h^{-1}([0, 3]) = [-1, 0] \cup [2, 3].

Check: h(โˆ’1)=3h(-1) = 3, h(0)=0h(0) = 0, h(2)=0h(2) = 0 and h(3)=3h(3) = 3.

Problem 4 ยท What the Definitions Guarantee

Given: a function f:Sโ†’Tf: S \to T, a subset AโІSA \subseteq S and a subset CโІTC \subseteq T. Which statement is true for every such ff, AA and CC?

โœ… Correct! The image collects outputs, so it lives in the codomain; the preimage collects inputs, so it lives in the domain.
โŒ Not quite. The preimage uses only ff: g(x)=x2g(x) = x^2 is not a bijection, yet gโˆ’1([1,4])=[โˆ’2,โˆ’1]โˆช[1,2]g^{-1}([1, 4]) = [-2, -1] \cup [1, 2].
โŒ Not quite. A nonempty target can miss the range: gโˆ’1({โˆ’1})=โˆ…g^{-1}(\{-1\}) = \emptyset for g(x)=x2g(x) = x^2.
โŒ Not quite. yโˆˆf(A)y \in f(A) says only that some element of AA is sent to yy. For g(x)=x2g(x) = x^2, 4โˆˆg([1,2])4 \in g([1, 2]), yet g(โˆ’2)=4g(-2) = 4 and โˆ’2โˆ‰[1,2]-2 \notin [1, 2].
Show solution

(c) is true straight from the definitions: f(A)={f(x)โˆฃxโˆˆA}f(A) = \{ f(x) \mid x \in A \} consists of values of ff, which lie in TT, and fโˆ’1(C)={xโˆˆSโˆฃf(x)โˆˆC}f^{-1}(C) = \{ x \in S \mid f(x) \in C \} is carved out of SS.

(a), (b) and (d) each fail for g(x)=x2g(x) = x^2: gg has no inverse function but has preimages; gโˆ’1({โˆ’1})=โˆ…g^{-1}(\{-1\}) = \emptyset; and โˆ’2-2 is sent to 4โˆˆg([1,2])4 \in g([1, 2]) without lying in [1,2][1, 2].

Solved: 0 / 4