Mathematical-Reasoning ยท Unit 25 ยท Video 5 ยท Interactive Practice

Preimages Preserve Unions and Intersections Exactly

IKey Formulas

ResultKindReason in one line
fโˆ’1(CโˆฉD)=fโˆ’1(C)โˆฉfโˆ’1(D)f^{-1}(C \cap D) = f^{-1}(C) \cap f^{-1}(D)EqualityMembership of xx depends on the single value f(x)f(x)
fโˆ’1(CโˆชD)=fโˆ’1(C)โˆชfโˆ’1(D)f^{-1}(C \cup D) = f^{-1}(C) \cup f^{-1}(D)EqualityThe same chain, with "or" in place of "and"
AโІfโˆ’1(f(A))A \subseteq f^{-1}(f(A))InclusionxโˆˆAx \in A gives f(x)โˆˆf(A)f(x) \in f(A)
f(fโˆ’1(C))โІCf(f^{-1}(C)) \subseteq CInclusionAn element of the image comes from some xx with f(x)โˆˆCf(x) \in C

Key Insight: f(fโˆ’1(C))f(f^{-1}(C)) is the part of CC that ff actually reaches, and fโˆ’1(f(A))f^{-1}(f(A)) adds to AA every input that shares an output with an element of AA. Either round trip can change the set.

IIThe If-and-Only-If Chain for One Input

Every line of the chain has the same truth value, because each step only unpacks a definition.

xโˆˆfโˆ’1(CโˆฉD)x \in f^{-1}(C \cap D) start
โ€…โ€ŠโŸบโ€…โ€Šf(x)โˆˆCโˆฉD\iff f(x) \in C \cap D definition of preimage
โ€…โ€ŠโŸบโ€…โ€Šf(x)โˆˆC\iff f(x) \in C and f(x)โˆˆDf(x) \in D definition of intersection
โ€…โ€ŠโŸบโ€…โ€Šxโˆˆfโˆ’1(C)\iff x \in f^{-1}(C) and xโˆˆfโˆ’1(D)x \in f^{-1}(D) definition of preimage, twice
โ€…โ€ŠโŸบโ€…โ€Šxโˆˆfโˆ’1(C)โˆฉfโˆ’1(D)\iff x \in f^{-1}(C) \cap f^{-1}(D) definition of intersection
xโˆˆfโˆ’1(CโˆชD)x \in f^{-1}(C \cup D) start
โ€…โ€ŠโŸบโ€…โ€Šf(x)โˆˆCโˆชD\iff f(x) \in C \cup D definition of preimage
โ€…โ€ŠโŸบโ€…โ€Šf(x)โˆˆC\iff f(x) \in C or f(x)โˆˆDf(x) \in D definition of union
โ€…โ€ŠโŸบโ€…โ€Šxโˆˆfโˆ’1(C)\iff x \in f^{-1}(C) or xโˆˆfโˆ’1(D)x \in f^{-1}(D) definition of preimage, twice
โ€…โ€ŠโŸบโ€…โ€Šxโˆˆfโˆ’1(C)โˆชfโˆ’1(D)\iff x \in f^{-1}(C) \cup f^{-1}(D) definition of union

Here f(x)=(x2+2)โ€Šmodโ€Š8f(x) = (x^2 + 2) \bmod 8 on R8R_8, with C={1,2,3}C = \{1, 2, 3\} and D={3,4,5}D = \{3, 4, 5\}, as in the video's modulo 88 example.

IIIWhere the Round Trips Leak

fโˆ’1(f(A))f^{-1}(f(A)) grows past AA when an outside input shares an output; f(fโˆ’1(C))f(f^{-1}(C)) shrinks below CC when CC holds a non-output.

IVQuiz Questions

Problem 1 ยท Justify the Middle Step

Given: the chain proving fโˆ’1(CโˆฉD)=fโˆ’1(C)โˆฉfโˆ’1(D)f^{-1}(C \cap D) = f^{-1}(C) \cap f^{-1}(D) contains the step "f(x)โˆˆCf(x) \in C and f(x)โˆˆDf(x) \in D โ€…โ€ŠโŸบโ€…โ€Š\iff xโˆˆfโˆ’1(C)x \in f^{-1}(C) and xโˆˆfโˆ’1(D)x \in f^{-1}(D)." What justifies this step?

โœ… Correct! xโˆˆfโˆ’1(C)x \in f^{-1}(C) means exactly f(x)โˆˆCf(x) \in C, and likewise for DD; no property of ff is needed.
โŒ Not quite. Neither side of this step mentions an intersection; the definition of intersection is used in the steps before and after it.
โŒ Not quite. The theorem holds for every function. The step only rewrites f(x)โˆˆCf(x) \in C as xโˆˆfโˆ’1(C)x \in f^{-1}(C).
โŒ Not quite. No image appears in this step: f(x)f(x) is a single output, and the sets involved are preimages.
Show solution

By definition, fโˆ’1(C)={xโˆˆSโˆฃf(x)โˆˆC}f^{-1}(C) = \{ x \in S \mid f(x) \in C \}, so for xโˆˆSx \in S, xโˆˆfโˆ’1(C)โ€…โ€ŠโŸบโ€…โ€Šf(x)โˆˆCx \in f^{-1}(C) \iff f(x) \in C; the same holds with DD. Replacing each half of the "and" statement by its equivalent gives the step. The definition reads in both directions, which is why the whole proof can be a chain of โ€…โ€ŠโŸบโ€…โ€Š\iff.

Problem 2 ยท Image, Then Preimage

Given: g:Rโ†’Rg: \mathbb{R} \to \mathbb{R}, g(x)=x2g(x) = x^2, and A=[0,2]A = [0, 2]. Find gโˆ’1(g(A))g^{-1}(g(A)).

โœ… Correct! g(A)=[0,4]g(A) = [0, 4], and x2โˆˆ[0,4]x^2 \in [0, 4] exactly when โˆ’2โ‰คxโ‰ค2-2 \le x \le 2. Each negative input shares its square with a positive one in AA.
โŒ Not quite. Image and preimage need not cancel: g(โˆ’1)=1=g(1)g(-1) = 1 = g(1) with 1โˆˆA1 \in A, so โˆ’1โˆˆgโˆ’1(g(A))-1 \in g^{-1}(g(A)).
โŒ Close, butโ€ฆ that is g(A)g(A), the halfway point. Now take its preimage.
โŒ Not quite. g(3)=9โˆ‰[0,4]g(3) = 9 \notin [0, 4], so 33 is not in the preimage; the bound on xx is 4=2\sqrt{4} = 2.
Show solution

Image: on [0,2][0, 2] the squares rise from 00 to 44, so g(A)=[0,4]g(A) = [0, 4].

Preimage: xโˆˆgโˆ’1([0,4])x \in g^{-1}([0, 4]) exactly when 0โ‰คx2โ‰ค40 \le x^2 \le 4, that is, โˆ’2โ‰คxโ‰ค2-2 \le x \le 2.

gโˆ’1(g(A))=[โˆ’2,2]โŠ‹[0,2]=A.g^{-1}(g(A)) = [-2, 2] \supsetneq [0, 2] = A.

Problem 3 ยท Preimage, Then Image

Given: h:Rโ†’Rh: \mathbb{R} \to \mathbb{R}, h(x)=โˆฃxโˆฃh(x) = |x|, and C=[โˆ’3,2]C = [-3, 2]. Find h(hโˆ’1(C))h(h^{-1}(C)).

โœ… Correct! hโˆ’1(C)=[โˆ’2,2]h^{-1}(C) = [-2, 2], and โˆฃxโˆฃ|x| on [โˆ’2,2][-2, 2] takes exactly the values [0,2][0, 2]. The negative part of CC is never an output.
โŒ Not quite. f(fโˆ’1(C))f(f^{-1}(C)) is only the part of CC that the function reaches, and โˆฃxโˆฃ|x| never takes a negative value.
โŒ Close, butโ€ฆ that is hโˆ’1(C)h^{-1}(C). Apply hh to it.
โŒ Not quite. h(1)=1โˆˆCh(1) = 1 \in C, so 1โˆˆhโˆ’1(C)1 \in h^{-1}(C) and 1โˆˆh(hโˆ’1(C))1 \in h(h^{-1}(C)).
Show solution

Preimage: โˆฃxโˆฃโˆˆ[โˆ’3,2]|x| \in [-3, 2] means โˆ’3โ‰คโˆฃxโˆฃโ‰ค2-3 \le |x| \le 2; the left inequality always holds, so the condition is โˆฃxโˆฃโ‰ค2|x| \le 2, and hโˆ’1(C)=[โˆ’2,2]h^{-1}(C) = [-2, 2].

Image: as xx runs over [โˆ’2,2][-2, 2], โˆฃxโˆฃ|x| takes every value from 00 to 22, so

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

Problem 4 ยท Reading a Gap

Given: f:Sโ†’Tf: S \to T and CโІTC \subseteq T such that some element cโˆˆCc \in C is not f(x)f(x) for any xโˆˆSx \in S. Which statement must be true?

โœ… Correct! Every element of f(fโˆ’1(C))f(f^{-1}(C)) is an output of ff, and cc is not, so cโˆˆCc \in C but cโˆ‰f(fโˆ’1(C))c \notin f(f^{-1}(C)).
โŒ Not quite. Other elements of CC may be outputs: for f(x)=(x2+2)โ€Šmodโ€Š8f(x) = (x^2 + 2) \bmod 8 on R8R_8, C={1,2,3}C = \{1, 2, 3\} contains the non-output 11, yet fโˆ’1(C)={0,1,3,4,5,7}f^{-1}(C) = \{0, 1, 3, 4, 5, 7\}.
โŒ Not quite. The element cc cannot come back: anything in f(fโˆ’1(C))f(f^{-1}(C)) has the form f(x)f(x).
โŒ Not quite. This concerns the other round trip, and it fails for A=SA = S: fโˆ’1(f(S))=Sf^{-1}(f(S)) = S for every function.
Show solution

If yโˆˆf(fโˆ’1(C))y \in f(f^{-1}(C)), then y=f(x)y = f(x) for some xโˆˆfโˆ’1(C)x \in f^{-1}(C), so yy is an output of ff. The element cc is not an output, so cโˆ‰f(fโˆ’1(C))c \notin f(f^{-1}(C)) while cโˆˆCc \in C. Hence f(fโˆ’1(C))โ‰ Cf(f^{-1}(C)) \ne C, and with f(fโˆ’1(C))โІCf(f^{-1}(C)) \subseteq C the inclusion is proper.

(a) fails whenever CC also contains an output; (d) fails for A=SA = S, since every input lies in fโˆ’1(f(S))f^{-1}(f(S)) and that set is a subset of SS.

Solved: 0 / 4