Advanced Lecture Notes on Functions Part

11: Advanced Composition & Graphical Transformations

For the Serious Aspirant

Contents

1 Recap: The Functional Menagerie 1

2 Function Composition: An In-Depth Perspective 1

2.1 Domain and Range of Composite Functions: Nuances . . . . . . . . . . . . . . . . . 1
2.2 Iterated Functions and Discrete Dynamical Systems . . . . . . . . . . . . . . . . . . 1
2.2.1 Fixed Points and Their Stability . . . . . . . . . . . . . . . . . . . . . . . . . 1
2.2.2 Cobweb Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

3 Graphical Transformations: A Rigorous Coordinate Approach 2

3.1 Standard Transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
3.2 Order of Transformations: Non-Commutativity . . . . . . . . . . . . . . . . . . . . . 2
3.3 Transformations Involving GIF, Fractional Part . . . . . . . . . . . . . . . . . . . . . 3

4 Functional Conjugacy: Transferring Dynamics 3

5 Problems for the Adept: Part 11 4

6 Answer Key (Part 11) 5

Part 11: Composition & Transformations Advanced Functions by Yuki

1 Recap: The Functional Menagerie

Thus far, we’ve built a robust understanding of functions from their set-theoretic foundations to their
specific properties like injectivity, surjectivity, and bijectivity. We’ve also dissected the behavior of
algebraic (polynomial, rational, irrational) and transcendental (exponential, logarithmic, trigonometric,
inverse trigonometric) functions. This part synthesizes these ideas, focusing on the powerful concepts
of function composition and graphical transformations, pushing into more advanced territory.

2 Function Composition: An In-Depth Perspective

Recall the definition: If f : A → B and g : B ′ → C with Ran(f ) ⊆ B ′ , then g ◦ f : A → C is
(g ◦ f )(x) = g(f (x)). The domain Dg◦f = {x ∈ Df | f (x) ∈ Dg }.

2.1 Domain and Range of Composite Functions: Nuances

While Ran(g ◦ f ) = g(Ran(f ) ∩ Dg ), a common point of confusion is when Ran(f ) is not entirely
contained in Dg .

Example 2.1. Domain of Composition

√
Let f (x) = x with Df = [0, ∞), √ so Ran(f ) = [0, ∞). Let g(x) = ln(x−2) with Dg = (2, ∞).
For (g ◦ f )(x) = g(f (x)) = ln( x − 2):

1. x ∈ Df =⇒ x ≥ 0.
√ √
2. f (x) ∈ Dg =⇒ x ∈ (2, ∞) =⇒ x > 2 =⇒ x > 4.

So, Dg◦f = (4, ∞). Here, Ran(f ) = [0, ∞) is not a subset of Dg = (2, ∞). We only consider
x for which f (x) falls into Dg .

2.2 Iterated Functions and Discrete Dynamical Systems

When f : X → X, we define f 1 = f , f 2 = f ◦ f , f n+1 = f ◦ f n . The sequence x0 , x1 = f (x0 ), x2 =
f (x1 ), . . . is the orbit of x0 .

2.2.1 Fixed Points and Their Stability

A point x∗ is a fixed point if f (x∗ ) = x∗ . If f is differentiable, the stability of a fixed point x∗ can
often be determined by |f ′ (x∗ )|:

• If |f ′ (x∗ )| < 1, x∗ is an attracting fixed point (orbits starting near x∗ converge to x∗ ).

• If |f ′ (x∗ )| > 1, x∗ is a repelling fixed point (orbits starting near x∗ move away).

• If |f ′ (x∗ )| = 1, the test is inconclusive (can be attracting, repelling, or neutral).

1
Part 11: Composition & Transformations Advanced Functions by Yuki

Example 2.2. Logistic Map Iteration (Conceptual)

Consider f (x) = rx(1−x) for x ∈ [0, 1] and r ∈ [0, 4]. Fixed points: x = rx(1−x) =⇒ x = 0
or 1 = r(1 − x) =⇒ 1 = r − rx =⇒ rx = r − 1 =⇒ x = (r − 1)/r. For x∗ = (r − 1)/r
to be in [0, 1], we need r ≥ 1. f ′ (x) = r − 2rx. |f ′ (0)| = |r|. Attracting if |r| < 1.
|f ′ ((r − 1)/r)| = |r − 2r((r − 1)/r)| = |r − 2(r − 1)| = |r − 2r + 2| = |2 − r|. So x∗ = (r − 1)/r
is attracting if |2 − r| < 1 =⇒ −1 < 2 − r < 1 =⇒ −3 < −r < −1 =⇒ 1 < r < 3. As r
increases beyond 3, this fixed point becomes repelling, and period-doubling bifurcations occur,
leading to chaotic behavior. This is a cornerstone of chaos theory.

2.2.2 Cobweb Plots

Visualizing xn+1 = f (xn ) by plotting y = f (x) and y = x.
y
y=x

y = f (x)
x2

x1
x
0 x0

3 Graphical Transformations: A Rigorous Coordinate Approach

We revisit transformations from a coordinate mapping perspective to understand their algebra. A
point (xP , yP ) is on the graph of y = f (x) if yP = f (xP ). A point (xN , yN ) is on the graph of the
transformed function yN = G(xN ). We seek the relationship between (xP , yP ) and (xN , yN ).

3.1 Standard Transformations

1. yN = f (xN ) + c: Here xN = xP , yN = yP + c. (xP , yP ) → (xP , yP + c). Vertical shift.

2. yN = f (xN − c): Let xP = xN − c =⇒ xN = xP + c. yN = f (xP ) = yP . (xP , yP ) →

(xP + c, yP ). Horizontal shift right by c.

3. yN = kf (xN ): xN = xP , yN = kyP . (xP , yP ) → (xP , kyP ). Vertical stretch/compression.

4. yN = f (kxN ): Let xP = kxN =⇒ xN = xP /k. yN = f (xP ) = yP . (xP , yP ) → (xP /k, yP ).

Horizontal stretch/compression.

3.2 Order of Transformations: Non-Commutativity

The order in which transformations are applied matters. Consider f (x) → af (bx + c) + d. This can
be seen as a sequence: 1. y1 = f (x) 2. y2 = f (x + c/b) (Horizontal shift by −c/b if we factor b:
f (b(x + c/b)) ) - this step is usually easier if b is factored out. Or y2 = f (bx) (Horizontal compression
by b) 3. y3 = f (b(x + c/b)) if factoring, or f (bx + c) by replacing x with x + c/b in f (bx). (This is
x → bx → bx + c). This can be tricky. A reliable way is to analyze the argument: To get f (bx + c)
from f (x): replace x with bx + c. This means xP = bxN + c. So xN = (xP − c)/b = xP /b − c/b. So
from (xP , yP ): 1. f (x) → f (bx): (xP , yP ) → (xP /b, yP ). 2. f (bx) → f (b(x − c/b)) = f (bx − c):
Here we replace x with x − c/b. So (xP /b, yP ) → ((xP /b) + c/b, yP ). This is not what we want.

2
Part 11: Composition & Transformations Advanced Functions by Yuki

Correct sequence for y = af (b(x + c′ )) + d (where c′ = c/b): Original point on y = f (x)

is (x0 , f (x0 )). New point is (x, y). x → x + c′ (shift left by c′ ): argument becomes x0 + c′ .
x + c′ → b(x + c′ ) (horizontal compression by b about x = −c′ ): argument becomes b(x0 + c′ ).
xnew = x0 /b − c′ . f → af (vertical stretch by a): ynew = af (b(xnew + c′ )). af → af + d (vertical
shift by d). So, xtransf ormed = xoriginal /b − c/b. ytransf ormed = ayoriginal + d.

Example 3.1. Order of Modulus and Shift

onsider f (x) = x − 2. 1. y = |f (x)| = |x − 2|. 2. y = f (|x|) = |x| − 2. These are different.

The order matters.

3.3 Transformations Involving GIF, Fractional Part

Example 3.2. Graph of y = [sin x] vs y = sin[x]

(Refer to Part 11 Problem 7.7 for detailed graphs and discussion from previous turn if desired).
y = [sin x]: Range is {−1, 0, 1}. Discontinuities when sin x is an integer. Period 2π. y = sin[x]:
Range is {sin k | k ∈ Z}. Graph is piecewise constant over [k, k+1). Not periodic in the standard
sense due to varying values of sin k.

4 Functional Conjugacy: Transferring Dynamics

This is a more advanced concept that highlights structural similarities between functions.

Definition 4.1. Functional Conjugacy

Two functions f : X → X and g : Y → Y are conjugate if there exists a bijection (often a

homeomorphism in topological contexts) h : X → Y such that g ◦ h = h ◦ f , or equivalently,
g = h ◦ f ◦ h−1 . The function h is called the conjugacy.

If f and g are conjugate:

• g n = (h ◦ f ◦ h−1 )n = h ◦ f n ◦ h−1 . Iterates of g are related to iterates of f .

• If x0 is a fixed point of f (f (x0 ) = x0 ), then y0 = h(x0 ) is a fixed point of g: g(y0 ) = g(h(x0 )) =

(h ◦ f )(x0 ) = h(f (x0 )) = h(x0 ) = y0 .

• If x0 is a periodic point of f with period k, then h(x0 ) is a periodic point of g with period k.

Example 4.1. Conjugacy with Logistic Map

The logistic map f (x) = rx(1 − x) for certain r values (e.g., r = 4) is conjugate to the tent
map or bit-shift map via a transformation like h(x) = sin2 (πx/2). Analyzing the simpler map
can reveal properties (like chaos) in the logistic map. For r = 4, f (x) = 4x(1 − x). Let
x = sin2 θ. f (sin2 θ) = 4 sin2 θ(1 − sin2 θ) = 4 sin2 θ cos2 θ = (2 sin θ cos θ)2 = sin2 (2θ). If
we let h(θ) = sin2 θ, and T (θ) = 2θ (mod π) (angle doubling map), then f (h(θ)) = h(T (θ)).
So f ◦ h = h ◦ T , or f = h ◦ T ◦ h−1 (if h is invertible on suitable domain). This shows f is
conjugate to an angle-doubling map.

3
Part 11: Composition & Transformations Advanced Functions by Yuki

5 Problems for the Adept: Part 11

Problem 5.1. prob:P11.1
Ä ä
Let f (x) = 1+x2x
2 and g(x) = ln
1+x
1−x
. Find the domain and range of g ◦ f . Is g ◦ f injective
or surjective on its domain to R?

Problem 5.2. prob:P11.2

If f (x + 2) = 1+f (x)
1−f (x)
for all x ∈ R, show that f (x) is periodic. What is its period? (Hint:
Calculate f (x + 4), f (x + 8)).

Problem 5.3. prob:P11.3

Sketch the graph of the region defined by |y| ≤ sin(|x|).

Problem 5.4. prob:P11.4

1
Let f (x) = 1−x . Find the number of real solutions to f 2025 (x) = cos(πx), where f n denotes
the n-th iteration of f . Carefully determine the domain of f 2025 (x).

Problem 5.5. prob:P11.5

Let f (x) = |x2 − 4|x| + 3|. Analyze its symmetry. Find the number of points where f (x) is not
differentiable. Sketch the graph of y = f (x) and y = [f (x)].

Problem 5.6. prob:P11.6

Consider f (x) = min(sin x, cos x) and g(x) = {f (x)}, where {.} is the fractional part function.
Determine the fundamental period of g(x). Sketch g(x) over one period.

Problem 5.7. prob:P11.7

If f (x) is a function such that f (x + T ) = k · f (x) for some constants k > 0, T > 0. If
g(x) = logb (f (x)) (for a suitable base b), show that g(x) is periodic if k = 1. What if k ̸= 1?
If f (x + T ) = f (x) + k, what can be said about g(x) = af (x) ?

4
Part 11: Composition & Transformations Advanced Functions by Yuki

6 Answer Key (Part 11)

Solutions (Part 11)
Ä ä
1. Problem 11.1: f (x) = 1+x 2x
2 , g(x) = ln
1+x
1−x
. Domain of f (x) is R. Range of f (x):
Let y = 1+x2 =⇒ y(1 + x ) = 2x =⇒ yx2 − 2x + y = 0. For x to be real,
2x 2

D ≥ 0 =⇒ (−2)2 − 4(y)(y) ≥ 0 =⇒ 4 − 4y 2 ≥ 0 =⇒ y 2 ≤ 1 =⇒ −1 ≤ y ≤ 1. So
Ran(f ) = [−1, 1]. Domain of g(x): We need 1−x 1+x
> 0. Critical points at x = −1, x = 1.
Sign analysis yields Dg = (−1, 1). For g(f (x)): We need f (x) ∈ Dg , so −1 < f (x) < 1.
We know Ran(f ) = [−1, 1]. So we need to exclude x for which f (x) = 1 or f (x) = −1.
f (x) = 1 =⇒ 1+x 2x
2 = 1 =⇒ 2x = 1 + x
2
=⇒ x2 − 2x + 1 = 0 =⇒ (x − 1)2 =
0 =⇒ x = 1. f (x) = −1 =⇒ 1+x 2x
2 = −1 =⇒ 2x = −1 − x
2
=⇒ x2 + 2x + 1 =
Ä ä
0 =⇒ (x + 1)2 = 0 =⇒ x = −1. So Dg◦f = R \ {−1, 1}. (g ◦ f )(x) = g 1+x 2x
=
Å ã Ç 2 å  
2

1+ 2x 1+x +2x Ä 2
ä Ä ä2
= ln (1+x) . Range of g ◦ f :
2 1+x2 1+x
ln 1− 1+x
2x = ln 1+x 2 −2x (1−x)2
= ln 1−x = 2 ln 1+x
1−x
1+x2 1+x2

Let u = As x varies in R \ {−1, 1}, u varies in R \ {0, −1} (as x → 1, u → ∞;

1+x
1−x
.
as x → −1, u → 0; as x → ∞, u → −1). So |u| varies in (0, ∞). Then ln |u| varies
in (−∞, ∞). So 2 ln |u| varies in R. Range is R. g ◦ f is surjective onto R. Is g ◦ f
injective? Let h(x) = 2 ln 1+x
1−x
. If x1 = 2, h(2) = 2 ln |3/ − 1| = 2 ln 3. If x2 = 1/2,
1+1/2
1−1/2
= 3/2
1/2
= 3. h(1/2) = 2 ln 3. Since h(2) = h(1/2) but 2 ̸= 1/2, g ◦ f is not
injective.
1+f (x)
1+f (x) 1+f (x+2) 1+ 1−f (x)
2. Problem 11.2: f (x + 2) = 1−f (x)
. f (x + 4) = f ((x + 2) + 2) = 1−f (x+2)
= 1+f (x) =
1− 1−f (x)

= −2f2(x) = − f (x)
(1−f (x))+(1+f (x))
(1−f (x))−(1+f (x))
1
. f (x + 8) = f ((x + 4) + 4) = − f (x+4)
1
= − −1/f1 (x) =
f (x). So f (x + 8) = f (x). The function is periodic with period 8.

3. Problem 11.3: |y| ≤ sin(|x|). 1. Start with y1 = sin x. 2. y2 = sin(|x|): For x ≥ 0,

y2 = sin x. For x < 0, y2 = sin(−x) = − sin x. This graph is symmetric about y-axis for
x ≥ 0 part, actually y2 is an even function. Keep sin x for x ≥ 0, reflect this part onto
x < 0. So it looks like sin x for x ∈ [0, π], then reflected, giving lobes. 3. |y| ≤ y2 : This
means −y2 ≤ y ≤ y2 . First, this requires yS 2 = sin(|x|) ≥ 0. sin(|x|) ≥ 0 when |x| ∈
[2nπ, (2n + 1)π] for n = 0, 1, 2, . . . . So x ∈ ∞n=0 [−(2n + 1)π, −2nπ] ∪ [2nπ, (2n + 1)π].
In these intervals, the region is bounded by y = sin(|x|) and y = − sin(|x|). The graph
will be a series of ”lens” shapes symmetric about the x-axis, located where sin(|x|) ≥ 0.
1 1
4. Problem 11.4 (Yuki’s Challenge): f1 (x) = 1−x
. f2 (x) = f1 (f1 (x)) = 1 =
1 1−x x−1

1− 1−x

1−x−1 = −x
= x
. f3 (x) = f1 (f2 (x)) = f1 x−1
x
= 1
1− x−1
= 1
x−(x−1)
x
= 1 = x.
1−x x x
So f3 (x) = x. This means f4 (x) = f1 (x), f5 (x) = f2 (x), etc. The functions repeat
with period 3. We need f 2025 (x). 2025 (mod 3) = (2 + 0 + 2 + 5) (mod 3) = 9
(mod 3) = 0. So f 2025 (x) = f 3 (x) = x. Equation is x = cos(πx). Domain: For f1 (x),
x ̸= 1. For f2 (x), x ̸= 0 (and f1 (x) must be defined, so x ̸= 1). For f3 (x), x ̸= 0, 1.
The overall domain for f 2025 (x) = x to be well-defined through iteration is R \ {0, 1}.
Solutions to x = cos(πx): Let h(x) = x − cos(πx). h′ (x) = 1 − (−π sin(πx)) =
1 + π sin(πx). If x = 0, 0 = cos(0) = 1, False. If x = 1, 1 = cos(π) = −1,
False. If x = −1, −1 = cos(−π) = −1. So x = −1 is a solution. Graph y = x
and y = cos(πx). They intersect at x = −1. For x > 0, cos(πx) oscillates. For
x ∈ (0, 1/2), cos(πx) ∈ (0, 1). x ∈ (0, 1/2). Possible intersection. Let x0 ∈ (0, 1/2).

5
Part 11: Composition & Transformations Advanced Functions by Yuki

h(0) = −1 < 0. h(1/2) = 1/2 − cos(π/2) = 1/2 > 0. So one root in (0, 1/2). For
x ∈ (1/2, 1), cos(πx) ∈ (−1, 0). x ∈ (1/2, 1). x > cos(πx). No root. For x ∈ (1, 3/2),
cos(πx) ∈ (−1, 0). x > cos(πx). No root. For x ∈ (−1/2, 0), cos(πx) ∈ (0, 1).
x ∈ (−1/2, 0). x < cos(πx). h(0) = −1. h(−1/2) = −1/2 − cos(−π/2) = −1/2 < 0.
At x = −1, h(−1) = −1 − cos(−π) = −1 − (−1) = 0. Are there other negative
solutions? If x < −1, y = x is below y = −1. y = cos(πx) oscillates between −1 and
1. They can only meet if x = −1 and cos(πx) = −1. Solutions are x = −1 and one
solution x0 ∈ (0, 1/2). Total 2 solutions.

5. Problem 11.5: f (x) = |x2 − 4|x| + 3|. Let g(t) = t2 − 4t + 3 = (t − 1)(t − 3) where
t = |x| ≥ 0. So f (x) = ||x|2 − 4|x| + 3| = |(|x| − 1)(|x| − 3)|. Symmetry: Since
f (x) depends only on |x|, f (−x) = || − x|2 − 4| − x| + 3| = ||x|2 − 4|x| + 3| = f (x).
So f (x) is an even function, symmetric about y-axis. Non-differentiability: Points where
argument of outer modulus is zero: (|x| − 1)(|x| − 3) = 0 =⇒ |x| = 1 or |x| = 3. So
x = ±1, ±3. Points where argument of inner modulus |x| causes issues: x = 0. At x = 0:
f (x) = |0 − 0 + 3| = 3. f ′ (x) near 0: t = |x|. f (x) = |(t − 1)(t − 3)|. For x → 0+ ,
t → 0+ . f (x) = |(x−1)(x−3)| = |x2 −4x+3| = x2 −4x+3. f ′ (0+ ) = −4. For x → 0− ,
t → 0+ . f (x) = |(−x − 1)(−x − 3)| = |(x + 1)(x + 3)| = |x2 + 4x + 3| = x2 + 4x + 3.
f ′ (0− ) = 4. Not differentiable at x = 0. At |x| = 1: e.g. x = 1. f (x) = 0. For x → 1+ ,
|x| = x. f (x) = |(x−1)(x−3)|. Near x = 1, x−1 > 0, x−3 < 0. So (x−1)(x−3) < 0.
f (x) = −(x − 1)(x − 3) = −(x2 − 4x + 3) = −x2 + 4x − 3. f ′ (x) = −2x + 4. f ′ (1+ ) = 2.
For x → 1− , |x| = x. f (x) = |(x − 1)(x − 3)|. Near x = 1− , x − 1 < 0, x − 3 < 0. So
(x − 1)(x − 3) > 0. f (x) = (x − 1)(x − 3) = x2 − 4x + 3. f ′ (x) = 2x − 4. f ′ (1− ) = −2.
Not differentiable at x = 1. By symmetry, also at x = −1. At |x| = 3: e.g. x = 3.
f (x) = 0. For x → 3+ , |x| = x. f (x) = |(x − 1)(x − 3)| = (x − 1)(x − 3) since
x − 1 > 0, x − 3 > 0. f (x) = x2 − 4x + 3. f ′ (x) = 2x − 4. f ′ (3+ ) = 2. For x → 3− ,
|x| = x. f (x) = |(x − 1)(x − 3)| = −(x − 1)(x − 3) since x − 1 > 0, x − 3 < 0.
f (x) = −x2 + 4x − 3. f ′ (x) = −2x + 4. f ′ (3− ) = −2. Not differentiable at x = 3.
By symmetry, also at x = −3. Total points of non-differentiability: 0, ±1, ±3 (5 points).
Graph y = f (x): Plot y = (t − 1)(t − 3) for t ≥ 0. Then take |(t − 1)(t − 3)|. Then
use t = |x|. y = [f (x)]: This will be a step function based on f (x). f (0) = 3, f (±1) =
0, f (±2) = |4 − 8 + 3| = | − 1| = 1, f (±3) = 0. Max between 1, 3 for t is at t = 2,
g(2) = −1, |g(2)| = 1.

6. Problem 11.6: f (x) = min(sin x, cos x).√This function has period √ 2π. Graph is known.
g(x) = {f (x)}. Range of f (x) is [−1, 1/ 2]. If f (x) ∈ [0, 1/ 2], then {f (x)} = f (x).
If f (x) ∈ [−1, 0), then f (x) is like −0
® . . . . [f (x)] = −1. {f (x)} = f (x) − [f (x)] =
f (x) if f (x) ≥ 0
f (x) − (−1) = f (x) + 1. So g(x) = . The period of f (x) is 2π.
f (x) + 1 if f (x) < 0
The transformation to {f (x)} will also result in a period of 2π.

7. Problem 11.7: 1. f (x+T ) = k ·f (x). g(x) = logb (f (x)). g(x+T ) = logb (f (x+T )) =
logb (k · f (x)) = logb k + logb (f (x)) = logb k + g(x). This is periodic if logb k = 0, which
means k = 1. Period is T . If k ̸= 1, g(x) is not periodic but satisfies g(x+T ) = g(x)+C.
Such functions are sometimes called ”arithmetically periodic” or ”functions with a linear
trend.” 2. f (x+T ) = f (x)+k. g(x) = af (x) . g(x+T ) = af (x+T ) = af (x)+k = af (x) ·ak =
g(x) · ak . This is periodic if ak = 1. This requires k = 0 (if a ̸= ±1) or a = 1 (trivial) or
a = −1 and k is even. If k = 0, f (x) is periodic with period T , then g(x) = af (x) is also
periodic with period T . If ak ̸= 1, it’s ”geometrically periodic.”