Advanced Lecture Notes on Functions Part

10: Functional Equations I Foundations and Cauchy’s Legacy

For the Serious Aspirant

Contents

1 Introduction to Functional Equations 1

2 Fundamental Substitution Techniques 1

3 Cauchy’s Functional Equations 2

3.1 Cauchy’s Additive Functional Equation . . . . . . . . . . . . . . . . . . . . . . . . . 2
3.2 Other Cauchy Functional Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.3 Jensen’s Functional Equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

4 Problems for the Adept: Part 10-I 5

5 Answer Key (Part 10-I) 6

Part 10: Functional Equations I Advanced Functions by Yuki

1 Introduction to Functional Equations

A functional equation is an equation where the unknown is a function, rather than a variable. The
goal is typically to find all functions f : D → C (where D is the domain and C is the codomain) that
satisfy the given equation for all values of the variables in their specified domains. Solving functional
equations often involves a blend of strategic substitutions, algebraic manipulation, and leveraging known
properties of functions (like injectivity, surjectivity, parity, periodicity, continuity, differentiability).
As Evan Chen aptly puts it in his handout ”Introduction to Functional Equations”, a common
approach involves:

1. Identifying candidate solutions by testing simple, standard functions.

2. Rigorously proving that these candidate solutions are indeed the *only* solutions that satisfy the
equation.

3. Verifying that the found solutions work by substituting them back into the original equation.
(The ”if” part, often trivial but essential for completeness).

This part lays the groundwork, focusing on fundamental techniques and the cornerstone equations
known as Cauchy’s functional equations.

2 Fundamental Substitution Techniques

The first line of attack for most functional equations is strategic substitution of variables. The aim is
to simplify the equation, find particular values of the function (e.g., f (0), f (1)), or reveal underlying
properties. Common substitutions include:

• Setting variables to specific constants: x = 0, y = 0, x = 1, y = 1, x = −1, etc. This is often

denoted as P (0, 0), P (x, 0), P (1, y), etc.

• Setting variables equal: y = x.

• Setting variables to expressions involving other variables or function values: y = −x, y =

1/x, y = f (x), etc.

• Swapping variables if the equation has some symmetry: If P (x, y) is the assertion, compare
P (x, y) with P (y, x).

1
Part 10: Functional Equations I Advanced Functions by Yuki

Example 2.1. Finding f (0) and Involutions

(Inspired by Evan Chen, Example 2.1) Find all functions f : R → R such that f (f (x)2 +f (y)) =
xf (x) + y for all x, y ∈ R.

Proof
Denote the assertion as P (x, y). P (0, y) =⇒ f (f (0)2 + f (y)) = 0 · f (0) + y = y.
This shows that f is surjective, because for any z ∈ R, we can choose y = z. Then
f (f (0)2 +f (y)) = z, meaning z is in the range of f . Since f is surjective, there exists some
u such that f (u) = 0. P (u, y) =⇒ f (f (u)2 + f (y)) = uf (u) + y =⇒ f (02 + f (y)) =
u · 0 + y =⇒ f (f (y)) = y for all y ∈ R. This means f is an involution. Since
f (f (y)) = y, f is bijective (it is its own inverse). Since f is surjective, there exists c such
that f (c) = 0. P (x, c) =⇒ f (f (x)2 + f (c)) = xf (x) + c =⇒ f (f (x)2 ) = xf (x) + c.
Since f (f (y)) = y, we have f (0) = c′ for some c′ . Then f (c′ ) = 0. So f (0) exists.
Since f (f (y)) = y, and f (u) = 0, then f (0) = f (f (u)) = u. So u = f (0). Thus,
f (f (0)) = 0. Set x = f (0) in f (f (x)2 ) = xf (x) + c. f (f (f (0))2 ) = f (0)f (f (0)) + c.
f (02 ) = f (0) · 0 + c =⇒ f (0) = c. So, f (f (x)2 ) = xf (x) + f (0). Since f (f (y)) = y,
replace x with f (t): f (f (f (t))2 ) = f (t)f (f (t)) + f (0) =⇒ f (t2 ) = f (t) · t + f (0). So
f (t2 ) = tf (t) + f (0) for all t ∈ R. If t = 0, f (0) = 0 · f (0) + f (0), which is f (0) = f (0).
If f (0) = 0, then f (t2 ) = tf (t). If f (0) = 0, then f (f (y)) = y =⇒ f (0) = 0. This is
consistent. The original equation: f (f (x)2 + f (y)) = xf (x) + y. Since f (t2 ) = tf (t),
we have tf (t) √ = (f (t))2 for t ≥ 0 is not necessarily
√
2
√ true. However, from 2f (t ) = tf2(t), if
t ≥ 0, let t = u for u ≥ 0. Then f (u) = uf ( u). If f (x) = x: f (x + y) = x + y.
xf (x) + y = x · x + y = x2 + y. This works. If f (x) = −x: f ((−x)2 − y) = f (x2 − y) =
−(x2 − y) = y − x2 . xf (x) + y = x(−x) + y = −x2 + y. This works. The problem states
that f (f (y)) = y implies f is injective. From f (t2 ) = tf (t) + f (0): If f (0) = 0, then
f (t2 ) = tf (t). Since f (f (y)) = y, f is odd if f (0) = 0. f (−t2 ) = −f (t2 ) = −tf (t).
Also f ((−t)2 ) = f (t2 ) = −tf (t)? If f (t2 ) = tf (t): Let t ≥ 0. f (t) = ±t. If f (x) = x,
x2 = x · x. If f (x) = −x, −(t2 ) = t(−t) = −t2 . Solutions are f (x) = x and f (x) = −x.
(Verification required, as in Evan Chen’s notes).

3 Cauchy’s Functional Equations

Augustin-Louis Cauchy (1789-1857) studied four fundamental functional equations that now bear his
name. Their solutions, especially over R with additional conditions, are foundational.

3.1 Cauchy’s Additive Functional Equation

The equation is f (x + y) = f (x) + f (y) for all x, y in the domain.

2
Part 10: Functional Equations I Advanced Functions by Yuki

Theorem 3.1. Solutions to Additive Cauchy Equation

Let f satisfy f (x + y) = f (x) + f (y).

1. If the domain is Q, then f (x) = cx for some constant c = f (1).

2. If the domain is R and any one of the following conditions holds:

• f is continuous at least at one point.

• f is monotonic on any interval.
• f is bounded (above or below) on any interval.

then f (x) = cx for some constant c = f (1).

Proof
Solution over Q Let P (x, y) be the assertion f (x + y) = f (x) + f (y).

1. P (x, 0) =⇒ f (x + 0) = f (x) + f (0) =⇒ f (x) = f (x) + f (0) =⇒ f (0) = 0.

2. For n ∈ N: P (x, x) =⇒ f (2x) = f (x) + f (x) = 2f (x). By induction, assume

f (kx) = kf (x) for some k ∈ N. f ((k + 1)x) = f (kx + x) = f (kx) + f (x) = kf (x) +
f (x) = (k + 1)f (x). So f (nx) = nf (x) for all n ∈ N. Let x = 1, then f (n) = nf (1).
Let c = f (1). So f (n) = cn for n ∈ N.

3. For n ∈ Z− : Let n = −m where m ∈ N. f (0) = 0 =⇒ f (n + (−n)) = 0 =⇒

f (n) + f (−n) = 0 =⇒ f (n) = −f (−n). f (n) = −f (m) = −(cm) = c(−m) = cn. So
f (n) = cn for all n ∈ Z (including f (0) = c · 0 = 0).

4. For q ∈ Q: Let q = p/r where p ∈ Z, r ∈ N. f (r · pr ) = rf ( pr ) =⇒ f (p) = rf ( pr ).

We know f (p) = cp. So cp = rf ( pr ). f ( pr ) = c pr . Thus, f (x) = cx for all x ∈ Q, where
c = f (1).

Proof
Solution over R with Continuity Assume f (x+y) = f (x)+f (y) for x, y ∈ R, and f is continuous
at some point x0 . Step 1: Continuity at 0 =⇒ Continuity everywhere. If f is continuous
at x0 , then limh→0 f (x0 + h) = f (x0 ). f (x0 + h) = f (x0 ) + f (h). So limh→0 (f (x0 ) + f (h)) =
f (x0 ) =⇒ f (x0 ) + limh→0 f (h) = f (x0 ). This implies limh→0 f (h) = 0. Since f (0) = 0, this
means f is continuous at 0. Now, for any a ∈ R: limh→0 f (a + h) = limh→0 (f (a) + f (h)) =
f (a) + limh→0 f (h) = f (a) + f (0) = f (a). So f is continuous everywhere.
Step 2: Extend from Q to R using continuity. We know f (q) = cq for all q ∈ Q, where
c = f (1). Let x ∈ R. There exists a sequence of rational numbers (qn ) such that qn → x
as n → ∞. Since f is continuous, f (x) = f (limn→∞ qn ) = limn→∞ f (qn ). Since qn ∈ Q,
f (qn ) = cqn . So f (x) = limn→∞ (cqn ) = c limn→∞ qn = cx. Thus, f (x) = cx for all x ∈ R.
(Similar proofs hold if monotonicity or boundedness on an interval is assumed, as these also
imply continuity for additive functions over R).

For the Curious

Hamel Bases and Discontinuous Solutions If no additional condition (like continuity, monotonic-
ity, or boundedness on an interval) is imposed on an additive function f : R → R, then there
exist infinitely many other solutions besides f (x) = cx. These solutions are highly pathological:

3
Part 10: Functional Equations I Advanced Functions by Yuki

their graphs are dense in R2 , meaning every open disk in the plane contains a point from the
graph. The existence of these solutions relies on the Axiom of Choice and the concept of a Hamel
basis for R as a vector space over Q. This is far beyond JEE scope but is a fascinating piece of
advanced mathematics. For JEE, assume that if a Cauchy-type equation appears, the problem
will either provide or imply one of the regularity conditions leading to the ”nice” solutions.

3.2 Other Cauchy Functional Equations

1. Exponential Cauchy: f (x + y) = f (x)f (y) for x, y ∈ R. If f (x0 ) = 0 for some x0 , then
f (x) = f ((x − x0 ) + x0 ) = f (x − x0 )f (x0 ) = 0. So f (x) ≡ 0. If f (x) ̸= 0 for any x, then
f (x) = f (x/2 + x/2) = (f (x/2))2 > 0. So f (x) is always positive. Let g(x) = ln f (x). Then
g(x+y) = ln(f (x+y)) = ln(f (x)f (y)) = ln f (x)+ln f (y) = g(x)+g(y). So g(x) is additive. If
f has a regularity condition (e.g., continuous), then g(x) = kx. f (x) = eg(x) = ekx = (ek )x = ax
for some a = ek > 0. Solutions: f (x) = 0 or f (x) = ax .

2. Logarithmic Cauchy: f (xy) = f (x) + f (y) for x, y ∈ R+ . Let x = eu , y = ev where

u, v ∈ R. Let g(t) = f (et ). f (eu ev ) = f (eu ) + f (ev ) =⇒ f (eu+v ) = f (eu ) + f (ev ).
g(u + v) = g(u) + g(v). So g(t) is additive. If f has a regularity condition (e.g., continuous),
g(t) = kt. f (et ) = kt =⇒ f (x) = k ln x = logb x for some base b = e1/k . Solutions: f (x) = 0
or f (x) = k ln x.

3. Multiplicative Cauchy: f (xy) = f (x)f (y) for x, y ∈ R+ . If f (x0 ) = 0 for some x0 > 0,
then f (x) = f (x/x0 · x0 ) = f (x/x0 )f (x0 ) = 0. So f (x) ≡ √ 0.√ If f (1) =√ 0, then f (x) =
f (x · 1) = f (x)f (1) = 0. If f (x) ̸= 0 for x > 0. f (x) = f ( x x) = (f ( x))2 > 0 (unless
f (x) = 1 is allowed from f (x)2 = 1 with negative values somewhere, but domain is R+ ). So
f (x) > 0. Let g(t) = ln f (et ). Then g(u + v) = ln f (eu+v ) = ln f (eu ev ) = ln(f (eu )f (ev )) =
ln f (eu ) + ln f (ev ) = g(u) + g(v). So g(t) = kt. ln f (et ) = kt =⇒ f (et ) = ekt = (et )k .
f (x) = xk . Solutions: f (x) = 0, f (x) = 1, or f (x) = xk .

3.3 Jensen’s Functional Equation

( x+y ) f (x)+f (y)
The equation is f 2
= 2
. This equation is closely related to affine functions and convexity.

Theorem 3.2. Solutions to Jensen’s Equation

If f : R → R is continuous and satisfies Jensen’s functional equation, then f (x) = ax + b for

some constants a, b.

Proof
( ) ( )
Let g(x) = f (x) − f (0). Then g(0) = f (0) − f (0) = 0. g x+y = f x+y − f (0) =
2 ( x+y ) g(x)+g(y)
2
f (x)+f (y)
− f (0). g(x)+g(y)
= f (x)−f (0)+f (y)−f (0)
= f (x)+f (y)
− f (0). So g 2 = . Set
2 2 2 2 ( u+v ) 2( )
y = 0: g(x/2) = g(x)+g(0)
2
= g(x)
2
. So g(2t) = 2g(t). Now, g(u + v) = g 2 · 2
= 2g u+v
2
=
g(u)+g(v)
2 2
= g(u) + g(v). So g(x) satisfies Cauchy’s additive equation. Since f is continuous,
g is continuous. Thus g(x) = ax for some a. f (x) − f (0) = ax =⇒ f (x) = ax + f (0). Let
f (0) = b. So f (x) = ax + b.

4
Part 10: Functional Equations I Advanced Functions by Yuki

4 Problems for the Adept: Part 10-I

Problem 4.1. prob:P10.1

Find all functions f : R → R such that f (x + y) − f (x − y) = 2f (y) for all x, y ∈ R, assuming

f is continuous. (Hint: This is related to D’Alembert’s equation or can be reduced).

Problem 4.2. prob:P10.2

Find all continuous functions f : R → R such that f (xy) = xf (y) + yf (x) for all x, y ∈ R.

Problem 4.3. prob:P10.3

(USAMO 2002, Evan Chen Ex 1.1) Find all functions f : R → R such that f (x2 − y 2 ) =
xf (x) − yf (y) for all x, y ∈ R.

Problem 4.4. prob:P10.4

Find all functions f : Q → Q such that f (1) = 2 and f (xy) = f (x)f (y) − f (x + y) + 1 for all
x, y ∈ Q. (Hint: Try to relate to f (x) = x + 1. Consider g(x) = f (x) − 1).

Problem 4.5. prob:P10.5

Let f : R+ → R satisfy f (x) + f (y) = f (x)f (y) + 1 − xy

1
for x, y ∈ R+ . If f (1) = 2, find f (x).

5
Part 10: Functional Equations I Advanced Functions by Yuki

5 Answer Key (Part 10-I)

Solutions (Part 10-I)

1. Problem 10.1: f (x+y)−f (x−y) = 2f (y) (1) Let P (x, y) be the assertion. P (0, y) =⇒
f (y) − f (−y) = 2f (y) =⇒ −f (y) = f (−y). So f is an odd function. Thus f (0) = 0.
P (x, x) =⇒ f (2x) − f (0) = 2f (x) =⇒ f (2x) = 2f (x) since f (0) = 0. This
implies f (nx) = nf (x) for n ∈ N by induction (if we assume f (kx) = kf (x), then
f (2kx) = 2f (kx) = 2kf (x) and f ((k + 1)x) can be tricky). However, f (2x) = 2f (x)
and f odd means f (nx) = nf (x) for n ∈ Z. Also, for x ∈ Q, f (x) = cx. (This line of
reasoning for f (nx) = nf (x) from f (2x) = 2f (x) is subtle for additive-like equations if
not directly Cauchy).
Let x = y: f (2y) − f (0) = 2f (y) =⇒ f (2y) = 2f (y) as f (0) = 0. Let y = x:
f (2x) − f (0) = 2f (x) =⇒ f (2x) = 2f (x). In (1), swap x, y: f (y + x) − f (y − x) =
2f (x). Since f is odd, f (y − x) = −f (x − y). f (x + y) + f (x − y) = 2f (x) (2).
Adding (1) and (2): 2f (x + y) = 2f (x) + 2f (y) =⇒ f (x + y) = f (x) + f (y).
This is Cauchy’s additive equation. Since f is continuous, f (x) = cx. Substitute back:
c(x + y) − c(x − y) = 2cy =⇒ cx + cy − cx + cy = 2cy =⇒ 2cy = 2cy. This holds
for any c. So f (x) = cx for any constant c.

2. Problem 10.2: f (xy) = xf (y) + yf (x). P (1, x) =⇒ f (x) = 1 · f (x) + xf (1) =⇒

xf (1) = 0. Since this must hold for all x, we must have f (1) = 0. P (x, 1) =⇒ f (x) =
xf (1) + 1f (x) =⇒ xf (1) = 0, same. Let g(x) = f (x)/x for x ̸= 0. (Assume f (0) = 0
by P (x, 0) =⇒ f (0) = xf (0) + 0 · f (x) =⇒ f (0)(1 − x) = 0 =⇒ f (0) = 0).
Then f (x) = xg(x). xyg(xy) = x(yg(y)) + y(xg(x)) = xyg(y) + xyg(x). For x, y ̸= 0,
g(xy) = g(x) + g(y). This is logarithmic Cauchy for g. If x > 0, y > 0, and f continuous
(implies g continuous on (0, ∞)), then g(x) = c ln x. So f (x) = cx ln x for x > 0. For
x < 0, let x = −a, a > 0. g(−a·y) = g(−a)+g(y) for y > 0. This definition domain for g
needs care. If f (x) = cx ln |x| for x ̸= 0, and f (0) = 0. LHS: c(xy) ln |xy| = cxy(ln |x| +
ln |y|). RHS: x(cy ln |y|)+y(cx ln |x|) = cxy ln |y|+cxy ln |x| = cxy(ln |x|+ln |y|). This
works. So f (x) = cx ln |x| for x ̸= 0, f (0) = 0.

3. Problem 10.3: f (x2 − y 2 ) = xf (x) − yf (y). P (x, y) is the assertion. P (y, x) =⇒

f (y 2 − x2 ) = yf (y) − xf (x). So f (−(x2 − y 2 )) = −(xf (x) − yf (y)). This means
f (−t) = −f (t) if t can be written as x2 − y 2 . Any real number can be. So f is
odd. f (0) = 0. P (x, 0) =⇒ f (x2 ) = xf (x) − 0f (0) = xf (x). Since f is odd,
f ((−x)2 ) = (−x)f (−x) =⇒ f√ (x2 ) = (−x)(−f (x)) = xf (x). √Consistent. Using
f (t2 ) = tf (t) (for t ≥ 0, use x = t, y = 0. For t < 0, f (t) = tf ( t) is problematic).
Actually, f (x2 ) = xf (x) holds for all x. If x < 0, let x = √ −a, a√> 0. f ((−a)2 ) =
−af (−a) =⇒ f (a ) = −a(−f (a)) = af (a). So f (t) = tf ( t) for t ≥ 0. And
2

f (t) = −|t|1/2 f (|t|1/2 )? No. We have f (x2 ) = xf (x). Original equation: f (x2 − y 2 ) =
f (x2 ) − f (y 2 ). (This is P (x2 , y 2 ) if f (t) = ct). Let u = x2 , v = y 2 . For u, v ≥ 0.
f (u − v) = f (u) − f (v) for u, v ≥ 0. This is Cauchy on domain restriction. With f odd,
it extends to all reals. f (s−t) = f (s)−f (t). Let s = S, t = 0 =⇒ f (S) = f (S)−f (0).
Let s = 0, t = T =⇒ f (−T ) = f (0) − f (T ) =⇒ f (−T ) = −f (T ) (reconfirming odd).
Let S = s, T = −t. f (s − (−t)) = f (s) − f (−t) =⇒ f (s + t) = f (s) + f (t). So f is
additive. From f (x2 ) = xf (x) and f (x) = cx: cx2 = x(cx) = cx2 . This is consistent.
So f (x) = cx is the solution.

4. Problem 10.4 (Yuki’s Challenge): f (1) = 2, f (xy) = f (x)f (y) − f (x + y) + 1. Let

g(x) = f (x)−1. Then f (x) = g(x)+1. g(xy)+1 = (g(x)+1)(g(y)+1)−(g(x+y)+1)+1.

6
Part 10: Functional Equations I Advanced Functions by Yuki

g(xy) + 1 = g(x)g(y) + g(x) + g(y) + 1 − g(x + y) − 1 + 1. g(xy) = g(x)g(y) +

g(x) + g(y) − g(x + y). This is g(x + y) + g(xy) = g(x)g(y) + g(x) + g(y). We
know f (1) = 2 =⇒ g(1) = f (1) − 1 = 1. Consider P (x, 1) in the original equation:
f (x) = f (x)f (1) − f (x + 1) + 1. f (x) = 2f (x) − f (x + 1) + 1 =⇒ f (x + 1) = f (x) + 1.
By induction, f (x + n) = f (x) + n for n ∈ N. Let x = 0, f (n) = f (0) + n. Let
x = 1, f (n + 1) = f (1) + n = 2 + n. Also f (n + 1) = f (0) + n + 1. So 2 + n =
f (0) + n + 1 =⇒ f (0) = 1. Since f (x + 1) = f (x) + 1 and domain is Q: For x ∈ Q,
f (x) = x + f (0) = x + 1. Let’s check if f (x) = x + 1 is a solution. f (1) = 1 + 1 = 2.
LHS: f (xy) = xy + 1. RHS: f (x)f (y) − f (x + y) + 1 = (x + 1)(y + 1) − ((x + y) + 1) + 1
= xy + x + y + 1 − x − y − 1 + 1 = xy + 1. LHS=RHS. So f (x) = x + 1 is the solution
for x ∈ Q.

5. Problem 10.5: f (x) + f (y) = f (x)f (y) + 1 − xy 1

for x, y ∈ R+ . f (1) = 2. Rearrange:
f (x)f (y) − f (x) − f (y) + 1 = xy
1
. (f (x) − 1)(f (y) − 1) = xy1
. Let g(x) = f (x) − 1. Then
g(1) = f (1) − 1 = 1. The equation becomes g(x)g(y) = xy 1
. Let y = 1. g(x)g(1) =
1
x
=⇒ g(x) · 1 = x =⇒ g(x) = x . So f (x) − 1 = x1 =⇒ f (x) = 1 + x1 .
1 1

Check: f (1) = 1 + 1/1 = 2. Correct. LHS: 1 + x1 + 1 + y1 = 2 + x1 + y1 . RHS:

(1 + x1 )(1 + y1 ) + 1 − xy1
= 1 + x1 + y1 + xy 1
+ 1 − xy1
= 2 + x1 + y1 . LHS=RHS. So
f (x) = 1 + 1/x is the solution.