Power rule — proof for integer, rational exponents: Question bank

Definition Symbols and stages used on THIS page (build them before you meet them)

The exponent $n$ — the little number sitting up high in $x^{n}$ : the count of how many copies of $x$ get multiplied ( $x^{3} = x \cdot x \cdot x$ , so $n = 3$ ). It lives in the top-right of the power, it is constant while $x$ varies, and the entire page is about the single formula $\frac{d}{d x} x^{n} = n x^{n - 1}$ . As $n$ ranges over the number line — positive integer, $0$ , negative integer, fraction, irrational — we need a different tool per range, which is why $n$ gets sorted into stages below.
The increment $h$ — a tiny nudge added to $x$ . Every derivative here comes from the limit definition $f^{'} (x) = lim_{h \to 0} \frac{f ( x + h ) - f ( x )}{h}$ : we push $x$ to $x + h$ , measure the change, divide by the nudge, then shrink $h$ to $0$ . In Figure 1 the horizontal gap between the two coral dots is $h$ , the butter arrow is the change in height, and the coral dashed line is their ratio (the secant slope) that becomes the true slope as $h$ shrinks.
The integer $m$ — in the negative-exponent case we write $n = - m$ with $m > 0$ a positive integer, so $x^{n} = x^{- m} = 1/ x^{m}$ . Naming $m$ lets us reuse the positive-integer result on $x^{m}$ .
The four stages are the labelled boxes in Figure 2: Stage 1 positive integer $n$ (binomial expansion), Stage 2 $n = 0$ (constant), Stage 3 negative integer $n = - m$ (common denominator), Stage 4 rational $n = p / q$ (implicit differentiation). A fifth idea — irrational $n$ — is reached by continuity.

Figure — Power rule — proof for integer, rational exponents

Intuition Read Figure 2 — how the proof

flows between stages Figure 2 is the road map. Stage 1 (lavender, top-left) is the foundation: the binomial expansion of $(x + h)^{n}$ for a positive integer $n$ . Follow the arrows out of it: one arrow feeds Stage 3 (butter) — because after taking a common denominator, Stage 3's numerator $x^{m} - (x + h)^{m}$ is literally a Stage-1 difference in the positive integer $m$ . A second arrow feeds Stage 4 (coral) — because the chain-rule step $\frac{d}{d x} y^{q} = q y^{q - 1} \frac{d y}{d x}$ is Stage 1 applied to the inner function $y$ . Stage 2 (mint) is the trivial constant case. From Stage 4 a "rationals dense" arrow reaches the irrational- $n$ box: pick rationals crowding in on the irrational exponent, apply Stage 4 to each, then pass to the limit. All arrows finally converge on the single boxed result $\frac{d}{d x} x^{n} = n x^{n - 1}$ . The picture is the dependency graph of the proof.

True or false — justify

The power rule

\frac{d}{d x} x^{n} = n x^{n - 1}

holds for every real

n

True as a result, but no single proof covers all

n

— the four stages handle rational

n

, and irrational

n

is reached afterwards by a continuity/limiting argument (see the edge cases), which is why the parent breaks it into pieces.

The binomial theorem alone proves the power rule for all exponents.

False.

(x + h)^{n}

only has a finite, cleanly-cancelling expansion when

n

is a non-negative integer; for

n = \frac{1}{2}

n = - 3

it becomes an infinite series with no last term to divide out.

\frac{d}{d x} 2^{x} = x 2^{x - 1}

False. That mixes up power rule (variable base, constant exponent) with an exponential (constant base, variable exponent); the correct answer is

2^{x} ln 2

, see Derivative of exponential functions.

For

n = 0

, the formula

n x^{n - 1}

gives

0 \cdot x^{- 1}

, which is undefined at

x = 0

, so the rule fails there.

False. Take the derivative from the definition:

\frac{( x + h ) ^{0} - x ^{0}}{h} = \frac{1 - 1}{h} = 0

for every

h \neq = 0

and every

x

(including

x = 0

), so the limit is

0

everywhere — the "

0 \cdot x^{- 1}

" is never actually evaluated; the leading factor

0

annihilates the whole product for

x \neq = 0

, and the definition itself supplies

0

x = 0

In Stage 4 (rational exponent) you may write

\frac{d}{d x} y^{q} = q y^{q - 1}

because it "looks like the power rule."

False —

y

is a function of $x$ , so the chain rule attaches

\frac{d y}{d x}

: the correct left side is

q y^{q - 1} \frac{d y}{d x}

The negative-integer proof (Stage 3, with

n = - m

) secretly relies on the positive-integer proof.

True. After taking a common denominator it uses

\frac{( x + h ) ^{m} - x ^{m}}{h} \to m x^{m - 1}

, which is Stage 1's result applied to the positive integer

m

— the stages are deliberately stacked.

Squaring both sides of

y = x

to get

y^{2} = x

changes the function, so the derivative we compute isn't the derivative of

x

False. Squaring is a valid, reversible bridge on the domain

x \geq 0

; it lets us use integer powers, and after solving for

\frac{d y}{d x}

we substitute back, so the result is genuinely

\frac{d}{d x} x

The power rule for an irrational exponent like

n = π

needs a brand-new proof unrelated to the rational stages.

False. We approximate

π

by a sequence of rationals (each handled by Stage 4) and pass to the limit; because

x^{n}

and

n x^{n - 1}

are continuous in $n$ (for fixed

x > 0

x^{n} = e^{n l n x}

is continuous in

n

), the formula survives the limit — no separate machinery required.

Spot the error

(x + h)^{1/2} = x^{1/2} + \frac{1}{2} x^{- 1/2} h

, so dividing by

h

gives

\frac{1}{2} x^{- 1/2}

." Where's the flaw?

The expansion is only the first two terms of an infinite series, not an exact equality — you can't legitimately cancel a finite expansion for a non-integer exponent, which is precisely why we use implicit differentiation instead.

"For

x^{- 2}

, expand

(x + h)^{- 2}

with the binomial theorem and cancel like Stage 1." What breaks?

The binomial theorem's finite form requires a non-negative integer exponent;

- 2

produces an infinite series, so nothing cleanly cancels — use the common-denominator trick of Stage 3 (write

- 2 = - m

with

m = 2

"Differentiate

y^{3} = x^{2}

: left side is

3 y^{2}

, right side is

2 x

, so

\frac{d y}{d x} = \frac{2 x}{3 y ^{2}}

." Find the mistake.

The left side is missing the chain-rule factor:

\frac{d}{d x} y^{3} = 3 y^{2} \frac{d y}{d x}

, giving

3 y^{2} \frac{d y}{d x} = 2 x

and

\frac{d y}{d x} = \frac{2 x}{3 y ^{2}}

— the final answer happens to match only because the writer omitted then "recovered" the factor by luck; the reasoning is wrong.

\frac{d}{d x} x^{n} = n x^{n - 1}

, so

\frac{d}{d x} x^{x} = x \cdot x^{x - 1} = x^{x}

." What's wrong?

x^{x}

has a variable base AND variable exponent — neither the power rule nor the exponential rule applies alone; you need logarithmic/implicit differentiation. (The true answer is

x^{x} (ln x + 1)

"In Stage 1, after dividing by the nudge

h

we get

n x^{n - 1} + (2 n) x^{n - 2} h + \dots

; set

h = 0

to finish." Is "set

h = 0

" the right phrase?

Strictly we take the limit

h \to 0

, not plug

h = 0

— plugging

h = 0

into the original

\frac{( x + h ) ^{n} - x ^{n}}{h}

gives

\frac{0}{0}

; the whole point is that after cancellation the limit exists and equals

n x^{n - 1}

Why questions

Why does the binomial theorem, specifically, get invoked for positive integers rather than any other expansion?

Because it exposes the first-order-in- $h$ term

n x^{n - 1} h

separately from all higher-

h

terms; after dividing by the nudge

h

and taking the limit, only that first-order coefficient survives — the binomial theorem is the tool that isolates it.

Why do all terms except

n x^{n - 1}

vanish as

h \to 0

Every other term in the divided expansion still carries at least one factor of the nudge

h

(the

(2 n) x^{n - 2} h

, etc.), and any quantity times a shrinking

h

tends to

0

Why does Stage 3 take a common denominator instead of expanding directly?

Combining

\frac{1}{( x + h ) ^{m}} - \frac{1}{x ^{m}}

produces the numerator

x^{m} - (x + h)^{m}

, which is exactly a Stage-1-shaped difference (in the positive integer

m

) we already know how to limit — it converts a new problem into a solved one.

Why must

q \neq = 0

in the rational exponent

n = p / q

A denominator of

0

makes

p / q

undefined, and the trick

y^{q} = x^{p}

collapses (

y^{0} = 1

carries no information), so there is no equation to differentiate.

Show, step by step, why

\frac{d}{d x} y^{q} = q y^{q - 1} \frac{d y}{d x}

— and why that step is "just Stage 1."

Write the outer power as

g (u) = u^{q}

with

u = y (x)

. Stage 1 (integer power rule) gives

g^{'} (u) = q u^{q - 1}

. The chain rule says the rate of change of

g (y (x))

is how fast $g$ responds to $u$ times how fast $u = y$ responds to $x$ :

\frac{d}{d x} g (y) = g^{'} (y) \cdot \frac{d y}{d x} = q y^{q - 1} \frac{d y}{d x}

. So Stage 1 supplies the

q y^{q - 1}

and the chain rule multiplies by the inner rate

\frac{d y}{d x}

; if

y

were just

x

then

\frac{d y}{d x} = 1

and we recover plain Stage 1.

Why does a small change in the exponent

n

produce only a small change in

x^{n}

(the fact continuity leans on)?

For a fixed base

x > 0

x^{n} = e^{n l n x}

, and

e^{\cdot}

is continuous; so if

n_{k} \to n

then

x^{n_{k}} \to x^{n}

and likewise

n_{k} x^{n_{k} - 1} \to n x^{n - 1}

— small wobble in

n

, small wobble in both sides, which is what lets the formula pass to the irrational limit.

Why does the derivative of

x

blow up near

x = 0

\frac{d}{d x} x = \frac{1}{2 x}

, and as

x \to 0^{+}

the denominator shrinks to

0

, so the slope tends to

+ \infty

— matching the graph's near-vertical steepness at the origin.

Edge cases

What is

\frac{d}{d x} x^{n}

x = 0

when

n = 1

f (x) = x

has

f^{'} (x) = 1 \cdot x^{0} = 1

everywhere, including

x = 0

; here

x^{0} = 1

by convention, so the "

x^{n - 1}

" factor is harmless.

Does the power rule give a derivative for

x^{1/2}

x = 0

No —

\frac{1}{2 x}

is undefined at

x = 0

; the function

x

has a vertical tangent there, so the derivative genuinely does not exist even though

0

is fine.

What about

x^{- 1}

x = 0

x^{- 1} = 1/ x

isn't even defined at

x = 0

(there's a vertical asymptote), so neither the function nor its derivative

- x^{- 2}

exists there — the rule only speaks about points in the domain.

x^{2/3}

differentiable at

x = 0

No —

\frac{2}{3} x^{- 1/3} \to \pm \infty

x \to 0

, giving a sharp cusp; the function is continuous at

0

but has no finite slope.

Where does the negative-exponent formula

- m x^{- m - 1}

break down as a limit argument?

x = 0

, since the working divides by

(x + h)^{m} x^{m} \to x^{2 m}

, which is

0

when

x = 0

— the derivation assumes

x \neq = 0

throughout, consistent with

x^{- m}

being undefined there.

Is the rational-exponent proof valid for negative

x

when

q

is even (e.g.

x^{1/2}

No —

x^{1/2}

isn't real for

x < 0

, so the equation

y^{q} = x^{p}

has no real

y

; the proof only holds where

x^{p / q}

is actually defined (here

x \geq 0

Is the rational-exponent proof valid for negative

x

when

q

is odd (e.g.

x^{1/3}

Yes for

x < 0

, since

x^{1/3}

is real (e.g.

(- 8)^{1/3} = - 2

); but watch the point

x = 0

, where

\frac{1}{3} x^{- 2/3} \to + \infty

gives a vertical tangent — so the formula holds on both sides of

0

but not at

0

How is

x^{n}

even defined for an irrational

n

, and on which bases?

Only for

x > 0

: we define

x^{n} = e^{n l n x}

, and

ln x

needs

x > 0

. For

x < 0

an irrational power like

(- 2)^{π}

has no real value (no consistent odd/even root trick survives an irrational exponent), so the continuity argument for irrational

n

lives strictly on the positive-base domain

x > 0

;

x = 0

is excluded because

ln 0

is undefined.

How does the power rule extend to an irrational exponent such as

x^{π}

, and what makes the rational approximations converge?

Choose rationals

r_{k} \to π

(e.g. truncating

π = 3.14159 \dots

3, \frac{31}{10}, \frac{314}{100}, \dots

); Stage 4 gives

\frac{d}{d x} x^{r_{k}} = r_{k} x^{r_{k} - 1}

for each. Since (for fixed

x > 0

) both

x^{n} = e^{n l n x}

and

n x^{n - 1}

are continuous in

n

, letting

r_{k} \to π

forces

r_{k} x^{r_{k} - 1} \to π x^{π - 1}

— the convergence of the inputs

r_{k} \to π

transfers to the outputs, giving

\frac{d}{d x} x^{π} = π x^{π - 1}

Can we use the power rule on

x^{0}

to conclude the derivative of any constant is

0

Not directly —

x^{0} = 1

is one specific constant; a general constant

c

isn't a power of

x

, but its derivative is

0

for the same underlying reason:

\frac{c - c}{h} = 0

in the limit definition for every

h \neq = 0

Recall One-line summary of the traps

Wrong tool for the exponent class, dropping the chain-rule factor, confusing $x^{n}$ with $n^{x}$ , forgetting that "the rule applies" still requires the point to be in the domain, and thinking irrational exponents need new machinery when continuity finishes the job. ::: Each of these is a place the staged proof structure protects you — if you can name which stage (or the continuity step) handles an exponent, you rarely fall in.

Connections

Power rule — proof for integer, rational exponents — the parent this bank drills.
Limit definition of the derivative — why the nudge $h \to 0$ is a limit, not a plug-in.
Binomial theorem — the finite-vs-infinite distinction behind Stage 1.
Chain rule — the factor everyone drops in Stage 4.
Implicit differentiation — the rational-exponent method.
Derivative of exponential functions — the $n^{x}$ contrast trap.
Quotient rule — an alternative for negative exponents.