Understanding Derivatives of Inverses (Including Inverse Trig)

Differentiating Inverse Functions

What an inverse function is (and what it is not)

An inverse function is a function that “undoes” another function. If a function $f$ takes an input $x$ and outputs $f(x)$, then its inverse $f^{-1}$ takes the output back to the original input:

$f(f^{-1}(x)) = x$

and

$f^{-1}(f(x)) = x$

These equations are the defining property of inverses.

A crucial warning: the notation $f^{-1}$ does **not** mean the reciprocal function $\frac{1}{f(x)}$. Students often confuse these because exponent $-1$ suggests “reciprocal,” but for functions, $-1$ in the superscript means “inverse function,” not “one over.”

Why inverses matter for derivatives

Derivatives describe rates of change and slopes. Inverse functions often show up when you solve equations for an input in terms of an output. For example, if a model gives $y = f(x)$ but you need $x$ as a function of $y$, you are using an inverse.

The key geometric idea is this: the graph of $y = f(x)$ and the graph of $y = f^{-1}(x)$ are reflections across the line $y = x$. Reflection across $y = x$ swaps the roles of “run” and “rise,” which strongly suggests that slopes should turn into reciprocals.

That intuition becomes the main derivative relationship:

• the slope of the inverse at a corresponding point is the reciprocal of the slope of the original function at the matching point.

But you have to be careful about where you evaluate each slope—because the points swap coordinates.

Conditions: when an inverse is differentiable

For the inverse-derivative rule to work in the way you’ll use it in AP Calculus AB, you need:

1. $f$ is one-to-one on the interval you care about (so the inverse exists as a function).
2. $f$ is differentiable at the relevant point, and its derivative is not zero there.

That second condition matters because the inverse derivative involves dividing by $f'$. If $f'(\text{something}) = 0$, the inverse has a vertical tangent there (slope undefined), so the inverse derivative is not finite.

The inverse derivative formula (what it says and how to use it)

If $y = f^{-1}(x)$, then by definition:

$f(y) = x$

Differentiate both sides with respect to $x$. On the left, $y$ depends on $x$, so you must use the chain rule:

$\frac{d}{dx}[f(y)] = f'(y)\frac{dy}{dx}$

and the right side differentiates to $1$:

$f'(y)\frac{dy}{dx} = 1$

Solve for $\frac{dy}{dx}$:

$\frac{dy}{dx} = \frac{1}{f'(y)}$

Now replace $y$ with $f^{-1}(x)$:

$\frac{d}{dx}[f^{-1}(x)] = \frac{1}{f'(f^{-1}(x))}$

That is the main rule you use.

A point-specific version (common on AP questions)

Often you are asked for $\left(f^{-1}\right)'(a)$ at a specific input $a$. Then it’s helpful to write:

$\left(f^{-1}\right)'(a) = \frac{1}{f'(f^{-1}(a))}$

This tells you the workflow:

1. Find $f^{-1}(a)$, which is the number $b$ such that $f(b) = a$.
2. Compute $f'(b)$.
3. Take the reciprocal.

Conceptually: the point $\left(a, f^{-1}(a)\right)$ lies on the inverse, which corresponds to the point $\left(f^{-1}(a), a\right)$ on the original.

Seeing the “reciprocal slope” idea geometrically

Suppose $f(2) = 5$. Then the point $\left(2,5\right)$ is on $y=f(x)$, and the point $\left(5,2\right)$ is on $y=f^{-1}(x)$.

If the tangent slope to $f$ at $x=2$ is $f'(2)=3$, then the inverse’s tangent slope at $x=5$ is:

$\left(f^{-1}\right)'(5) = \frac{1}{3}$

So the reciprocal is real, but notice the input switches from $2$ to $5$.

Worked example 1: using the formula at a point

Problem. A function $f$ is differentiable and one-to-one. You are given $f(3)=7$ and $f'(3)=4$. Find $\left(f^{-1}\right)'(7)$.

Step 1: interpret $f^{-1}(7)$.
Since $f(3)=7$, it means $f^{-1}(7)=3$.

Step 2: apply the inverse derivative formula.

$\left(f^{-1}\right)'(7) = \frac{1}{f'(f^{-1}(7))} = \frac{1}{f'(3)} = \frac{1}{4}$

Common pitfall: using $\frac{1}{f'(7)}$ instead. The derivative must be evaluated at the input to $f$ (here $3$), not at the input to $f^{-1}$ (here $7$).

Worked example 2: finding $\left(f^{-1}\right)'(a)$ from a table

AP questions often provide a table of values for $f$ and $f'$.

Problem. Suppose a table gives $f(1)=4$ and $f'(1)=-2$. Find $\left(f^{-1}\right)'(4)$.

Because $f(1)=4$, we know $f^{-1}(4)=1$. Then:

$\left(f^{-1}\right)'(4) = \frac{1}{f'(1)} = \frac{1}{-2} = -\frac{1}{2}$

Worked example 3: using implicit differentiation without memorizing the formula

Sometimes you can derive the same result quickly in-context.

Problem. Let $f(x)=x^3+x$ and let $g(x)=f^{-1}(x)$. Find $g'(2)$.

Step 1: convert the inverse relationship.
If $g=f^{-1}$, then:

$f(g(x))=x$

Here that is:

$[g(x)]^3+g(x)=x$

Step 2: differentiate both sides.

$3[g(x)]^2g'(x)+g'(x)=1$

Factor out $g'(x)$:

$g'(x)(3[g(x)]^2+1)=1$

So:

$g'(x)=\frac{1}{3[g(x)]^2+1}$

Step 3: find $g(2)$ by solving $f(u)=2$.
We need $u$ such that $u^3+u=2$. Testing $u=1$ gives $1+1=2$, so $g(2)=1$.

Then:

$g'(2)=\frac{1}{3(1)^2+1}=\frac{1}{4}$

This matches the inverse-derivative logic: $f'(x)=3x^2+1$, so $f'(1)=4$ and the inverse derivative at $2$ is $\frac{1}{4}$.

What can go wrong (and how to self-check)

1. Mixing up inputs. If you’re computing $\left(f^{-1}\right)'(a)$, you need the point on $f$ where the output is $a$. A good self-check is to explicitly write “find $b$ such that $f(b)=a$.”
2. Forgetting the inverse exists only on a one-to-one interval. If $f$ isn’t one-to-one, you can sometimes restrict its domain to make it one-to-one—but you must know which branch you’re using.
3. Ignoring the possibility of a vertical tangent. If $f'(b)=0$, then $\left(f^{-1}\right)'(a)$ is undefined (the inverse has a vertical tangent there).

Exam Focus

• Typical question patterns:
  • Given values like $f(c)=a$ and $f'(c)=m$, find $\left(f^{-1}\right)'(a)$.
  • Use a table of $f$ and $f'$ values to compute inverse-derivative values.
  • Given a graph of $f$ with a tangent slope at a point, infer the slope of $f^{-1}$ at the reflected point.
• Common mistakes:
  • Using $\frac{1}{f'(a)}$ instead of $\frac{1}{f'(f^{-1}(a))}$.
  • Treating $f^{-1}(x)$ as $\frac{1}{f(x)}$.
  • Forgetting to verify/find $f^{-1}(a)$ (the matching input on the original function).

Differentiating Inverse Trigonometric Functions

What “inverse trig” actually means

Inverse trigonometric functions reverse trig functions, but only after restricting trig functions to be one-to-one.

For example, sine is not one-to-one on all real numbers because many angles share the same sine value. To define an inverse, we restrict sine to a principal interval where it is one-to-one, then define:

• $y=\arcsin(x)$ means $\sin(y)=x$ with $y$ restricted to a principal range.
• $y=\arccos(x)$ means $\cos(y)=x$ with $y$ restricted to a principal range.
• $y=\arctan(x)$ means $\tan(y)=x$ with $y$ restricted to a principal range.

These principal ranges matter because they control the sign of cosine or sine that appears in derivative work.

Domains and ranges you must know

For AP Calculus AB, you should know these standard principal definitions:

• $\arcsin(x)$ has domain $[-1,1]$ and range $\left[-\frac{\pi}{2},\frac{\pi}{2}\right]$.
• $\arccos(x)$ has domain $[-1,1]$ and range $[0,\pi]$.
• $\arctan(x)$ has domain $(-\infty,\infty)$ and range $\left(-\frac{\pi}{2},\frac{\pi}{2}\right)$.

Why you care: when you use identities like $\sin^2(y)+\cos^2(y)=1$, you will need to decide whether $\cos(y)$ is positive or negative. The principal range tells you.

A quick notation table (so you don’t misread the symbols)

Important: $\sin^{-1}(x)$ means $\arcsin(x)$, not $\frac{1}{\sin(x)}$.

Why inverse trig derivatives matter

Inverse trig derivatives appear constantly inside bigger differentiation problems because they are common antiderivatives in later units and common inverse relationships in modeling.

Also, they are a perfect “integration of ideas” topic: you need inverse functions, implicit differentiation, trig identities, and the chain rule.

Deriving the derivative of $\arcsin(x)$ (from scratch)

Let:

$y=\arcsin(x)$

By definition, that means:

$\sin(y)=x$

Differentiate both sides with respect to $x$. Use the chain rule on the left:

$\cos(y)\frac{dy}{dx}=1$

So:

$\frac{dy}{dx}=\frac{1}{\cos(y)}$

Now we need to rewrite $\cos(y)$ in terms of $x$. From $\sin(y)=x$ and the Pythagorean identity:

$\cos^2(y)=1-\sin^2(y)=1-x^2$

So $\cos(y)=\sqrt{1-x^2}$ or $\cos(y)=-\sqrt{1-x^2}$. Which one is correct?

Because $y=\arcsin(x)$ is restricted to $\left[-\frac{\pi}{2},\frac{\pi}{2}\right]$, cosine is nonnegative on that interval. Therefore:

$\cos(y)=\sqrt{1-x^2}$

Substitute back:

$\frac{d}{dx}[\arcsin(x)]=\frac{1}{\sqrt{1-x^2}}$

This derivative only makes sense for $|x|<1$ as a function derivative (endpoints involve vertical tangents).

Deriving the derivative of $\arccos(x)$

Let:

$y=\arccos(x)$

Then:

$\cos(y)=x$

Differentiate:

$-\sin(y)\frac{dy}{dx}=1$

So:

$\frac{dy}{dx}=-\frac{1}{\sin(y)}$

Use $\sin^2(y)=1-\cos^2(y)=1-x^2$. On the range $[0,\pi]$, sine is nonnegative, so:

$\sin(y)=\sqrt{1-x^2}$

Therefore:

$\frac{d}{dx}[\arccos(x)]=-\frac{1}{\sqrt{1-x^2}}$

Notice it is the negative of the $\arcsin$ derivative.

Deriving the derivative of $\arctan(x)$

Let:

$y=\arctan(x)$

Then:

$\tan(y)=x$

Differentiate:

$\sec^2(y)\frac{dy}{dx}=1$

So:

$\frac{dy}{dx}=\frac{1}{\sec^2(y)}$

Use the identity $\sec^2(y)=1+\tan^2(y)$ and substitute $\tan(y)=x$:

$\sec^2(y)=1+x^2$

Thus:

$\frac{d}{dx}[\arctan(x)]=\frac{1}{1+x^2}$

This one works for all real $x$.

Using these derivatives correctly: the chain rule is not optional

On the AP exam, inverse trig functions almost always appear as part of a composite, like $\arcsin(2x)$ or $\arctan(x^2+1)$. You must combine the inverse trig derivative with the chain rule.

If $y=\arcsin(u)$ where $u=u(x)$, then:

$\frac{dy}{dx}=\frac{u'}{\sqrt{1-u^2}}$

Similarly:

$\frac{d}{dx}[\arccos(u)]=-\frac{u'}{\sqrt{1-u^2}}$

and

$\frac{d}{dx}[\arctan(u)]=\frac{u'}{1+u^2}$

Here $u'$ means $\frac{du}{dx}$.

Worked example 1: differentiating a composite with $\arcsin$

Problem. Find $\frac{d}{dx}[\arcsin(2x)]$.

Let $u=2x$, so $u'=2$. Apply the chain rule form:

$\frac{d}{dx}[\arcsin(2x)]=\frac{2}{\sqrt{1-(2x)^2}}=\frac{2}{\sqrt{1-4x^2}}$

A natural domain note: the inside of arcsine must satisfy $-1\le 2x\le 1$, so the original function is defined for $-\frac{1}{2}\le x\le \frac{1}{2}$, and the derivative has vertical tangent behavior near the endpoints.

Worked example 2: differentiating $\arctan$ with algebra inside

Problem. Find $\frac{d}{dx}[\arctan(x^2+1)]$.

Let $u=x^2+1$, so $u'=2x$. Then:

$\frac{d}{dx}[\arctan(x^2+1)]=\frac{2x}{1+(x^2+1)^2}$

A common mistake is to write $\frac{1}{1+x^2+1}$, which incorrectly “drops the square” on $u$.

Worked example 3: mixing inverse trig with other derivative rules

Problem. Find $\frac{d}{dx}[x\arccos(x)]$.

This requires the product rule. Let:

$y=x\arccos(x)$

Then:

$y'=1\cdot \arccos(x)+x\cdot \frac{d}{dx}[\arccos(x)]$

Use the derivative formula for $\arccos(x)$:

$\frac{d}{dx}[\arccos(x)]=-\frac{1}{\sqrt{1-x^2}}$

So:

$y'=\arccos(x)-\frac{x}{\sqrt{1-x^2}}$

Connecting inverse trig derivatives back to inverse-function derivatives

You can think of inverse trig derivatives as special cases of the inverse-function derivative rule.

For instance, sine and arcsine are inverses on the restricted sine interval. The general inverse rule says:

$\frac{d}{dx}[f^{-1}(x)]=\frac{1}{f'(f^{-1}(x))}$

If $f(x)=\sin(x)$ (with the restricted domain where it’s one-to-one), then $f^{-1}(x)=\arcsin(x)$ and $f'(x)=\cos(x)$, so:

$\frac{d}{dx}[\arcsin(x)] = \frac{1}{\cos(\arcsin(x))}$

Then the triangle/identity work converts $\cos(\arcsin(x))$ into $\sqrt{1-x^2}$. This viewpoint helps you remember that the inverse derivative involves “reciprocal of derivative evaluated at the inverse,” not just a random formula.

Common conceptual traps with inverse trig

1. Confusing $\arcsin(x)$ with $\frac{1}{\sin(x)}$. The latter is $\csc(x)$. Inverse trig is an inverse function, not a reciprocal.
2. Forgetting the chain rule. If the inside isn’t exactly $x$, you need to multiply by the inside derivative.
3. Sign mistakes from ignoring principal ranges. When deriving, deciding that $\cos(y)=\sqrt{1-x^2}$ uses the fact that $y$ is in the principal range for $\arcsin$. If you don’t respect the range, you can get the wrong sign.

Exam Focus

• Typical question patterns:
  • Differentiate expressions like $\arcsin(u(x))$, $\arccos(u(x))$, or $\arctan(u(x))$, often combined with product/quotient rules.
  • Use implicit differentiation with a step like $\sin(y)=x$ to justify a derivative formula or to differentiate an equation involving inverse trig.
  • Evaluate a derivative at a specific point (sometimes requiring you to compute the inside value first, then plug into the derivative).
• Common mistakes:
  • Writing $\frac{1}{\sqrt{1-x}}$ or $\frac{1}{\sqrt{1-x^2}}$ but forgetting the inside substitution, leading to forms like $\sqrt{1-(2x)}$ instead of $\sqrt{1-(2x)^2}$.
  • Dropping parentheses when squaring, such as turning $(x^2+1)^2$ into $x^2+1^2$.
  • Mixing up inverse trig with reciprocal trig, especially interpreting $\sin^{-1}(x)$ as $\csc(x)$.