Apply Inverse to Both Sides: Isolate a Variable Inside a Function

On this page: The Principle | Conditions | Failure Modes | EE Questions | Retrieval Practice | Practice Ground | Solve a Problem | Related Principles | FAQ | How This Fits

---

The Principle

The move: When a function $f$ is one-to-one and $y$ is in the range of $f$, apply $f^{-1}$ to both sides of $y = f(x)$, producing $f^{-1}(y) = f^{-1}(f(x))$.

The invariant: This preserves the solution set.

Pattern: $y = f(x) \quad\Longrightarrow\quad f^{-1}(y) = f^{-1}(f(x))$

The two conditions must both hold. Compare a valid application against an invalid attempted one:

Legal ✓	Illegal ✗
$y = e^x \Rightarrow \ln y = \ln(e^x)$ — $f(t)=e^t$ is one-to-one; $y > 0 \in \mathrm{ran}(e^t)$	$y = x^2 \Rightarrow \sqrt y = x$ on $\mathbb{R}$ — $f$ is not one-to-one; the correct result is $\sqrt(x^2) =

---

Conditions of Applicability

Condition: $f\ \mathrm{one{-}to{-}one}$; $y\in\mathrm{ran}(f)$

In plain language: every output of $f$ must come from exactly one input, and $y$ must be a value that $f$ can produce.

Before applying, check: (1) Is $f$ one-to-one — does every output value come from exactly one input? (2) Is the right-hand side value $y$ in the range of $f$ — could $f(x) = y$ have a solution at all?

If either condition is violated: Applying $f^{-1}$ may produce an extraneous solution, a multi-valued result, or an undefined expression. The equation that results is not equivalent to the original.

• If $f$ is not one-to-one (e.g., $f(x)=x^2$ on $\mathbb{R}$): the inverse function does not exist as a single-valued map. You must first restrict the domain of $f$ to a region where it is one-to-one, or handle both solution branches explicitly.
• If $y \notin \mathrm{ran}(f)$ (e.g., asking $e^x = -1$): the equation has no solution. Applying $\ln$ to $-1$ is undefined; the move produces a domain error rather than a solution.

Want the complete framework behind this guide? Read Masterful Learning.

---

Common Failure Modes

Failure mode: Apply $f^{-1}$ to both sides of $y = x^2$ without restricting the domain → $\sqrt{y} = |x|$, not $x$. The student writes $x = \sqrt{y}$ and misses the solution branch $x = -\sqrt{y}$.

Debug: Ask “Is $f$ one-to-one?” Graph $f$ mentally or recall its symmetry: if any horizontal line crosses the graph more than once, $f$ is not one-to-one on that domain. For $x^2$ on $\mathbb{R}$, the answer is no: every positive $y$ has two preimages.

---

Failure mode: Attempt to apply $f^{-1}$ with a $y$-value outside $\mathrm{ran}(f)$ (e.g., solving $e^x = -5$) → the expression $\ln(-5)$ is undefined in the reals; the step fails silently if the author does not recognize the range violation.

Debug: Ask “Can $f(x)$ equal this $y$-value?” For exponentials: $e^x > 0$ always — any $y \le 0$ is outside the range and the equation has no real solution.

---

Elaborative Encoding

Use these questions to build deep understanding. (See Elaborative Encoding for the full method.)

Within the Principle

• Why does $f^{-1}(f(x)) = x$ — what property of one-to-one functions guarantees this simplification works?
• What changes in the move if $f$ is one-to-one only on a restricted domain (e.g., $f(x)=x^2$ on $[0, \infty)$)? Is the move still valid under that restriction?

For the Principle

• Suppose $f$ is one-to-one and $y \in \mathrm{ran}(f)$. After applying $f^{-1}$ to both sides and obtaining $f^{-1}(y) = f^{-1}(f(x))$, what does it mean to say the two equations $y = f(x)$ and $x = f^{-1}(y)$ are equivalent?
• How would you verify that a given function is one-to-one before deciding to apply this move?

Between Principles

• How does Apply Inverse to Both Sides relate to the input-recovery form of Inverse Cancellation ($f^{-1}(f(x)) = x$)? Which principle cleans up the right side after the move is made?

Generate an Example

• Construct an equation of the form $y = f(x)$ where $f$ is one-to-one on a restricted domain only, and show explicitly why simply applying $f^{-1}$ without stating the domain restriction leads to a missing solution.

---

Retrieval Practice

Answer from memory, then click to reveal and check. (See Retrieval Practice for the full method.)

State the apply-inverse move in one sentence: _____If f is one-to-one and y is in the range of f, apply f-inverse to both sides of y = f(x) to get f-inverse(y) = f-inverse(f(x)).

Write the canonical move pattern: _____$y=f(x) \Rightarrow f^{-1}(y)=f^{-1}(f(x))$

State the canonical condition: _____$f\ \mathrm{one{-}to{-}one}; y\in\mathrm{ran}(f)$

---

Practice Ground

Use these exercises to build move-selection fluency. (See Self-Explanation for how to use worked examples effectively.)

Procedure Walkthrough

Starting from $25 = 5^x$, isolate $x$ using the apply-inverse move.

Assume $f(t) = 5^t$ (one-to-one on $\mathbb{R}$) and $25 \in \mathrm{ran}(5^t)$ since $5^2 = 25 > 0$.

Step	Expression	Operation
0	$25 = 5^x$	—
1	$\log_5(25) = \log_5(5^x)$	Apply $f^{-1} = \log_5$ to both sides; valid — $f$ is one-to-one and $25 \in \mathrm{ran}(f)$
2	$\log_5(25) = x$	Inverse cancellation: $\log_5(5^x) = x$
3	$x = 2$	Evaluate $\log_5(25) = 2$ (since $5^2 = 25$)

---

Drills

Goal micro-chain — reach the isolated variable form

For each equation, check both conditions (one-to-one and $y \in \mathrm{ran}(f)$), then apply $f^{-1}$ to both sides and simplify to isolate $x$.

---

Assume $f$ is one-to-one and $y \in \mathrm{ran}(f)$. Apply $f^{-1}$ to both sides and isolate $x$.

$3 = e^x$

Reveal

$f^{-1} = \ln$. Apply to both sides:

$\ln 3 = \ln(e^x) = x$

---

Check both conditions, then apply $f^{-1}$ to both sides and isolate $x$.

$7 = 2x + 1, \quad f(t) = 2t + 1$

Reveal

$f(t) = 2t+1$ is one-to-one (non-zero slope); $7 \in \mathrm{ran}(f)$. $f^{-1}(t) = \dfrac{t-1}{2}$.

$x = \frac{7-1}{2} = 3$

---

Check both conditions, then apply $f^{-1}$ to isolate $x$.

$8 = x^3, \quad f(t) = t^3$

Reveal

$f(t) = t^3$ is one-to-one; $8 \in \mathrm{ran}(f)$. $f^{-1}(t) = t^{1/3}$.

$8^{1/3} = (x^3)^{1/3} \implies x = 2$

---

Check both conditions, then apply $f^{-1}$ to isolate $x$.

$\log_2(x) = 5$

Write the equation in the form $y = f(x)$ first, identifying $f$.

Reveal

Rewrite: $f(t) = \log_2(t)$, $y = 5$. $\log_2$ is one-to-one; $5 \in \mathrm{ran}(\log_2) = \mathbb{R}$. $f^{-1}(t) = 2^t$.

$2^5 = 2^{\log_2(x)} \implies x = 32$

---

Forward step — apply the move once and state the result

Identify $f$ and $f^{-1}$, verify the two conditions, then write the transformed equation after applying $f^{-1}$ to both sides. (Do not yet simplify the right side with inverse cancellation.)

---

State $f$ and $f^{-1}$, verify both conditions, and write the transformed equation (one step only).

$4 = \sqrt{x}, \quad \text{domain } x \ge 0$

Reveal

$f(t) = \sqrt{t}$ on $[0,\infty)$ — one-to-one; $4 \ge 0$ so $4 \in \mathrm{ran}(f)$. $f^{-1}(t) = t^2$.

$4^2 = (\sqrt{x})^2$

The right side simplifies via Inverse Cancellation — a separate move.

---

State $f$ and $f^{-1}$, verify both conditions, and write the transformed equation (one step only).

$-2 = \ln(x)$

Reveal

$f(t) = \ln t$ on $(0,\infty)$ — one-to-one; $-2 \in \mathbb{R} = \mathrm{ran}(\ln)$. $f^{-1}(t) = e^t$.

$e^{-2} = e^{\ln(x)}$

The right side simplifies via Inverse Cancellation — a separate move.

---

Condition failure — domain violation. Verify whether both conditions hold before attempting the move.

$-4 = e^x$

Reveal

$f(t) = e^t$ is one-to-one ✓. But $y = -4$: since $e^t > 0$ for all real $t$, we have $-4 \notin \mathrm{ran}(e^t)$ ✗.

The second condition fails. The move cannot be applied — there is no real $x$ satisfying $e^x = -4$. Applying $\ln(-4)$ is undefined in $\mathbb{R}$.

Correct conclusion: no real solution.

---

Near-miss — condition failure: not one-to-one. A student writes ”$y = x^2$, so $\sqrt{y} = x$.” What condition is violated?

Reveal

$f(t) = t^2$ on $\mathbb{R}$ is not one-to-one — both $x = 3$ and $x = -3$ satisfy $x^2 = 9$. Condition 1 fails; $f^{-1}$ does not exist as a single-valued map on $\mathbb{R}$.

To recover both solutions: write $\sqrt{y} = |x|$, giving $x = \sqrt{y}$ or $x = -\sqrt{y}$. Or restrict to $x \ge 0$.

---

Eligibility check: From the list below, identify which equations are eligible for the apply-inverse move as stated (with unrestricted domain), and which are not. For each ineligible case, state which condition fails.

1. $5 = e^x$
2. $9 = x^2$ (domain $\mathbb{R}$)
3. $-1 = \ln(x)$
4. $0 = e^x$
5. $8 = 2^x$

Reveal

Equation	$f$ one-to-one?	$y \in \mathrm{ran}(f)$?	Eligible?
$5 = e^x$	✓ ($e^t$ strictly increasing)	✓ ($5 > 0$)	Yes — apply $\ln$ to both sides
$9 = x^2$, $\mathbb{R}$	✗ (not one-to-one on $\mathbb{R}$)	—	No — condition 1 fails
$-1 = \ln(x)$	✓ ($\ln$ one-to-one on $(0,\infty)$)	✓ ($-1 \in \mathbb{R} = \mathrm{ran}(\ln)$)	Yes — apply $e^{(\cdot)}$ to both sides
$0 = e^x$	✓	✗ ($e^t > 0$; $0 \notin \mathrm{ran}(e^t)$)	No — condition 2 fails; no real solution
$8 = 2^x$	✓ ($2^t$ strictly increasing)	✓ ($8 = 2^3 > 0$)	Yes — apply $\log_2$ to both sides

---

Canonicalization — simplify after applying the inverse

After applying $f^{-1}$ to both sides, the right side contains $f^{-1}(f(x))$. Simplify to the standard form $x$ using inverse cancellation.

---

What is the canonical simplified form of the right side?

$\ln(e^x) = ?$

Reveal

$\ln(e^x) = x$ — inverse cancellation with $f(t) = e^t$, $f^{-1}(t) = \ln t$.

(Valid because $f$ is one-to-one and $x \in \mathrm{dom}(f)$.)

---

What is the canonical simplified form of the right side?

$\log_3(3^x) = ?$

Reveal

$\log_3(3^x) = x$ — inverse cancellation with $f(t) = 3^t$, $f^{-1}(t) = \log_3 t$.

---

Solve a Problem

Apply what you’ve learned with Problem Solving.

Problem: Starting from $50 = 10^{2x-1}$, isolate $x$ using the apply-inverse move. Verify both conditions before applying $f^{-1}$.

Condition check: $f(u) = 10^u$ is one-to-one (strictly increasing); $50 > 0$ so $50 \in \mathrm{ran}(10^u)$. Both conditions hold.

Full solution

Step	Expression	Move
0	$50 = 10^{2x-1}$	—
1	$\log_{10}(50) = \log_{10}(10^{2x-1})$	Apply $f^{-1} = \log_{10}$ to both sides
2	$\log_{10}(50) = 2x - 1$	Inverse cancellation: $\log_{10}(10^u) = u$
3	$2x = \log_{10}(50) + 1$	Add $1$ to both sides
4	$x = \dfrac{\log_{10}(50) + 1}{2}$	Divide both sides by $2$

---

Related Principles

Principle	Relationship
Inverse Definition	Prerequisite — defines what makes $f$ one-to-one and what $f^{-1}$ is as a relation; Apply Inverse to Both Sides operationalizes that relation as a move
Inverse Cancellation	Follow-on step — in this solve-chain context, uses the input-recovery form to simplify $f^{-1}(f(x))$ to $x$ after the apply-inverse move; the two principles typically appear together
Composition Expansion	Structural complement — both involve domain conditions and a two-step evaluation chain; composition chains functions forward while inverse application unwinds them
Log-Exponential Rewrite	Concrete successor — once a logarithmic equation is isolated, this is the specific rewrite move that often follows inverse-style reasoning in practice

---

FAQ

What does “Apply Inverse to Both Sides” mean?

It means: if you have an equation $y = f(x)$ and $f$ has an inverse, you apply $f^{-1}$ to the left side and to the right side simultaneously, producing $f^{-1}(y) = f^{-1}(f(x))$. That is the move. Simplifying $f^{-1}(f(x))$ to $x$ is a subsequent step governed by Inverse Cancellation.

When is the apply-inverse move valid?

Two conditions must hold: (1) $f$ must be one-to-one — every output comes from exactly one input, so $f^{-1}$ exists as a single-valued function. (2) $y$ must be in the range of $f$ — the equation $f(x)=y$ must have a solution at all. If either condition fails, the move is not valid as stated.

What goes wrong if $f$ is not one-to-one?

If $f$ is not one-to-one, it has no single-valued inverse. Applying "$f^{-1}$" picks only one of the preimages and silently discards the others — producing a missing-solution error. For example, “solving” $9 = x^2$ by writing $x = \sqrt{9} = 3$ misses the solution $x = -3$.

What goes wrong if $y$ is not in the range of $f$?

The equation $f(x) = y$ has no solution. The expression $f^{-1}(y)$ is not defined in the intended real setting — there is no valid input for $f^{-1}$ to act on. The result is a domain error rather than a solution.

How does this move relate to log-exponential rewriting?

The log-exponential rewrite ($y = \log_b x \Leftrightarrow b^y = x$) is a specific instance of the apply-inverse pattern: $f(t) = b^t$ with $f^{-1}(t) = \log_b t$ (or vice versa). The general apply-inverse principle covers any one-to-one function, including trigonometric inverses, cube roots, and custom function inverses.

Do I need to include the condition check every time?

Yes — explicitly, especially when there is any risk the condition could fail (e.g., squaring, exponential, piecewise functions). For elementary linear functions ($f(x) = ax + b$, $a \ne 0$) the one-to-one check is trivial, but stating it once builds the habit that catches harder cases.

---

How This Fits in Unisium

Unisium treats Apply Inverse to Both Sides as a condition-critical move — one where skipping the one-to-one check causes silent errors that look like valid algebra. Practice sessions in the Unisium Study System embed the condition verification as the first required step in every drill and use near-miss cases (like $y = x^2$ on $\mathbb{R}$) to build the checking habit.

The move ends the moment both sides are wrapped in $f^{-1}$: the equation is now $f^{-1}(y) = f^{-1}(f(x))$. The subsequent step that collapses $f^{-1}(f(x))$ to $x$ is Inverse Cancellation — a distinct principle. This guide covers the first half of that two-step sequence.

Explore further:

• Functions Subdomain Map — Return to the functions hub to see how inverse moves, logarithms, and composition are grouped together in the release-prep cluster
• Inverse Definition — The representational prerequisite: what a one-to-one function is and why the inverse relation becomes a function
• Inverse Cancellation — The cleanup step that typically follows the apply-inverse move in a standard solve chain, using the input-recovery form $f^{-1}(f(a)) = a$
• Log-Exponential Rewrite — A common specialized follow-on when the inverse relationship is logarithmic or exponential
• Elaborative Encoding — Build deep understanding of why both conditions are necessary, not just convenient

Ready to master Apply Inverse to Both Sides? Start practicing with Unisium or explore the full learning framework in Masterful Learning.

Masterful Learning

The study system for physics, math, & programming that works: retrieval, connection, explanation, problem solving, and more.

Read the book (opens in new tab) ISBN 979-8-2652-9642-9