1. Start with continuity and derivative definitions
Continuity at an interior input requires the function value to exist, the two-sided function limit to exist, and those values to agree:
Differentiability at an interior input requires a finite real limit of the difference quotient:
Continuity compares nearby outputs with a base output. Differentiability compares limiting rates. At a domain endpoint, continuity uses the available side; derivative endpoint conventions must also be identified explicitly. A point must belong to the original function's domain before either its continuity there or its derivative there can be established.
Use Topic 2.2: derivative definition and notation and continuity at a point to review these definitions. Derivative values estimated in Topic 2.3 do not themselves prove differentiability.
2. Differentiability implies continuity
Core concept: A finite derivative at a point forces continuity there, but continuity alone does not guarantee a derivative. A continuous function can fail to be differentiable at a corner, cusp or vertical tangent; a point of discontinuity also cannot have a derivative. Use definitions and one-sided difference quotients to justify each conclusion.
DIFFERENTIABILITY IMPLIES CONTINUITY
If is differentiable at , then is continuous at
Symbolic Form:
Contrapositive (Also True):
Plain English: If you can take the derivative at a point, the function must be continuous there. Any break in the function means no derivative at that point!
THE CONVERSE IS FALSE!
It is NOT true that: If is continuous at , then is differentiable at
Continuity does NOT guarantee differentiability! A function can be continuous but still fail to have a derivative (corners, cusps, vertical tangents).
Visual Summary:
- All differentiable functions are continuous
- Some continuous functions are NOT differentiable
- At a point where a function is discontinuous, it is not differentiable there. It may still be differentiable at other inputs.
A stronger condition: Differentiability requires a single finite limiting rate, not merely connected-looking behavior. A graph that appears smooth at a chosen scale is not a proof that the derivative exists.
The derivative can only be defined where the original function is defined. In symbols, . At a hole, even a finite limit of the nearby function values does not supply the missing base value. Review types of discontinuities for the domain checks.
3. Why differentiability forces continuity: the proof
Theorem proof
A finite derivative forces continuity
1 · State the assumption and the goal
Given: exists
To Prove: is continuous at , i.e.,
We need to show:
2 · Rewrite the output change using the quotient
Multiply and divide by :
3 · Use the two finite limits
Use limit properties to separate:
Recognize the derivative:
4 · Conclude continuity at the same input
Therefore: , so is continuous at ∎
Why the proof works: The difference-quotient limit exists as a finite real number. For , multiplying the quotient by recovers . The product tends to zero by the limit laws. This does not use an undefined quotient at .
4. Four common ways a finite derivative can fail
When a finite does not exist
The following are four common reasons a finite derivative fails to exist at an input. They are not an exhaustive list: an oscillating difference quotient can also fail to converge.
Case 1
A discontinuity rules out a derivative
1 · Identify the condition
If has any discontinuity at (jump, removable, infinite), then does NOT exist.
Why: By the theorem, differentiability requires continuity. No continuity → no derivative!
2 · Recognize a jump, hole or asymptote
Examples:
- Jump discontinuity: at
- Removable discontinuity: at (if not defined there)
- Infinite discontinuity: at
Case 2
A corner has two finite rates
1 · Identify the condition
At a corner, the left and right difference-quotient limits are finite but unequal. The function is continuous at the corner, but there is no single tangent slope.
Test: Check if (one-sided derivatives don't match)
2 · Read the rates for absolute value
Classic Example: at
- Left derivative:
- Right derivative:
- Since , does NOT exist
3 · Recognize related corner examples
Other Examples:
- at
- at
- Any piecewise function with different slopes meeting at a point
Case 3
A cusp has opposite unbounded rates
1 · Identify the condition
At the standard cusp illustrated here, the function is continuous while its one-sided difference quotients diverge with opposite signs. There is no finite derivative.
Characteristic: A pointed peak or valley with oppositely unbounded one-sided rates. This differs from both a finite-slope corner and a same-sign vertical-tangent example.
2 · Compute the defining quotient
For the cusp , the base-point quotient is . It tends to from the left and from the right.
3 · Distinguish a cusp from a corner
Classic Example: or at
- As , slope
- As , slope
- The derivative does not exist (infinite from both sides)
Difference from a corner: A corner has two distinct finite one-sided rates. A cusp of this type has oppositely unbounded one-sided rates. Infinity is not a real derivative value.
Case 4
A vertical tangent has matching unbounded signs
1 · Identify the condition
If the function is continuous at an interior input and both one-sided difference quotients diverge to the same signed infinity, the graph has a vertical tangent there. The tangent line exists geometrically, but it has no finite slope, so the ordinary derivative does not exist.
Use the defining quotient: Test from each side. The statement alone does not distinguish a cusp from a vertical tangent, nor does it establish that the original function is defined and continuous at the base point.
2 · Compute the defining quotient
For the cube-root vertical tangent, tends to from both sides.
3 · Compare with other root examples
Classic Example: or at
- As ,
- The function is continuous but not differentiable at
Other Examples:
- at (from the right)
- at
Summary Table:
| Case | What Happens | Classic Example |
|---|---|---|
| Discontinuity | Function has break | Step function, at 0 |
| Corner | Slopes differ (finite) | at 0 |
| Cusp | Slopes infinite, opposite | at 0 |
| Vertical Tangent | Slope infinite (same direction) | at 0 |
Corner
Cusp
Vertical tangent
Jump discontinuity
The first three functions are continuous at zero. Open and filled markers distinguish excluded and included points. Plot edges show the viewing window.
Domain example
A hole and its continuous extension
1 · Keep the original excluded input
The original rational function excludes one. Cancelling its common factor does not create a value or a derivative there.
2 · Define the new base value explicitly
If a new extension explicitly assigns the value two at one, it becomes the line everywhere, and that extended function has derivative one there. The derivative belongs to the specified function, not just a simplified expression.
Endpoint example
Check the available side
1 · Recognize the domain endpoint
The square-root example has an endpoint at zero.
2 · Check for a finite right-hand rate
Its right-hand quotient diverges as . Thus it has no finite right-hand derivative; there is also no available left side for an ordinary two-sided derivative.
5. Define and compare one-sided derivatives
Left-Hand Derivative (from the left):
Right-Hand Derivative (from the right):
Key Fact:
At an interior point, the ordinary derivative exists exactly when the two one-sided difference-quotient limits exist as finite real numbers and are equal. At a domain endpoint, identify a requested one-sided derivative separately; an unavailable side cannot silently be supplied.
How to Use One-Sided Derivatives:
- For piecewise functions: Verify the defined boundary value and continuity first. Then calculate the one-sided difference quotients using that same boundary value. With polynomial or other smooth branches that meet continuously, their endpoint branch slopes give those one-sided derivatives.
- For corners: Calculate both one-sided derivatives; if they differ, no derivative exists
- For continuity: Even if function is continuous, one-sided derivatives must match for differentiability
Here and denote the left and right derivative limits using the actual value . They are not simply the limits of a formula for as its input approaches the point. A finite derivative can exist even when the derivative function is not continuous nearby.
6. Five worked examples with complete justification
Example 1
A continuous absolute-value corner
1 · Check the defined value and continuity
Problem: Determine if is differentiable at
Check continuity:
Continuous at
2 · Rewrite the absolute value by side
Rewrite as piecewise:
3 · Compare the two finite slopes
Find left-hand derivative:
For : , so
Find right-hand derivative:
For : , so
4 · Justify why the derivative fails
Compare:
Conclusion: NOT differentiable at because the one-sided derivatives don't match. This is a corner.
Example 2
Can one parameter satisfy both conditions?
1 · Read the branches and the base value
Problem: For which is the following function differentiable at two?
Preserve the base value: .
2 · Find the only continuous choice
Continuity: The left limit is four and the right limit is , so continuity requires , or .
3 · Compute both quotients for that choice
Use that continuity value: With , the one-sided derivative limits are
They differ, so the only continuous choice gives a corner.
4 · Explain why no parameter works
Explain the conflicting conditions: Matching the two branch slopes would require , but then the right limit would be eight while the defined value remains four. Branch-slope matching is insufficient when continuity fails.
Answer: No value of makes this specified function differentiable at two. Continuity requires , whereas matching branch slopes would require ; these cannot both hold.
Example 3
Match the values and the slopes
1 · Identify the join and the unknowns
Problem: Find and so that is differentiable at
2 · Set the continuity condition
Continuity condition:
Equation 1:
3 · Compare the one-sided rates after continuity
Differentiability condition:
Equation 2:
Continuity requires . Only after imposing that relation does the right-hand quotient simplify to the branch slope :
The left-hand quotient tends to two. Thus and , preserving both the base value and the derivative.
4 · Solve the system and check the join
Solve the system:
From Equation 2:
Substitute into Equation 1:
Answer: and
Result: is differentiable everywhere
Example 4
Locate and justify a cusp
1 · Identify the domain and continuity
Problem: Identify all points where is not differentiable
Check continuity: is continuous everywhere (cube root is defined for all real numbers)
2 · Test the defining quotient at the candidate
The defining quotient at the cusp input two is
It diverges negatively from the left and positively from the right, proving the claimed failure. Away from two the displayed derivative formula is valid.
3 · Check all other inputs using the power rule
Find the derivative:
Check where doesn't exist:
When , i.e.,
At , the formula for the derivative away from the join has a zero denominator. To decide whether exists, use the defining quotient at the actual base value.
As ,
Answer: Not differentiable at
Reason: The function is continuous and the defining quotients diverge with opposite signs, so the graph has a cusp.
Example 5
Recognize a vertical tangent
1 · Confirm continuity at the base point
Problem: Show that has a vertical tangent at
Check continuity:
Continuous at
2 · Use the quotient and compare its signs
The defining quotient at zero is . Both sides diverge positively. The tangent is the vertical line , but a finite derivative value does not exist.
3 · Check nearby derivatives as an extension
Find the derivative:
Check behavior at :
The slopes blow up to from both sides
Conclusion: The function is continuous at . The graph has a vertical tangent there, so there is no finite derivative.
A continuous corner
A differentiable join
In both examples, the left branch supplies the filled join value and the right branch approaches that same point. The curves continue beyond the viewing window.
7. A reliable differentiability checklist
The Complete Differentiability Test (Step-by-Step)
To determine if is differentiable at :
- Step 1: Check if is in the domain
- If is undefined → NOT differentiable
- Step 2: Check continuity at
- Verify
- If discontinuous → NOT differentiable (by theorem)
- Step 3: Check for corners/cusps
- For piecewise: Check if
- For absolute values: Rewrite and check one-sided derivatives
- If one-sided derivatives differ → NOT differentiable
- Step 4: Check for vertical tangents
- Compute the one-sided difference quotients at the base input. Use finite equality for differentiability; same-sign unbounded rates for a continuous vertical-tangent example; opposite signs for a cusp of the type illustrated.
- If a defining quotient is unbounded, there is no ordinary finite derivative. Keep this conclusion distinct from simply seeing a formula undefined after substitution.
- Step 5: If all tests pass
- Conclude differentiability only after establishing a finite two-sided difference-quotient limit, or applying a valid differentiability rule with its hypotheses. A visual checklist alone is not sufficient.
8. Know the limits of visual shortcuts
Enrichment 1
Oscillation supplies another failure case
1 · Establish continuity at zero
Define and for nonzero inputs. Since , the function is continuous at zero. However,
2 · Compute the difference quotient
3 · Explain why the rate has no limit
oscillates without a limit. Thus the function is continuous but not differentiable at zero, even though this behavior does not fit a simple corner/cusp/vertical-tangent sketch.
Enrichment 2
Differentiable does not mean the derivative is continuous
1 · Define the function at zero
Define and for nonzero inputs. Its base-point quotient satisfies
2 · Bound the defining quotient
3 · Distinguish the derivative from its limit
Therefore . For nonzero inputs, , which has no limit at zero. This is why a test involving only a nearby derivative formula cannot replace the defining quotient at the point.
These examples are extensions of the core topic. They clarify that differentiability is a precise finite-limit condition, not a synonym for an informal visual notion of “smooth.”
9. Recognition guides, strategies and common mistakes
Essential Tips
- Memorize the theorem: Differentiable → Continuous (NOT the reverse!)
- Check continuity first: If discontinuous, automatically not differentiable
- Absolute value needs a check: If a differentiable inner function crosses zero with a nonzero derivative, its absolute value has a corner there. A zero alone does not guarantee a corner: is differentiable at zero, and even crosses with a flat derivative and remains differentiable there.
- Fractional exponents: Watch for vertical tangents at (e.g., )
- One-sided derivatives: Must exist AND be equal for differentiability
- Piecewise functions: Always check boundary points carefully
Quick Recognition Guide
Spotting Non-Differentiable Points:
| If you see... | Likely problem | What to check |
|---|---|---|
| Absolute value | Corner | One-sided derivative limits at candidate zeros of the inner function |
| Piecewise function | Corner or discontinuity | Continuity and one-sided derivatives at boundaries |
| with | Possible cusp, vertical or endpoint behavior; classify from the defining quotient. | Original domain, continuity and one-sided difference quotients; account for odd/even root domains. |
| Rational function | Discontinuity | Excluded inputs from the original denominator; check any separately defined extension |
| , | Cusp | One-sided derivatives at critical point |
Memory Tricks
Mnemonic for Non-Differentiable Cases:
"Don't Create Crazy Vertical tangents!"
- Discontinuities
- Corners
- Cusps
- Vertical tangents
Common Mistakes to Avoid
- Mistake 1: Thinking continuous implies differentiable (FALSE!)
- Mistake 2: Forgetting to check continuity before testing differentiability
- Mistake 3: Only checking one-sided limit instead of one-sided derivatives
- Mistake 4: Not checking BOTH continuity AND matching derivatives for piecewise functions
- Mistake 5: Assuming smooth-looking graphs are differentiable everywhere (check corners!)
- Mistake 6: Confusing vertical tangent with vertical asymptote (different concepts!)
- Mistake 7: Assuming every zero or sign change inside an absolute value creates a corner. Compute the one-sided rates; a flat zero can remain differentiable.
- Mistake 8: Forgetting that must be a FINITE number to exist
10. Four practice problems with solutions
Decide whether a finite ordinary derivative exists at each stated input, and justify the answer.
1. Shifted absolute value
at .
Check continuity and the one-sided rates
Practice 1 · Solution
Shifted absolute value
1 · Check the function value and limits
The function value and both function limits are zero, so it is continuous there. For nonzero input changes,
2 · Compare the one-sided rates
3 · Decide differentiability
The unequal finite rates form a corner; no derivative exists.
2. A real fractional power
at zero, interpreted as .
Check the cusp and real domain
Practice 2 · Solution
A real fractional power
1 · Check the real domain
The function is defined and continuous for every real input. Its derivative away from zero is . At zero use the quotient itself:
2 · Evaluate the defining quotient
3 · Classify the two unbounded rates
It tends to from negative inputs and from positive inputs. Hence zero is a cusp, with no finite derivative.
3. A continuous piecewise join
Test differentiability at one. The function name here is ; use for an input change.
Check the defined value and both quotients
Practice 3 · Solution
A continuous piecewise join
1 · Use the actual boundary value
, and both one-sided function limits are two. The function is continuous. However,
2 · Compare the two difference quotients
3 · Explain the continuous corner
The one-sided derivatives differ, so the continuous join is a corner and is not differentiable.
4. A square-root endpoint
at zero.
Check the endpoint convention and finite-rate requirement
Practice 4 · Solution
A square-root endpoint
1 · Check right continuity at the endpoint
The domain is , and the function is continuous from the right at zero. For positive input changes,
2 · Evaluate the available quotient
3 · State the finite-rate conclusion
There is no finite right-hand derivative, and no ordinary two-sided derivative at this endpoint. The graph has a vertical tangent from its available side. For , the derivative formula is valid; it is not a finite derivative at zero.
11. Explain differentiability in an AP-style response
What the AP® Exam Expects:
- Know the theorem: State that differentiability implies continuity
- Check both conditions: For piecewise, verify continuity AND matching derivatives
- Justify your answer: Explain WHY the function is/isn't differentiable
- Use proper notation: Write and clearly
- Show one-sided limits: Calculate left and right derivatives separately
- Identify the case: State if it's a corner, cusp, vertical tangent, or discontinuity
- Graph interpretation: Be able to identify non-differentiable points from graphs
Common FRQ Formats:
- "Determine if f is differentiable at . Justify your answer."
- "Find the values of and that make f both continuous and differentiable at "
- "Explain why the derivative does not exist at "
- "Is f continuous at ? Is f differentiable at ? Justify."
- "Identify all points where the function is not differentiable"
- "Show that if exists, then f is continuous at " (proof question)
12. Quick reference
| Concept | Key Fact |
|---|---|
| Main Theorem | Differentiable → Continuous (NOT reverse!) |
| Contrapositive | Discontinuous → Not differentiable |
| 4 Non-Diff Cases | Discontinuity, Corner, Cusp, Vertical tangent |
| One-Sided Test | exists at an interior input exactly when and both values are finite. |
| Corner Example | at 0 |
| Vertical Tangent Example | at 0 |
| Cusp Example | at 0 |
13. Connections and next study steps
Topic 2.4 connects to:
- Topic 2.1-2.2: Foundation for understanding what derivatives mean
- Topic 2.5+: You need to know where derivatives exist before using rules
- Unit 5: The Mean Value Theorem needs continuity on a closed interval and differentiability on its interior. Optimization also requires attention to endpoints and points where derivatives fail.
- Later derivative tests: First and second derivatives provide sufficient tests for behavior when their hypotheses hold. Concavity is not defined merely by the existence of a second derivative.
- Unit 6: A continuous integrand gives the standard differentiation conclusion in the Fundamental Theorem of Calculus. This theorem is distinct from the differentiability-implies-continuity result studied here.
- Real-world: Many phenomena have points where instantaneous rates don't exist
Remember: Differentiability implies continuity at the same input. Continuity does not imply differentiability. Preserve the original domain, verify the base value, then check the defining one-sided rates. The ordinary derivative must be finite. For a piecewise join, both continuity and matching finite one-sided derivatives are required.
