If a function has a derivative at a point, it must be continuous there — a slope can't exist across a gap or a jump. So differentiability is the stronger property: every differentiable function is continuous, but plenty of continuous functions fail to be differentiable. That arrow only points one way.
What you're looking at: the logic in both directions. The forward arrow always holds; the backward arrow is crossed out because continuity alone is not enough.
Flip “differentiable \(\Rightarrow\) continuous” around: not continuous \(\Rightarrow\) not differentiable. So the instant you spot a jump, hole, or asymptote, you're done — no derivative can live there.
The interesting cases are the ones that are continuous but still fail. That's the next section.
When a continuous function has no derivative at a point, it's one of four culprits. Tap each to see the graph and the reason. The first three are continuous — the curve is connected, but it isn't smooth.
What you're looking at: the live graph for the selected failure mode, with the trouble point marked.
Here's the deepest way to see it. A function is differentiable at a point exactly when, if you zoom in far enough, the curve becomes indistinguishable from a straight line — the tangent. Zoom in on a smooth point and it flattens to a line. Zoom in on a corner and it stays a corner forever.
What you're looking at: the full curve, with a magnifier lens you slide along it. Inside the lens you see the curve up close. On the smooth curve, every spot looks like a straight line through the lens. Switch to the corner and slide the lens over the kink — it stays sharp, no matter how close you look.
“Has a derivative” and “looks like a line up close” are the same statement. Whatever the lens reveals as a straight line, its slope is the derivative there.
The AP exam's favorite version of this: a function defined in pieces, glued at a seam. To check differentiability there, run two tests in order. First continuous? (do the pieces meet?) Then smooth? (do the one-sided slopes match?). It must pass both.
What you're looking at: left piece is \(f(x)=x^2\) for \(x\le 1\); right piece is the line \(mx+b\) for \(x \gt 1\). Tune \(m\) and \(b\) and watch the two tests.
On the exam you're often just shown a graph and asked where \(f\) is not differentiable. Scan for the four culprits: breaks (not continuous), corners, cusps, and vertical tangents. Here's a graph with several — can you spot them before revealing?
What you're looking at: a single continuous-looking graph with four trouble points. Each is a different failure mode.
Everywhere else, the graph is a smooth curve — differentiable. The derivative exists at every point except these four.
Differentiable always implies continuous — a slope can’t exist across a break. This direction never fails.
Continuous does not imply differentiable. \(|x|\) is continuous at 0 but has a corner — no slope there.
A derivative exists at a point only when the left and right slopes agree. At a corner they differ; at a smooth point they match.
Not continuous \(\Rightarrow\) not differentiable. Spot a jump, hole, or asymptote and you’re instantly done — no derivative there.
Five problems — the implication, the failure modes, and the piecewise seam test (including a solve-for-\(k\)). Try each before revealing.
Eight fast checks across the whole lesson.
The single most common slip. Continuity is necessary but not sufficient — a continuous corner like \(|x|\) has no derivative. Always check smoothness separately.
It's differentiable \(\Rightarrow\) continuous, not the reverse. Memorize the one-way arrow; the backward direction is false.
On a piecewise problem, matching the values makes it continuous — but you still must match the one-sided slopes for differentiability. Two tests, not one.
\(\sqrt[3]{x}\) is continuous and even has a tangent line at 0 — but it's vertical, so the slope is infinite and \(f'(0)\) doesn't exist.
“Make it differentiable” usually needs two conditions (continuity and matching slopes), which may give two unknowns. Set up both equations.
A corner has two different finite slopes; a cusp has slopes running to \(\pm\infty\). Both kill the derivative, but name them correctly if asked.