A function can be continuous at a point and still not differentiable there1. The origin of is the example.
The derivative at a point is defined as the limit of the average rate of change. At the limit to examine is the following.
| Approach | |
|---|---|
The value differs according as approaches from the right or from the left, so the limit does not exist.
For the function is the line and for it is . The two join at the origin, but their slopes disagree. That is why the graph comes to a point there.
Since joins up without a break at the origin it is continuous, but the slope is undetermined, so it is not differentiable. The converse does hold: wherever a function is differentiable it is continuous.
The distinction matters when looking for a maximum or a minimum. The function attains its minimum at , yet does not hold there.
Those three kinds are the candidates for a maximum or a minimum.
| Function | What happens at the origin |
|---|---|
| slopes of and from the two sides | |
| slopes of and from the two sides | |
| slope from both sides, a vertical tangent |
The first two are pointed shapes and the last has a vertical tangent; none of them is differentiable at the origin.
Whether a function is differentiable can be judged by whether the graph looks like a line under magnification. The origin of approaches a line the more it is magnified, while the origin of stays a V however far one zooms in.
Away from the origin is differentiable, with for and for . The derivative is a function with a step, matching , and only the origin is left undefined.
A function built with absolute values or with corners is handled by splitting the domain at the corners. Inside each piece differentiation is straightforward, and only the corners need separate treatment.
The V on the graph is , the two lines are and , and the large dot is the non-differentiable point .