equals for and for . It keeps the right half of a parabola and reflects the left half through the origin, turning the even function into an odd one. It is the simplest well-known example of a function that can be differentiated once but not twice.
Both the domain and the range are all of the real numbers. Since , the function is odd, with rotational symmetry about the origin. It is positive for and negative for , and the origin is its only zero.
| Range of | |||
|---|---|---|---|
| undefined | |||
The derivative is for and for , which can be written together as a single formula.
At the origin the one-sided derivatives are both and therefore agree, so is well defined. The derivative thus exists on the whole real line and is continuous. Since the function increases throughout and has no extrema.
The derivative is continuous, but being an absolute value it is not differentiable at the origin. The second derivative is for and for , values that disagree at . The function is therefore continuously differentiable once but not twice. In a single short formula it shows that smoothness comes in degrees.
From the sign of the second derivative, the curve is concave up for and concave down for , so the concavity reverses at the origin, making it an inflection point. The tangent there is horizontal and yet there is no extremum, just as for .
| Function | Slope at the origin | The origin | How often differentiable |
|---|---|---|---|
| inflection point | any number of times | ||
| inflection point | once only |
The difference is that can be differentiated any number of times, whereas this function stumbles on the second derivative. The two graphs meet at and , with bulging further out near the origin.
On the side where , solving gives ; where , solving gives . Together the inverse is written as follows.
This is the so-called signed square root, the even function rebuilt as an odd one. It is constructed in exactly the same way as the original function.
The antiderivative is , which equals for and for ; differentiating on either interval returns . Being odd, the function integrates to over any interval symmetric about the origin.
Air resistance proportional to the square of the speed is written, once direction is taken into account, as proportional to . Writing simply would leave the drag pointing backwards even when the body moves backwards; multiplying by keeps the magnitude quadratic while making the direction always oppose the motion. A plot of drag against velocity is exactly this curve. In numerical analysis it also serves as a test function for checking the order of smoothness a method requires.