Derivative at a point

In mathematics, the derivative of a function is a measure of how rapidly the function changes locally when its argument changes.

In order to define the derivative, a difference quotient is constructed. The derivative of the function f at a is the limit

$f'(a)=\lim _{h\to 0}{f(a+h)-f(a) \over h}$

as h approaches zero, if this limit exists. If the limit exists, then f is said to be differentiable at a. If a function is differentiable in all the points in which it is defined, then it is said to be differentiable.

If a function is differentiable in a point, it is also continuous in that point. The reverse is not true, which we shall soon see:

Example. Consider the function $f=|x|$, (where $|x|$ is the absolute value of x). The function is continuous in the point 0, since when zero is approached from either side, the limit of the function is zero. We study the derivative in the point zero:

$f'(0)=\lim _{h\to 0}{|0+h|-|0| \over h}.$

We see that this expression has the limit 1 when we approach zero from the right side, but the limit -1 when we approach from the left side. Hence, the function is not differentiable.

Some notational styles

These are all equivalent ways to denote the derivative of a function f in the point x.

$f'(x)\!$

$\mathrm {D} f(x)\!$

${\frac {df}{dx}},\quad {\frac {d}{dx}}f,\quad {\frac {dy}{dx}}\quad \mathrm {with} \quad y=f(x).$

Multivariable calculus

The extension of the concept of derivative to multivariable functions, or vector-valued functions of vector variables, may be achieved by considering the derivative as a linear approximation to a differentiable function. In the one variable case we can regard $x\mapsto f(a)+f'(a)(x-a)$ as a linear function of one variable which is a close approximation to the function $x\mapsto f(x)$ at the point $x=a$.

Let $F:\mathbf {R} ^{n}\rightarrow \mathbf {R} ^{m}$ be a function of n variables. We say that F is differentiable at a point $a\in \mathbf {R} ^{n}$ if there is a linear function $\mathrm {D} F:\mathbf {R} ^{n}\rightarrow \mathbf {R} ^{m}$ such that

${\frac {\Vert F(a+h)-F(a)-\mathrm {D} F(h)\Vert }{\Vert h\Vert }}\rightarrow 0{\hbox{ as }}\Vert h\Vert \rightarrow 0\,$

where $\Vert \cdot \Vert$ denotes the Euclidean distance in $\mathbf {R} ^{n}$.

The derivative $\mathrm {D} F$, if it exists, is a linear map and hence may be represented by a matrix. The entries in the matrix are the partial derivatives of the component functions of F_j with respect to the coordinates x_i. If F is differentiable at a point then the partial derivatives all exist at that point, but the converse does not hold in general.

Formal derivative

The derivative of the monomial Xⁿ may be formally defined as $D:X^{n}\mapsto n.X^{n-1}$ and this extends to a linear map D on the polynomial ring $R[X]$ over any ring R. Similarly we may define D on the ring of formal power series $R[[X]]$.

The map D is a derivation, that is, an R-linear map such that

$D:fg\mapsto f.Dg+Df.g\,.$