Consider the example of the multivariable function
\begin{equation}
f(x,y) = \frac{2}{x^2+y^2+1}\text{.}\tag{6.5.1}
\end{equation}
First let’s find the partial derivative at \((1,0)\text{.}\) The partial derivative of \(f\) with respect to \(x\) is given by
\begin{equation*}
f_x(x,y) = \frac{-4x}{(x^2+y^2+1)^2}
\end{equation*}
and at the point \((1,0)\) we have
\begin{equation*}
f_x(1,0) = \frac{-4}{(1^2+0^2+1)^2} = -\frac{4}{4}=-1\text{.}
\end{equation*}
Let’s take another approach. If we consider the partial derivative at
\((1,0)\text{,}\) we know that
\(y\) is fixed at
\(0\text{.}\) Referring back to
Subsection 6.4.1 we should be able to find this path on the function. This can be accomplished by allowing
\begin{equation*}
x(t)=t \qquad y(t)=0 \qquad z(t) = f(t,0)= \frac{2}{t^2+0^2+1} = \frac{2}{t^2+1}\text{.}
\end{equation*}
in \(\mathbf{r}(t) = \langle x(t), y(t), z(t) \rangle\text{.}\) This parametric equation follows the function along \(y=0\) and passes through \((1,0, 1)\) when \(t=1\text{.}\)
This parametric equation has a derivative, which we learned about in
Section 5.2. The derivative at of this path is given by
\(\mathbf{r}\'(t) = \langle 1,0, \frac{-4t}{(t^2+1)^2} \rangle\) At
\(t=1\) (the value at which the path passing through
\((1,0,1)\)),
\(\mathbf{r}\'(1) = \langle 1,0, \frac{-4}{(1^2+1)^2} \rangle = \langle 1,0, -1 \rangle \text{.}\)
Notice we have that
\begin{equation*}
f_x(1,0) = -1 \qquad \text{and} \qquad \mathbf{r}\'(1) =\langle 1,0, -1 \rangle \text{.}
\end{equation*}
Let’s rewrite this,
\begin{equation*}
f_x(1,0) = \frac{-1}{1} \qquad \text{and} \qquad \mathbf{r}\'(1) =\langle 1,0, -1 \rangle \text{.}
\end{equation*}
It is no coincidence that these are similar. The partial \(f_x(1,0)= \frac{-1}{1}\) can be interpreted using the typical \(\frac{\text{rise}}{\text{run}}\) for slope. This means at the point \((x=1,y=0)\) the rate of elevation change is \(-1\) (down 1) for a one unit change in the positive \(x\) direction. For the rate of change \(\mathbf{r}\'(1) =\langle 1,0, -1 \rangle\) we can interpret the \(0\) as we are not changing in the \(y\) direction This corresponds to \(y\) being fixed when we consider partials with respect to \(x\text{.}\) The other values are interpreted the same as before. That is, a one unit change in \(x\) corresponds to an elevation change of \(-1\text{.}\)
You can explore the function and the partial with respect to \(x\) using the tool below.