Skip to content
UoL CS Notes

Partial Derivatives & Multi-variable Functions

COMP116 Lectures

Consider a function such as:

\[z=-(3x^2+2y^2+xy)\]

What are the values of $x$ and $y$ which maximise $z$?

Are there any such values?

Partial Derivatives

For the prior function:

  1. We need values of $x$ and $y$ which maximise $z$.
  2. We know $x$ to maximise $z$ when $y$ is fixed.
  3. We know $y$ to maximise $z$ when $x$ is fixed.
  4. But we want to do things simultaneously.
  5. Assume that $y$ is fixed and find the “right” $x$.
  6. Assume that $x$ is fixed and find the “right” $y$.

From this we can make a set of simultaneous equations to solve.

Example

For $z=-(3x^2+2y^2+xy)$.

  1. Differentiate as if $y$ was a constant:

    \[f_x(x,y)=-(6x+y)\]
  2. Differentiate $f(x,y)$ as if $x$ was a constant.

    \[f_y(x,y)=-(4y+x)\]
  3. Find the values of $x,y$ for which $f_x(x,y)=0,$ $f_y(x,y)=0$ by solving the simultaneous equations.

Notation

For functions of two variables ($z=f(x,y)$) we have:

  • $f_x(x,y)$ - The first partial derivative of $f(x,y)$ with respect to $x$.
  • $f_y(x,y)$ - The first partial derivative of $f(x,y)$ with respect to $y$.

Also used are:

\[\frac{\partial z}{\partial x}; \frac{\partial z}{\partial y}\]

For a function of $n$ variables:

\[z = f(x_1,x_2,\ldots,x_k,\ldots,x_n)\]

this style of notation will be used:

\[\frac{\partial z}{\partial x_k}\]

Example Continued

From the prior example the two simultaneous equations were:

\[\begin{aligned} -6x-y&=0\\ -x-4y&=0 \end{aligned}\]

The only solution is: $x=0, y=0$.

Therefore, this function is critical at the point $(0,0)$.

To test if this is maximum we need an analogue of the second derivative test.

Second Derivative Test with 2 Variables

  1. With 2 variables there are 4 possible forms of the second order partial derivative.
  2. With $n$ variables there are $n^2$.

How do we combine these?

The 4 forms: $(f_{xx},f_{xy},f_{yx},f_{yy})$ or:

\[\frac{\partial^2z}{\partial x^2},\frac{\partial^2z}{\partial x\partial y},\frac{\partial^2z}{\partial y\partial x},\frac{\partial^2z}{\partial y^2}\]

Interpretation of Notation

  • $f_{xx}$ - The partial derivative of $f_x$ with respect to $x$.
  • $f_{xy}$ - The partial derivative of $f_y$ with respect to $x$.

For $f(x,y)=-(3x^2+2y^2+xy)$:

\[\begin{aligned} f_{xx}&=-6\\ f_{xy}&=-1\\ f_{yx}&=-1\\ f_{yy}&=-4 \end{aligned}\]

In general $f_{xy}$ and $f_{yx}$ are identical functions.

Using the Test

There is a precondition:

\[(f_{xx}f_{yy}-(f_{xy})^2)(\alpha, \beta)>0\]
  • If $f_{xx}(\alpha, \beta)>0$ the point is a minimum.
  • If $f_{xx}(\alpha, \beta)<0$ the point is a maximum.
  • If $f_{xx}(\alpha, \beta)=0$ no conclusion can be made.

Where $(\alpha, \beta)$ is the critical point.