Calculus of Two and More Variables
Open in the Padhaiko app1.1 Partial differentiation
For a function of two variables, the partial derivative with respect to is the ordinary derivative taken while is held constant:
1.1.1 Partial derivatives of first and higher order
Higher order partial derivatives
Differentiating and again gives the second order derivatives
Schwarz's (Clairaut's) theorem: if and are continuous at a point, then there. So a function of two variables has three distinct second derivatives, and a function of variables has .
1.1.2 Homogeneous function: Euler's theorem for two and three variables
Homogeneous function
A function is homogeneous of degree if for all . For example has degree , has degree , and has degree . The same definition extends to .
Euler's theorem
If is homogeneous of degree in (and ), then
Second-order consequence (differentiate the relation again):
Proof
Put , in and differentiate with respect to using the chain rule: . Setting gives the result.
1.1
If , show that .
A homogeneous function of : let . Replacing by gives , so is homogeneous of degree .
Euler's theorem on : . Since , and :
Solving:
.
1.1.3 Total derivatives and differentials, composite and implicit functions
Total differential and chain rules
- Total differential. For : . It is the linear approximation of the change in .
- Total derivative. If , : .
- Composite function. If with , :
- Implicit function. If defines as a function of :
For defining : and .
1.1.4 Jacobians and their properties
Jacobian
For and the Jacobian of with respect to is the determinant
For three functions of three variables it is the determinant of all first partial derivatives.
Properties of Jacobians
- Reciprocal: .
- Chain rule: if are functions of and of then .
- Functional dependence: and are functionally dependent (one is a function of the other) exactly when identically.
- Change of variables: in a multiple integral .
1.1.4
Find for , , and hence . Also find the Jacobian of the polar transformation , .
The partial derivatives: , , , .
The determinant:
By the reciprocal property, .
Polar case: .
, , and .
1.2 Extreme values of two and three variables. Lagrange's multiplier
Maxima and minima of
A point is a local maximum (minimum) if is greater (less) than at all nearby points. A necessary condition is at (a stationary point). Let , , at the point:
| Condition at the stationary point | Conclusion |
|---|---|
| and | local maximum |
| and | local minimum |
| saddle point (no extremum) | |
| test fails; examine the function near the point |
Lagrange's method of multipliers
To find the extreme values of subject to a constraint , form
and solve together with for . The same method works for two variables, and with two constraints one uses two multipliers .
1.3 Application in optimization of function of two variables in one constraint
Optimisation problems are solved in three stages: translate the quantity to be maximised or minimised into , write the restriction as , then use Lagrange's method (or eliminate one variable). The nature of the stationary point comes from the geometry of the problem, or from the second derivative test.
1.2
- Find the extreme values of .
- Find the rectangle of greatest area that can be inscribed in the ellipse with sides parallel to the axes.
Part 1 — stationary points: gives and gives . The points are , , , .
Part 1 — classification: , , , so .
- : , , so a minimum, .
- : , , so a maximum, .
- and : , so saddle points.
Part 2 — Lagrange's method: the corner in the first quadrant gives area . Take :
Eliminating (multiply the first by , the second by and subtract): , i.e. . The constraint then gives , so and .
Part 2 — the area: (the general result is ). This is the maximum, since the area can be made arbitrarily small and the parametrisation has largest value .
- Minimum value at ; maximum value at ; and are saddle points.
- Greatest area , with the corner at , i.e. sides and .
Practice Exercises — Partial derivatives, Jacobians and extrema
If , where and , find at .
Chain rule: , , , :
Substituting , : .
At : . (Check: , so .)
, equal to at .
If , find , and evaluate it at when .
Implicit differentiation: let . Then and :
At , : the point lies on the curve (), and .
; the slope at is .
Show that the Jacobian for , , is . What does this say about when two of are equal?
The determinant:
Column operations: replace by and by (the determinant is unchanged). Row 1 becomes , row 2 becomes and row 3 becomes .
Expanding along row 1:
Since , we get .
Meaning: when any two of coincide, and then are functionally dependent there.
, which vanishes when two of the variables are equal.
Find the maximum and minimum values of .
Stationary points: and . If : gives . If : gives . The points are , , , .
Second derivatives: , , .
- : , , and , a maximum, .
- : , , and , a minimum, .
- : , , , saddle points.
Maximum value at ; minimum value at (local extremes); are saddle points.
Use Lagrange's multipliers to find the minimum value of subject to .
Setting up: , so and . Hence and .
Using the constraint: , so , and , .
The value: . It is a minimum: it is the squared distance from the origin to the line, and the distance to a line is least at the foot of the perpendicular (any other point of the line is farther away, for instance gives ).
Minimum value at .