Professional Documents
Culture Documents
Partial differentiation
√
Example 1.1 What is the maximal domain of the real function g defined by g(x) = x2 + 3x + 2?
Solution : The key point is that the square root only gives a real result if the argument is non-negative.
Hence we need to identify those x for which x2 + 3x + 2 ≥ 0. Factorizing this quadratic gives
x2 + 3x + 2 = (x + 2)(x + 1) ≥ 0 ⇐⇒ x ≤ −2 or x ≥ −1,
(To see this, draw the graph of (x + 2)(x + 1)). Hence the maximal domain is (−∞, −2] ∪ [−1, ∞). ¤
p
Example 1.2 Determine the maximal domain of the function f defined by f (x, y) = 1 − x2 − y 2 .
Solution : Since the square root is defined if and only if 1 − x2 − y 2 ≥ 0, f has maximal domain
D = {(x, y) : x2 + y 2 ≤ 1}, which is the unit disk centre (0, 0). ¤
p
Example 1.3 Sketch the graph of f (x, y) = − 1 − 2x − x2 − y 2 .
Example
p 1.4 By considering the level curves and the cross-sections x = 0 and y = 0, obtain a sketch of
z = x2 + y 2 .
1
p
Figure 1.1: Graph of f (x, y) = − 1 − 2x − x2 − y 2
2x + y + 4z = 1,
where x, y, z ≥ 0.
Solution : We consider the cross-section with the coordinate planes x = 0 (y, z-plane), y = 0 (x, z-plane)
and z = 0 (x, y-plane).
The cross-section of 2x + y + 4z = 1 with x = 0 is the line y + 4z = 1 (lying in the y, z-plane). This
passed through the points (0, 0, 41 ) and (0, 1, 0). In a similar way we obtain the cross-section with the other
coordinate planes; 2x + 4z = 1 in the x, z-plane, passing through (0, 0, 41 ) and ( 12 , 0, 0) and 2x + y = 1 in the
x, y-plane, passing through (0, 1, 0) and ( 21 , 0, 0).
A sketch of the plane is shown in Figure 1.4. ¤
∂f ∂f
Example 1.6 Find and where
∂x ∂y
Solution :
(a) To calculate the partial x derivative, we think of y as a constant and differentiate in the usual way
with respect to x. Hence, we have
∂f ∂ ∂
= y 2 (x3 ) + (x) = 3x2 y 2 + 1.
∂x ∂x ∂x
For the y derivative, we think of x as a constant and differentiate with respect to y;
∂f ∂ ∂
= x3 (y 2 ) + (x) = 2x3 y.
∂y ∂y ∂y
(b) Let u = x + xy. We have f (x, y) = sin u. Then by the chain rule (for a function of one variable),
∂f d ∂u
= (sin u) = cos u × (1 + y) = (1 + y) cos(x + xy).
∂x du ∂x
2
Figure 1.2: Cross sections
It was only for the purpose of explanation that the substitution u was made. When familiar with the process,
the chain rule may be used directly
∂f ∂
= cos(x + xy) (x + xy) = x cos(x + xy).
∂y ∂y
µ ¶
∂z x
Example 1.7 Find where z = sin−1 and x, y > 0.
∂x x+y
[Note that sin−1 u is the inverse sine function (sometimes written as arcsin u), and not the reciprocal 1/ sin u.
The domain of sin−1 is [−1, 1] and, since x, y > 0, x/(x + y) lies in this domain.]
∂z d ∂u
= (sin−1 u) .
∂x du ∂x
We have
d 1 1 |x + y| x+y
(sin−1 u) = √ =s µ ¶2 = p(x + y)2 − x2 = p2xy + y 2 ,
du 1 − u2 x
1−
x+y
3
p
Figure 1.3: The cone z = x2 + y 2
∂ ∂
∂u (x).(x + y) − x. (x + y) y
= ∂x ∂x = .
∂x (x + y) 2 (x + y)2
Hence
∂z y 1
= p .
∂x x + y 2xy + y 2
¤
4
and similarly,
∂r ∂r
uy = f 0 (r) , uz = f 0 (r) .
∂y ∂z
Now
∂ 2 ∂ 2 ∂r
(r ) = (x + y 2 + z 2 ) i.e., 2r = 2x.
∂x ∂x ∂x
Therefore
∂r x
= ,
∂x r
and similarly,
∂r y ∂r z
= , = .
∂y r ∂z r
Thus we have
x2 0 y2 z2 (x2 + y 2 + z 2 ) 0
xux + yuy + zuz = f (r) + f 0 (r) + f 0 (r) = f (r) = rf 0 (r),
r r r r
as required. ¤
Example 1.9 Determine all second order derivatives of u = sin xy and verify that uxy = uyx .
∂ ∂ ¡ ¢
uxx = (ux ) = y cos xy = −y 2 sin xy,
∂x ∂x
∂ ∂ ∂ ¡ ¢
uxy = (ux ) = (y). cos xy + y cos xy = cos xy − yx sin xy,
∂y ∂y ∂y
∂ ∂ ∂ ¡ ¢
uyx = (uy ) = (x). cos xy + x cos xy = cos xy − xy sin xy,
∂x ∂x ∂x
∂ ∂ ¡ ¢
uyy = (uy ) = x cos xy = −x2 sin xy.
∂y ∂y
Example 1.10 Let u = f (x/y), where f is an arbitrary (twice differentiable, with continuous second deriva-
tive) function of one variable. Show that
xux + yuy = 0,
5
Solution : Using the chain rule, we have,
µ ¶ µ ¶ µ ¶
x ∂ x 1 x
ux = f 0 = f0 ,
y ∂x y y y
µ ¶ µ ¶ µ ¶
x ∂ x x x
uy = f 0 = − 2 f0 .
y ∂y y y y
So, µ ¶ µ ¶
1 x x 0 x
xux + yuy = x f 0 − y 2f = 0.
y y y y
[Although we could proceed by calculating uxx , uxy and uyy and taking the appropriate combination, it is
much less work to deduce the final part as indicated below.]
Since xux + yuy = 0, its x- and y-derivatives must also equal 0. Hence
and
xuxy + yuyy + uy = 0. (2)
Taking x × (1) + y × (2) [the need to have the correct coefficient for uxx and uyy dictates the choice of this
combination of (1) and (2)] we get
as required.
¤
Example 1.11 Let w = u2 + v 2 where u = sin θ and v = cos φ. Use the chain rule to calculate wθ and wφ
in terms of θ and φ.
∂u ∂w ∂v ∂w
wθ = + = cos θ.2u + 0.2v = 2 cos θ sin θ = sin 2θ,
∂θ ∂u ∂θ ∂v
and
∂u ∂w ∂v ∂w
wφ = + = 0.2u + (− sin φ).2v = −2 sin φ cos φ = −2 sin 2φ.
∂φ ∂u ∂φ ∂v
¤
∂f
= x2 + y + 9,
∂x
where f is a function of two independent variables x and y.
6
Solution : Integrating with respect to x and treating y as fixed gives
Z
x3
f= x2 + y + 9dx = + xy + 9x + A(y) ,
y fixed 3
∂2f
= 2x,
∂x∂y
where A is an arbitrary function depending on the fixed variable y. Integrating with respect to y, holding x
fixed then gives
Z Z
2 2
f= x + A(y)dy = x y + A(y)dy + B(x) = x2 y + C(y) + B(x).
x fixed x fixed
B is an arbitrary functionof the fixed variable x. Since A wasR an arbitrary function of y its intergral is also
and arbitrary function of y so let’s call this function C(y) = x fixed A(y)dy. ¤
Example 1.14 By changing variables from (x, y) to (u, v), where u = xy, v = x/y, solve the PDE
∂z ∂z
x +y = 2x2 sin(xy).
∂x ∂y
7
and so, after the change of variable the PDE becomes,
∂z
2u = 2uv sin u,
∂u
i.e.,
∂z
= v sin u.
∂u
Then Z
z= v sin u du = −v cos u + A(v),
v fixed
³ ´
and in terms of x and y this is z = − xy cos(xy) + A x
y , where A is an arbitrary function. ¤
∂f ∂f
Example 1.15 By changing variables from (x, y) to (u, v), where u = x3 /y, v = x, find ∂x and ∂y in terms
of parital derivatives with respect to u and v. Hence, solve the PDE
∂f ∂f 6x5
x + 3y = .
∂x ∂y y
∂f ∂u ∂f ∂v ∂f 3x2 ∂f ∂f
= + = +1 ,
∂x ∂x ∂u ∂x ∂v y ∂u ∂v
and
∂f ∂u ∂f ∂v ∂f −x3 ∂f ∂f
= + = 2 −0 .
∂y ∂y ∂u ∂y ∂v y ∂u ∂v
Therefore, µ ¶ µ ¶
∂f ∂f 3x2 ∂f ∂f −x3 ∂f ∂f
x + 3y =x + +y =x .
∂x ∂y y ∂u ∂v y ∂u ∂v
Inverting the change of variables we have
v3
x = v, y= ,
u
and so, after the change of variable the PDE becomes,
∂f
v = 6uv 2 ,
∂v
i.e.,
∂f
= 6uv.
∂u
Then Z
f= 6uv dv = 3v 2 u + A(u),
u fixed
5
³ ´
x3
and in terms of x and y this is f = 3 xy + A y , where A is an arbitrary function. ¤