1 Derivative of Composite and Implicit Functions

1.1 Partial Derivative of Composite Function

Theorem 1. Let z = f (x, y) and x = g(t), y = h(t) be defined in an appropriate domain and have continuous
FIRST PARTIAL DERIVATIVES. Then the composite function z(t), defined by

z(t) = f [x(t), y(t)] = F (t)

is differentiable and its derivative given by

dz ∂z dx ∂z dy
= + (1)
dt ∂x dt ∂y dt
Proof. Let t be a fixed value and let x, y, z be the corresponding values of the functions g, h and f . Then for
given ∆t (increment of t),∆x and ∆y are determined as

∆x = g(t + ∆t) − g(t)

∆y = h(t + ∆t) − h(t)

while ∆z is then determined as

∆z = f (x + ∆x, y + ∆y) − f (x, y)

By the Fundamental lemma, we have

∂z ∂z
∆z = ∆x + ∆y + α∆x + β∆y
∂x ∂y
Hence
∆z ∂z ∆x ∂z ∆y ∆x ∆y
= + +α +β
∆t ∂x ∆t ∂y ∆t ∆t ∆t
∆y dy
As ∆t approaches zero, ∆x∆ t and ∆ t approaches the derivatives
dx
dt and dt , respectively, while α and β ap-
proaches zero, since ∆x and ∆y approaches zero. Hence
∆z ∂z dx ∂z dy dx dy
lim = + +0 +0
∆t→0 ∆t ∂x dt ∂y dt dt dt
that is 1 as required.

Equation 1 is called the derivative of the composite function of the chain rule.
dz
Example 2. Find if z = f (x, y) = x2 + xy, x = et and y = sin t.
dt
Solution 3. By using 1 and
∂z ∂z dx dy
= 2x + y, = x, = et and = cos t
∂x ∂y dt dt

dz
= (2x + y)et + x cos t
dt
dz
= (2et + sin t)et + et cos t
dt
Note that, the same result can be obtained by the first substituting x = et and y = sin t into z0x2 + xy and then
differentiating

z = e2t + et sin t
dz
= 2e2t + et sin t + et cos t = (2et + sin t)et + et cos t
dt

1
Example 4. a. Use the chain rule to find the derivative of z = f (x, y) = xy w.r.t. t along the path x = cos t,
y = sin t.
π
b. What is the derivative’s value at t = .
2
Solution 5. a.
dz ∂z dx ∂z dy
= +
dt ∂x dt ∂y dt
∂z ∂z dx dy
= y = sin t, = x = cos t, = − sin t, = cos t
∂x ∂y dt dt
dz
= y(− sin t) + x cos t = − sin t sin t + cos t cos t = − sin2 t + cos2 t = cos 2t
dt
In this case we can check the result with a more direct calculation. As a function of t,

1 dz d 1
z = xy = cos t sin t = 2 sin 2t so = ( sin 2t) = cos 2t.
dt dt 2

b. In either case,
dz π
( )t= π2 = cos(2. ) = cos π = −1.
dt 2
d2 z
Example 6. Let z = f (x, y), x = g(t), y = h(t). Evaluate of the second order derivative 2 of the composite
dt
function.

Solution 7. Let x = g(t) = x(t) and y = h(t) = y(t), then z = f (x, y) = f (x(t), y(t)) = F (t) is composite
function. We should use Equation 1.

d2 z d dz d ∂z dx ∂z dy
= ( )= ( + )
dt2 dt dt dt ∂x dt ∂y dt
d ∂z dx ∂z d2 x d ∂z dy ∂z d2 y
= ( ) + + ( ) +
dt ∂x dt ∂x dt2 dt ∂y dt ∂y dt2
d ∂z d ∂z
Applying Equation 1 for ( ) and ( ) we get
dt ∂x dt ∂y

d ∂z ∂ 2 z dx ∂ 2 z dy
( ) = +
dt ∂x ∂x2 dt ∂y∂x dt
d ∂z 2
∂ z dy ∂ 2 z dx
( ) = +
dt ∂y ∂y 2 dt ∂y∂x dt
d2 z 2
∂ z dx ∂ 2 z dy dx ∂z d2 x ∂ 2 z dy ∂ 2 z dx dy ∂z d2 y
= ( 2 + ) + + ( + ) +
dt2 ∂x dt ∂y∂x dt dt ∂x dt2 ∂y 2 dt ∂y∂x dt dt ∂y dt2
d2 z 2
∂ z dx 2 2 2 2
∂ z dx dy ∂ z dy 2 ∂z d x ∂z d y 2
= ( ) +2 + 2( ) + + .
dt2 2
∂x dt ∂x∂y dt dt ∂y dt ∂x dt2 ∂y dt2

Theorem 8. Let z = f (x, y) and x = g(u, v), y = h(u, v) be defined in an appropriate domain and have
continuous first partial derivatives. Then the composite function z(u, v) is defined by

z(u, v) = f [g(u, v) + h(u, v)] = F (u, v)

differentiable and its derivative given by

∂z ∂z ∂x ∂z ∂y ∂z ∂z ∂x ∂z ∂y
= + , = + (2)
∂u ∂x ∂u ∂y ∂u ∂v ∂x ∂v ∂y ∂v

2
Proof. If v be hold fixed, x and y reduce to function of the single variable u and we can apply Theorem 1 at
once, obtaining
∂z ∂z ∂x ∂z ∂y
= + (3)
∂u ∂x ∂u ∂y ∂u
Similarly, if u is hold fixed, x and y reduce to function of the single variable v and we obtain
∂z ∂z ∂x ∂z ∂y
= + (4)
∂v ∂x ∂v ∂y ∂v

Here [in Equation 3 and 4] all ordinary derivatives in

dz ∂z dx ∂z dy
= +
dt ∂x dt ∂y dt
now become partial derivatives. Equation 2 is called the derivative of the composite function of the chain rule.

Example 9. Given the composite function z = f (x, y) = x2 + y 2 , x = u cos v, y = u sin v.

∂z ∂z
• Find , .
∂u ∂v
• Evaluate total(exact) differential of z(u, v).

Solution 10. •
∂z ∂z ∂x ∂z ∂y
= + = 2x cos v + 2y sin v = 2u cos2 v + 2u sin2 v = 2u
∂u ∂x ∂u ∂y ∂u
∂z ∂z ∂x ∂z ∂y
= + = 2x(−u sin v) + 2y(u cos v) = −2u2 cos v sin v + 2u2 cos v sin v = 0
∂v ∂x ∂v ∂y ∂v

•
∂z ∂z
dz = du + dv = 2udu + 0dv = 2udu
∂u ∂v
Note that, the same result can be obtained by first substituting x = u cos v, y = u sin v into z = x2 + y 2 and
then differentiating w.r.t. u and v.

z = u2 cos2 v + u2 sin2 v = u2
∂z ∂z
= 2u, =0
∂u ∂v

1.2 The Chain Rule for the Functions defined on Surfaces

If we are interested in the temperature u = f (x, y, z) at the points (x, y, z) on a globe in space, we might
prefer to think of x, y and z as functions of variables t and s that give the points longitudes and latitudes. If
x = g(s, t), y = h(s, t) and z = k(s, t), we could then express the temperature as a function of s and t with the
composite function

u = f [g(s, t), h(s, t), k(s, t)].

u would have partial derivatives w.r.t. both s and t that could be calculated in the following way.
If u = f (x, y, z), x = g(s, t), y = h(s, t), z = k(s, t) are differentiable, then u is a differentiable function of s and
t and its partial derivatives are given by the equations
∂u ∂u ∂x ∂u ∂y ∂u ∂z
= + +
∂s ∂x ∂s ∂y ∂s ∂z ∂s
∂u ∂u ∂x ∂u ∂y ∂u ∂z
= + +
∂t ∂x ∂t ∂y ∂t ∂z ∂t

3
In general, if z = f (x, y, t, . . . ) and x = g(u, v, w, . . . ), y = h(u, v, w, . . . ), t = p(u, v, w, . . . ), . . . then

∂z ∂z ∂x ∂z ∂y ∂z ∂t
= + + + ...
∂u ∂x ∂u ∂y ∂u ∂t ∂u
∂z ∂z ∂x ∂z ∂y ∂z ∂t
= + + + ...
∂v ∂x ∂v ∂y ∂v ∂t ∂v
∂z ∂z ∂x ∂z ∂y ∂z ∂t
= + + + ...
∂w ∂x ∂w ∂y ∂w ∂t ∂w

The generalization of formula 1 to the case o a function z = f (x1 , x2 , . . . , xn ) whose n arguments depend on a
single variable t, xi = xi (t), i = 1, 2, . . . , n is given by
n
dz ∂z dx1 ∂z dx2 ∂z dxn X ∂z dxi
= + + ··· + = (5)
dt ∂x1 dt ∂x2 dt ∂xn dt ∂xi dt
i=1

The generalization of Equation 2 to the case of a function z = f (x1 , x2 , . . . , xn ) whose n arguments depend on
m new independent variables t1 , . . . , tm , xi = xi (t1 , . . . , tj , . . . , tn ), i = 1, 2, . . . , n is
n
dz ∂z dx1 ∂z dx2 ∂z dxn X ∂z dxi
= + + ··· + = (j = 1, 2, . . . , m) (6)
dtj ∂x1 dtj ∂x2 dtj ∂xn dtj ∂xi dtj
i=1

These rules are known as chain rules and are basic for computation of composite functions. Equations 1,2,5
are coincide statements of the relations between the derivatives involved. In equation 1, z = f [g(t), h(t)] is the
dz dx dy
function of t whose derivative is denoted by , while and stand for g 0 (t) and h0 (t), respectively. The
dt dt dt
∂z ∂z ∂z ∂z
derivatives and , which could b written ( )y and ( )x stand for fx (x, y) and fy (x, y). In equation
∂x ∂y ∂x ∂y
∂z ∂z
2, z = f [g(u, v), h(u, v)] is the function whose derivative w.r.t. u is denoted by , or ( )v . A more prease
∂u ∂u
statement of equation 2 would be as follows:
∂z ∂z ∂x ∂z ∂y
( )v = ( )y ( )v + ( )x ( )v
∂u ∂x ∂u ∂y ∂u
∂z ∂z ∂x ∂z ∂y
( )u = ( )y ( )u + ( )x ( )u
∂v ∂x ∂v ∂y ∂v
dz
Example 11. If z = f (x1 , x2 , x3 ) = 2x21 − x2 x3 + x1 x23 and x1 = 2 sin t, x2 = t2 − t + 1, x3 = 3e−t . Find at
dt
t = 0.

Solution 12. Since x1 = x1 (t) = 2 sin t, x2 = x2 (t) = t2 −t+1, x3 = x3 (t) = 3e−t then z = f [x1 (t), x2 (t), x3 (t)] =
F (t) is a composite function of t and

dz ∂z dx1 ∂z dx2 ∂z dx3

= + +
dt ∂x1 dt ∂x2 dt ∂x3 dt
dz dx1 dx2 dx3
= (4x1 + x23 ) + (−x3 ) + (−x2 + 3x1 x3 )
dt dt dt dt
The parameter value t = 0 corresponds to the point (x1 = 0, x2 = 1, x3 = 0) on the curve (x1 (t), x2 (t), x3 (t))
t=0

x1 = 2 sin t = 2 sin 0 = 0 x1 = 0
2 2
x2 = t − t + 1 = 0 − 0 + 1 = 1 x2 = 1
−t 0
x3 = 3e = 3e = 3 x3 = 3
(x1 , x2 , x3 ) = (0, 1, 3)

4
and
dx1 dx1
= 2 cos t |t=0 = 2 cos t = 2 cos 0 = 2
dt dt
dx2 dx2
= 2t − 1 |t=0 = 2t − 1 = −1
dt dt
dx3 dx3
= −3e−t |t=0 = −3e−t = −3
dt dt
So
dz
= (4.0 + 32 )(2) + (−2)(−1) + (−1 + 3.0.3)(−3) = 18 + 3 + 3 = 24
dt