Sie sind auf Seite 1von 12

Chapter 2

Differentiation of multivariable
functions

2.1 Partial derivatives


In this section we are going to generalize the notion of derivative to the two-
variable case (similarly, one can do it for any number of variables). The
principal difference is that now we have several variables and we can differ-
entiate with respect to any of them.

2.1.1 Definition of partial derivatives


Let f (x, y) be a function and (a, b) a point in the domain of f . We would
like to find the rate of change of f with respect to x at (a, b). Since only the
variable x is changing, we fix the value of y, that is, we have y = b. With y
being fixed, the function f (x, y) becomes a functions of one variable x
f (x, b) = g(x).
Now If g 0 (a) exists, then we call it the partial derivative of f (x, y) with respect
∂f
to x at the point (a, b) and denote it (a, b). Thus
∂x
∂f
(a, b) = g 0 (a).
∂x
g(x) is a one-variable function, hence by definition we have
g(a + h) − g(a) f (a + h, b) − f (a, b)
g 0 (a) = lim = lim .
h→0 h h→0 h

1
CHAPTER 2. DIFF. OF MULTIVARIABLE FUNCTIONS 2

It follows that
∂f f (a + h, b) − f (a, b)
(a, b) = lim
∂x h→0 h
∂f
provided that the limit exists. Essentially, (a, b) is the rate of change of
∂x
the function f (x, y) at the point (a, b) with respect to x when y = b is fixed.
Recall the geometric meaning of the derivative in the one-variable case:
it represents the slope of the tangent line to the graph of the function at the
given point. That is, g 0 (a) is the slope of the tangent line to the graph of
g(x) at the point x = a. Now let us recall how the function g(x) is defined:
it is the function f with y being fixed. The graph of g(x) is a curve on the
graph of f (x, y). This curve is the cross-section of Graph(f ) by the plane
y = b.

Figure 2.1.1: The graph of g as the cross-section of Graph(f ) by the plane


y=b

In the plane y = b we have the following picture

z
z = g(x)

x
0

which is the curve z = g(x). The slope of this curve at x = a is equal to


∂f
(a, b).
∂x
CHAPTER 2. DIFF. OF MULTIVARIABLE FUNCTIONS 3

z
z = g(x)

(a, f (a, b))


x
0

∂f
Similarly, we can define (a, b). Fix x = a in the function f (x, y) and
∂y
obtain a function of y only

f (a, y) = h(y).

Again, If h0 (a) exists, then we call it the partial derivative of f (x, y) with
∂f
respect to y at the point (a, b) and denote it (a, b). Thus
∂y
∂f
(a, b) = h0 (a)
∂y
and we have
∂f f (a, b + h) − f (a, b)
(a, b) = lim ,
∂y h→0 h
∂f
provided that the limit exists. (a, b) is the rate of change of the function
∂y
f (x, y) at the point (a, b) with respect to y when x = a is fixed.

Figure 2.1.2: The graph of h as the cross-section of Graph(f ) by the plane


x=a
CHAPTER 2. DIFF. OF MULTIVARIABLE FUNCTIONS 4

∂f
The geometric meaning of (a, b) is similar to the meaning of the partial
∂y
derivative by x: the curve z = h(y) is the cross-section of Graph(f ) by the
∂f
plane x = a, hence, the slope of this curve at y = b is equal to (a, b).
∂y
z
z = h(y)

(b, f (a, b))

y
0

∂f ∂f
Notice that the limit definitions of (a, b) and (a, b) given above can
∂x ∂y
be generalized to any point (x, y) in the domain of f . That is, one can define
∂f f (x + h, y) − f (x, y)
(x, y) = lim
∂x h→0 h
and
∂f f (x, y + h) − f (x, y)
(x, y) = lim
∂y h→0 h
∂f
simply by replacing a by x and b by y in the initial definitions. Thus, (x, y)
∂x
∂f
and (x, y) can be viewed as new two-variable functions whose domains
∂y
consist of all pairs (x, y) for which the corresponding limits exist.
Finally, a few words about the notation. Actually, the notation for partial
derivatives varies for different textbooks and you are supposed to be familiar
with the most common ways to denote them: if z = f (x, y) then
∂f ∂ ∂f ∂z
(x, y) = f (x, y) = = = fx (x, y) = Dx f,
∂x ∂x ∂x ∂x
∂f ∂ ∂f ∂z
(x, y) = f (x, y) = = = fy (x, y) = Dy f.
∂y ∂y ∂y ∂y
CHAPTER 2. DIFF. OF MULTIVARIABLE FUNCTIONS 5

2.1.2 How to calculate partial derivatives


First of all, using the limit definitions given above, one can estimate the
values of partial derivatives.
Recall that
∂f f (a + h, b) − f (a, b)
(a, b) = lim
∂x h→0 h
and
∂f f (a, b + h) − f (a, b)
(a, b) = lim .
∂y h→0 h
Hence, for values of h close to 0 we can write approximate equalities as follows
∂f f (a + h, b) − f (a, b)
(a, b) ≈
∂x h
and
∂f f (a, b + h) − f (a, b)
(a, b) ≈ .
∂y h
Alternatively, if a1 ≈ a and b1 ≈ b then
∂f f (a1 , b) − f (a, b)
(a, b) ≈
∂x a1 − a
and
∂f f (a, b1 ) − f (a, b)
(a, b) ≈ .
∂y b1 − b
Example 2.1.1. Suppose the function f (x, y) is given by the table below.
x\y 1 2 3
0 2 −1 −2
1 0 1 2
2 1 0 3
∂f ∂f
Estimate (1, 2) and (1, 2).
∂x ∂y
∂f
Solution. Estimate (1, 2) first. Have a look at the provided values of x:
∂x
notice that there are two values of x “close” to 1 (that’s the best we have
based on the given data). Namely, we can choose a1 = 0, or a1 = 2. Hence,
we obtain
∂f f (0, 2) − f (1, 2) −1 − 1
(1, 2) ≈ = =2
∂x 0−1 −1
CHAPTER 2. DIFF. OF MULTIVARIABLE FUNCTIONS 6

for a1 = 0, and
∂f f (2, 2) − f (1, 2) 0−1
(1, 2) ≈ = = −1
∂x 2−1 1
∂f
for a1 = 0. Both choices give legitimate estimates of (1, 2).
∂x
∂f
Now we do the same for (1, 2): this time we have again two values of
∂y
y close to 2. We obtain
∂f f (1, 1) − f (1, 2) 0−1
(1, 2) ≈ = =1
∂y 1−2 −1
for b1 = 1, and
∂f f (1, 3) − f (1, 2) 2−1
(1, 2) ≈ = =1
∂y 3−2 1
for b1 = 1. Notice that in this case the estimates coincide, but it is not what
happens in general.
One can compute partial derivatives precisely using differentiation tech-
niques from Calculus I without using the limit definitions given in the previ-
ous sub-section. Suppose f (x, y) is a function. Hence,
∂f
• in order to compute , consider y as a constant and differentiate
∂x
f (x, y) with respect to x,
∂f
• in order to compute , consider x as a constant and differentiate
∂y
f (x, y) with respect to x,
∂f ∂f
Example 2.1.2. Calculate and if
∂x ∂y

f (x, y) = x3 y 2 + ln(x2 + y 2 ).

Solution. Using the rules outlined above we obtain


∂f ∂ 3 1 ∂ 2
= y2 · (x ) + 2 2
· (x + y 2 )
∂x ∂x x + y ∂x
CHAPTER 2. DIFF. OF MULTIVARIABLE FUNCTIONS 7

2x
= y 2 · (3x2 ) +
x2 + y 2
2x
= 3x2 y 2 + 2
x + y2
treating y as a constant and applying the Chain rule, and
∂f ∂ 2 1 ∂ 2
= x3 · (y ) + 2 2
· (x + y 2 )
∂y ∂y x + y ∂y
2y
= x3 · (2y) +
x2
+ y2
2y
= 2x3 y + 2
x + y2
now treating x as a constant and again applying the Chain rule.
∂f ∂f
Example 2.1.3. Calculate (0, π) and (0, π) if
∂x ∂y

f (x, y) = (x2 + y) sin(x − y).

Solution. We obtain
∂f ∂ 2 ∂
= (x + y) · sin(x − y) + (x2 + y) · (sin(x − y))
∂x ∂x ∂x

= 2x sin(x − y) + (x2 + y) cos(x − y) · (x − y)
∂x
= 2x sin(x − y) + (x2 + y) cos(x − y)
treating y as a constant and applying the product and the Chain rules, and
∂f ∂ 2 ∂
= (x + y) · sin(x − y) + (x2 + y) · (sin(x − y))
∂y ∂y ∂y

= 1 · sin(x − y) + (x2 + y) cos(x − y) · (x − y)
∂y
= sin(x − y) − (x2 + y) cos(x − y)
now treating x as a constant and again applying the product and the Chain
rules.
CHAPTER 2. DIFF. OF MULTIVARIABLE FUNCTIONS 8

Finally we evaluate the obtain functions at the point (0, π):

∂f
(0, π) = 0 · sin(0 − π) + (0 + π) · cos(0 − π) = −π,
∂x
∂f
(0, π) = sin(0 − π) − (0 + π) · cos(0 − π) = π.
∂y

2.1.3 Implicit differentiation in the multi-variable case


Recall from Calculus I that the equation G(x, y) = 0 defines a function
y = g(x) implicitly near the point (x0 , y0 ) if

• G(x0 , y0 ) = 0, that is, the point (x0 , y0 ) belongs to the curve defined
by the equation G(x, y) = 0,

• y0 = g(x0 ) which means that the point (x0 , y0 ) belongs to the graph of
g,

• G(x, g(x)) = 0 is an identity which holds for every x close to x0 - that


is not only (x0 , y0 ), but all points of the graph of g close to (x0 , y0 )
belong to the curve defined by G(x, y) = 0.

A very similar definition can be given in the two-variable case (and in


higher dimensions).
An equation F (x, y, z) = 0 in the variables x, y, and z defines a surface in
the 3D-space. This surface in general is not the graph of any function (it may
not pass the vertical line test, where vertical means “along the z-axis”), but
small pieces of the surface may be considered as the graphs of two-variable
functions. More precisely, we say that the equation F (x, y, z) = 0 defines a
function z = f (x, y) implicitly near a point (x0 , y0 , z0 ) if

1. F (x0 , y0 , z0 ) = 0, that is, the point (x0 , y0 , z0 ) belongs to the surface


defined by the equation F (x, y, z) = 0,

2. z0 = f (x0 , y0 ) which means that the point (x0 , y0 , z0 ) belongs to the


graph of f ,

3. F (x, y, f (x, y)) = 0 is an identity which holds for every (x, y) close to
(x0 , y0 ) - that is not only (x0 , y0 , z0 ), but all points of the graph of f
close to (x0 , y0 , z0 ) belong to the surface defined by F (x, y, z) = 0.
CHAPTER 2. DIFF. OF MULTIVARIABLE FUNCTIONS 9

Now, suppose a function z = f (x, y) is implicitly defined by the equation


∂f ∂f
F (x, y, z) = 0. How to compute and ? The method is again very
∂x ∂y
∂f
similar to what we do in the one-variable case: in order to find we
∂x
differentiate the equation F (x, y, z) = 0 with respect to x treating y as a
∂f
constant and z as a function of x and y. Similarly we find .
∂y
∂f ∂f
Example 2.1.4. Calculate and if f (x, y) is implicitly defined by the
∂x ∂y
equation
x3 + y 3 + z 3 + 6xyz = 1

Solution. Apply the rule for finding partial derivatives keeping in mind that
∂f
z = f (x, y). At first we differentiate with respect to x to find .
∂x
∂ 3 ∂
(x + y 3 + z 3 + 6xyz) = (1)
∂x ∂x
∂ 3 ∂ 3 ∂ 3 ∂
(x ) + (y ) + (z ) + (6xyz) = 0
∂x ∂x ∂x ∂x
 
2 2 ∂z ∂ ∂z
3x + 0 + 3z · + 6y (x) · z + x · =0
∂x ∂x ∂x
 
2 2 ∂z ∂z
3x + 3z · + 6y z + x · =0
∂x ∂x
∂f ∂z
Finally, = and we express it from the obtained equation:
∂x ∂x
∂f x2 + 2yz
=− 2 .
∂x z + 2xy
Similarly, we differentiate with respect to y.
∂ 3 ∂
(x + y 3 + z 3 + 6xyz) = (1)
∂y ∂y
∂ 3 ∂ 3 ∂ 3 ∂
(x ) + (y ) + (z ) + (6xyz) = 0
∂y ∂y ∂y ∂y
CHAPTER 2. DIFF. OF MULTIVARIABLE FUNCTIONS 10
 
2 2 ∂z ∂ ∂z
0 + 3y + 3z · + 6x (y) · z + y · =0
∂y ∂y ∂y
 
2 2 ∂z ∂z
3y + 3z · + 6x z + y · =0
∂y ∂y
and, eventually
∂f ∂z y 2 + 2xz
= =− 2 .
∂y ∂y z + 2xy
We are going to return to the topic of implicit differentiation in later
sections.
CHAPTER 2. DIFF. OF MULTIVARIABLE FUNCTIONS 11

2.1.4 Exercises
1. Calculate the partial derivatives of the following functions using the
limit definitions.
∂f ∂f
(a) f (x, y) = xy 2 , find (2, 1) and (2, 1).
∂x ∂y
p ∂g ∂g
(b) g(x, y) = x2 + y 2 , find (3, 4) and (3, 4).
∂x ∂y

2. Suppose the function f (x, y) is given by the table below.

x\y −1 0 1
2 −1 −2 −2
3 0 −1 −3
4 1 0 1

∂f ∂f
Estimate (3, 0) and (3, 0).
∂x ∂y

3. Find the partial derivatives of the given functions.

(a) f (x, y) = x2 y 2 − xy + 1
(b) g(x, y) = (2x − y)9
p
(c) h(x, y) = xy x2 + y 2
y
(d) k(x, y) = x+y

(e) m(u, v) = arctan(uv)

∂f ∂f
4. Using implicit differentiation find and .
∂x ∂y
(a) x2 + y 2 + z 2 = 1
(b) ez = xyz
(c) xz + y ln x = z 3
CHAPTER 2. DIFF. OF MULTIVARIABLE FUNCTIONS 12

Answers:
∂f ∂f
1. (a) (2, 1) = 1, (2, 1) = 4.
∂x ∂y
∂g 3 ∂g 4
(b) (3, 4) = , (3, 4) = .
∂x 5 ∂y 5

∂f ∂f
2. (3, 0) ≈ 1 using a1 = 2 and (3, 0) ≈ 1 using a1 = 4;
∂x ∂x
∂f ∂f
(3, 0) ≈ −1 using b1 = −1 and (3, 0) ≈ −2 using b1 = 1.
∂y ∂y

∂f ∂f
3. (a) = 2xy 2 − y, = 2x2 y − x.
∂x ∂y
∂g ∂g
(b) = 18(2x − y)8 , = −9(2x − y)8 .
∂x ∂y
∂h p x2 y ∂h p xy 2
(c) = y x2 + y 2 + p , = x x2 + y 2 + p .
∂x x2 + y 2 ∂y x2 + y 2
∂k −y ∂k x
(d) = 2
, = .
∂x (x + y) ∂y (x + y)2
∂m v ∂m u
(e) = 2
, = .
∂u 1 + (uv) ∂v 1 + (uv)2

∂z x ∂z y
4. (a) =− , =− .
∂x z ∂y z
∂z yz ∂z xz
(b) = z , = z .
∂x e − xy ∂y e − xy
∂z xz + y ∂z ln x
(c) = 2
, = 2 .
∂x x(3z − x) ∂y 3z − x

Das könnte Ihnen auch gefallen