diff. of functions

CHAPTER 16
Differentiable Functions of Several Variables
Ü16.1. The Differential and Partial Derivatives
Let w f´x y zµ be a function of the three variables x y z. In this chapter we shall explore how to
evaluate the change in w near a point ´x0 y0 z0µ, and make use of that evaluation. For functions of one
variable, this led to the derivative: dw dx is the rate of change of w with respect to x. But in more than
one variable, the lack of a unique independent variable makes this more complicated. In particular, the
rates of change may differ, depending upon the direction in which we move. We start by using the one
variable theory to define change in w with respect to one variable at a time.
Definition 16.1 Suppose we are given a function w f´x y zµ. The partial derivative of f with respect
to x is defined by differentiating f with respect to x, considering y and z as being held constant. That is,
at a point ´x0 y0 z0µ, the value of the partial derivative with respect to x is
(16.1)
∂ f
∂x
´x0 y0 z0µ
d
dx
f´x y0 z0µ lim
h 0
f´x0 · h y0 z0µ f´x0 y0 z0µ
h
Similarly, if we keep x and z constant, we define the partial derivative of f with respect to y by
(16.2)
∂ f
∂y
d
dy
f´x0 y z0µ
and by keeping x and y constant, we define the partial derivative of f with respect to z by
(16.3)
∂ f
∂z
d
dz
f´x0 y0 zµ
Example 16.1 Find the partial derivatives of f´x yµ x´1 · xyµ2.
Thinking of y as a constant, we have
(16.4)
∂ f
∂x
´1 · xyµ
2
· x´2´1 · xyµyµ ´1 · xyµ´1 · 3xyµ
235

Chapter 16 Differentiable Functions of Several Variables 236
Now, we think of x as constant and differentiate with respect to y:
(16.5)
∂ f
∂y
x´2´1 · xyµxµ 2x2
´1 · xyµ
Example 16.2 The partial derivatives of f´x y zµ xyz are
(16.6)
∂ f
∂x
yz
∂ f
∂y
xz
∂ f
∂z
zy
Of course, the partial derivatives are themselves functions, and when it is possible to differentiate the
partial derivatives, we do so, obtaining higher order derivatives. More precisely, the partial derivatives are
found by differentiating the formula for f with respect to the relevant variable, treating the other variable
as a constant. Apply this procedure to the functions so obtained to get the second partial derivatives:
(16.7)
∂2 f
∂x2
∂
∂x
´
∂ f
∂x
µ
∂2 f
∂y∂x
∂
∂y
´
∂ f
∂x
µ
∂2 f
∂x∂y
∂
∂x
´
∂ f
∂y
µ
∂2 f
∂y2
∂
∂y
´
∂ f
∂y
µ
Example 16.3 Calculate the second partial derivatives of the function in example 1.
We have f´x yµ x´1 · xyµ2
, and have found
(16.8)
∂ f
∂x
´1 · xyµ´1 · 3xyµ
∂ f
∂y
2x2
´1 · xyµ
Differentiating these expressions, we obtain
(16.9)
∂2 f
∂x2
´1 · xyµ´3yµ · y´1 · 3xyµ 4y· 6xy2
(16.10)
∂2
f
∂y∂x
´1 · xyµ´3xµ · x´1 · 3xyµ 4x· 6x2
y
(16.11)
∂2 f
∂x∂y
4x´1 · xyµ · 2x2
y 4x· 6x2
y
(16.12)
∂2 f
∂y2
2x2
´xµ 2x3
Notice that the second and third lines are equal. This is a general fact: the mixed partials (the middle
terms above) are equal when the second partials are continuous:
(16.13)
∂2 f
∂y∂x
∂2 f
∂x∂y

Ü16.1 The Differential and Partial Derivatives 237
This is not easily proven, but is easily verified by many examples. Thus ∂2 f ∂x∂y can be calculated in
whatever is the most convenient order. Finally, we note an alternative notation for partial derivatives;
(16.14) fx
∂ f
∂x
fy
∂ f
∂y
fxx
∂2 f
∂x2
fxy
∂2 f
∂x∂y
fyy
∂2 f
∂y2
etc
Example 16.4 Let f´x yµ ytanx· xsecy. Show that fxy fyx.
We calculate the first partial derivatives and then the mixed partials in both orders:
(16.15) fx ysec2
x· secy fy tanx· xsecytany
(16.16) fyx sec2
x· secytany fxy sec2
x· secytany
The partial derivatives of a function w f´x y zµ tell us the rates of change of w in the coordinate
directions. But there are many directions at a point on the plane or in space: how do we find these
rates in other directions? More generally, if two or three variables are changing, how do we explore the
corresponding change in w? The answer to these questions starts with the generalization of the idea of the
differential as linear approximation. For a function of one variable, a function w f´xµ is differentiable
if it is can be locally approximated by a linear function
(16.17) w w0 · m´x x0µ
or, what is the same, the graph of w f´xµ at a point ´x0 y0µ is more and more like a straight line, the
closer we look. The line is determined by its slope m f¼´x0µ. For functions of more than one variable,
the idea is the same, but takes a little more explanation and notation.
Definition 16.2 Let w f´x y zµ be a function defined near the point ´x0 y0 z0µ. We say that f is
differentiable if it can be well- approximated near ´x0 y0 z0µ by a linear function
(16.18) w w0 a´x x0µ · bý y0µ · c´z z0µ
In this case, we call the linear function the differential of f at ´x0 y0 z0µ, denoted d f´´x0 y0 z0µ. It
is important to keep in mind that the differential is a function of a vector at the point; that is, of the
increments ´x x0 y y0 z z0µ.
If f´x yµ is a function of two variables, we can consider the graph of the function as the set of
points ´x y zµ such that z f´x yµ. To say that f is differentiable is to say that this graph is more and
more like a plane, the closer we look. This plane, called the tangent plane to the graph, is the graph of
the approximating linear function, the differential. For a precise definition of what we mean by “well”
approximated, see the discussion in section 16.3. The following example illustrates this meaning.
Example 16.5 Let f´x yµ x2 · y. Find the differential of f at the point (1,3). Find the equation of the
tangent plane to the graph of z f´xµ at the point.
We have ´x0 y0µ ´1 3µ, and z0 f´x0 y0µ 4. Express z 4 in terms of x 1 and y 3:
(16.19) z 4 x2
· y 4 ´1 · ´x 1µµ
2
· ´3 · ý 3µµ 4

(16.20) 1 · 2´x 1µ · ´x 1µ
2
· 3 · ý 3µ simplifing to
(16.21) z 4 2´x 1µ · y 3 · ´x 1µ
2
Comparing with (16.18), the first two terms give the differential. ´x 1µ2 is the error in the approxima-
tion. The equation of the tangent plane is
(16.22) z 4 2´x 1µ · y 3 or z 2x· y 1
If we just follow the function along the line where y y0 z z0, then (16.18) becomes just w w0
a´x x0µ; comparing this with definition 16.1, we see that a is the derivative of w in the x-direction, that
is a ∂w ∂x. Similarly b ∂w ∂y and c ∂w ∂z. Finally, since the variables x y z are themselves
linear, we have that dx is x x0,and so forth. This leads to the following restatement of the definition of
differentiability:
Proposition 16.1 Suppose that w f´x y zµ is differentiable at ´x0 y0 z0µ. Then
(16.23) dw
∂ f
∂x
dx·
∂ f
∂y
dy·
∂ f
∂z
dz
There are a variety of ways to use formula (16.23), which we now illustrate.
Example 16.6 Let
(16.24) z f´x yµ x2
xy· y3
Find the equation of the tangent plane to the graph at the point (2,-1).
At ´x0 y0µ ´2 1µ, we have z0 f´x0 y0µ 6. We calculate
(16.25)
∂ f
∂x
2x y
∂ f
∂y
x· 3y2
so, at (2,-1), ∂ f ∂x 5 ∂ f ∂y 1. Substituting these values in (16.18) we obtain
(16.26) z 6 5´x 2µ · ý· 1µ or z 5x· y 3
An alternative approach is to differentiate equation (16.24) implicitly:
(16.27) dz 2xdx xdy ydx· 3y2
dy
Evaluating at (2,-1), we have z0 6, and dz 4dx 2dy· dx· 3dy. This is the equation of the tangent
plane, with the differentials dx dy dz replaced by the increments x 2 y· 1 z 5:
(16.28) z 6 4´x 2µ 2ý· 1µ · ´x 2µ · 3ý· 1µ
which is the same as (16.26).
Example 16.7 Find the equation of the tangent plane to the graph of the function z x2 · xy y at
(2,-1, 1).

First, we calculate the differential
(16.29) dz 2xdx· xdy· ydx dy
and then evaluate it at the point:
(16.30) dz 4dx· 2dy dx dy 3dx· dy
We now get the equation of the tangent plane by replacing the differentials by the increments:
(16.31) z 1 3´x 2µ · ý· 1µ or z 3x· y 4
Example 16.8 Find the points at which the graph of z f´x yµ x2 2xy· y has a horizontal tangent
plane.
The horizontal plane through the point ´x0 y0 z0µ has the equation z z0 0. Thus our points are
those where d f 0; i.e., solutions of the pair of equations
(16.32)
∂ f
∂x
0
∂ f
∂y
0
Calculating, we get 2x 2y 0 2x· 1 0, so x 1 2 y 1 2 and our point is ´1 2 1 2µ.
Example 16.9 Given the function z x2 xy· y3, in what direction, at the point (1,1,1) is the rate of
change of z equal to zero?
The differential of z is dz ´2x yµdx· ´ x· 3y2µdy, so at (1,1,1), we have dz dx· 2dy. This is
zero for the direction in which dx 2dy; that is along the line of slope -1/2. Thus the answer is given
by a vector in that direction, for example: 2I· J.
Example 16.10 Suppose that we have designed a cylindrical silo of base radius 6 meters and height 10
meters, and we are asked to increase the radius by .25 m and the height by .2 m. By (approximately) how
much do we increase the volume?
The volume of a cylinder of radius r and height h is V πr2h. To answer this question, we consider
the linear approximation of volume, so we take the differential of V:
(16.33) dV 2πrhdr · πr2
dh
Now, in our case r 5 h 10 dr 25 dh 2, so we calculate
(16.34) dV 2π´5µ´10µ´ 25µ · π´5µ
2
´ 2µ π´25 · 5µ 30π cubic meters
By looking at figure 1, we can identify the two terms in the increment of volume: the first is the volume
of the shell of width dr around the cylinder, and the second is the volume of the cap of height dh. The
negligible part is the volume 2πdrdh of the washer at the top of width dr and height dh.

Figure 16.1
dh
h
r dr
Proposition 16.2 (The Chain Rule). Let w f´x y zµ be a function deﬁned in a region R in space. Sup-
pose that γ is a curve in R given parametrically by x x´tµ y y´tµ z z´tµ, with t 0 corresponding
to ´x0 y0 z0µ. Then, considering w f´x´tµ y´tµ z´tµµ as a function of t along γ, we have
(16.35)
dw
dt
∂w
∂x
dx
dt
·
∂w
∂y
dy
dt
·
∂w
∂z
dz
dt
That is, the rate of change of w with respect to t along γ is given by (16.35). We shall give an
explanation of this formula in section 3.
Example 16.11 Let w f´x y zµ xy · y2
z. Consider the curve given parametrically by x t y
t2 z lnt. Find dw dt at t 2.
Differentiating,
(16.36)
∂w
∂x
y
∂w
∂y
x· 2yz
∂w
∂z
y2
(16.37)
dx
dt
1
dy
dt
2t
dz
dt
1
t
so
(16.38)
dw
dt
y´1µ · ´x· 2yzµ´2tµ · y2
´
1
t
µ
At t 2 we calculate x 2 y 4 and z ln2, giving
(16.39)
dw
dt
4 · ´2 · 8ln2µ´4µ · 16 2 20 · 32ln2
Example 16.12 Let z x2 · y2. Find the maximum value of z on the ellipse given parametrically by
x cost y 2sint.

We need to find the points at which dz dt 0. Now
(16.40)
∂z
∂x
2x
∂z
∂y
2y
dx
dt
sint
dy
dt
2cost
and thus
(16.41)
dz
dt
2xsint · 4ycost
Since x cost and y 2sint, this gives dz dt 2sint cost · 8sint cost Set this equal to zero to
obtain 6sint cost 0, or 3sin´2tµ 0, which has the solutions t 0 ¦π 2 π, and x ¦1 y 0 or
x 0 y ¦2. The corresponding values of z are thus 1 and 4, so 1 is the minimum value and 4 is the
maximum value on the ellipse.
We can think of an equation of the form f´x y zµ 0 as defining z implicitly as a function of x and y,
in the sense that we could solve for z, given specific values of x and y. However, just as in one dimension,
we need not solve for z to find the partial derivatives. If we take the differential of the defining equation
f´x y zµ 0 we get
(16.42) fxdx· fydy· fzdz 0 so that dz fx
fz
dx fy
fz
dy
The coefficient of dx is thus ∂z ∂x, and the coefficient of dy is ∂z ∂y. Of course if fz 0, these are not
defined. But if fz 0, then this method works.
Proposition 16.3 Suppose that f is a differentiable function of ´x y zµ near the point ´x0 y0 z0µ, and
that fz´´x0 y0 z0µ 0. Then the equation f´x y zµ 0 defines z implicitly as a function of x y and
(16.43)
∂z
∂x
fx
fz
and
∂z
∂x
fy
fz
Example 16.13 Given f´x y zµ z3 · 3xz2 · y2z, find expressions for ∂z ∂x and ∂z ∂y where z is
defined implicitly as a function of ´x yµ by the equation f´x y zµ 5. Evaluate these at the point (1,1,1).
First we calculate the partial derivatives:
(16.44)
∂ f
∂x
3z2 ∂ f
∂y
2yz
∂ f
∂z
3z2
· 6zx· y2
so that
(16.45)
∂z
∂x
3z2
3z2 · 6zx· y2
∂z
∂y
2yz
3z2 · 6zx· y2
The values at (1,1,1) are ∂z ∂x 3 10 ∂z ∂x 1 10.

Ü16.2. Gradients and Vector methods
Let w f´x y zµ, where f is a differentiable function. To put the formula for the differential,(16.23), in
vector form, we introduce the gradient of the function f:
(16.46) ∇f
∂ f
∂x
I·
∂ f
∂y
J ·
∂ f
∂z
K
and the vector differential dX dxI · dyJ · dzK. We interpret dX as a small change in the vector X.
Then (16.17) can be rewritten as
(16.47) dw
∂ f
∂x
dx·
∂ f
∂y
dy·
∂ f
∂z
dz ´∇fµ ¡dX
This leads to the following vectorial form of the chain rule.
Proposition 16.4 (The Gradient Form of the Chain Rule). Let w f´x y zµ be a function deﬁned
in a region R in space. Suppose that γ is a curve in R given parametrically by X X´tµ, with t 0
corresponding to X0. Then, considering w f´X´tµµ as a function of t along γ, we have
(16.48)
dw
dt
´∇fµ ¡ dX
dt
evaluated at X0.
The partial derivatives tell us the rate of change of the function f in the coordinate directions. Using
the gradient, we can calculate the rate of change in any direction.
Deﬁnition 16.3 Let w f´x y zµ be differentiable in a neighborhood of X0. For any vector V, let
X´tµ X0 · tV parametrize the line through X0 in the direction V. The derivative of f along V is
(16.49) DV f´X0µ lim
t 0
f´X0 · tVµ f´X0µ
t
Propostion 16.5. Given the differentiable function f and a vector V, we have
(16.50) DV f´X0µ ∇f ¡V
The right hand side of (16.49) is the derivative of f along the line in the direction of V. That line is
parametrized by X´tµ X0 · tV, so dX dt V. Now, by the chain rule
(16.51) DV f´X0µ
d
dt
f´X´tµµ
∂ f
∂x
dx
dt
·
∂ f
∂y
dy
dt
·
∂ f
∂z
dz
dt
∇f ¡ dX
dt
∇f ¡V
If we replace V by a unit vector U, then the parameter t represents distance along the line, since X´tµ
X0 t U t. We say that the line is parametrized by arc length, and refer to DU f as the directional
derivative of f in the direction U.
Example 16.14 Let f´x yµ x3 3x2 · xy· 7 and U 0 6I 0 8J. Find DU f´1 2µ.
We have fx 3x2 6x· y fy x. Evaluating at (1,-2), we have ∇f´1 2µ 5I· J. Thus
(16.52) DU f´1 2µ ∇f´1 2µ ¡U 3 0 8 3 8

Ü16.2 Gradients and Vector methods 243
Example 16.15 For f as above, find the direction U at (1,-2) in which DU f 0.
Let U aI· bJ. We must solve
(16.53) ∇f´1 2µ ¡U ´ 5I· J ¡áI· bJµ 5a · b 0
This gives b 5a. Since U is a unit vector, we have a2 · b2 26a2 1, so a 1
Ô
26 b 5
Ô
26 will
do Thus
(16.54) U
I· 5J
Ô
26
We also have the answer U. Notice that both these vectors are unit vectors in the direction of ∇f .
Example 16.16 Let γ be parametrized by X´tµ t2I· lntJ·tK, and let w f´x yµ xyz. Find dw dt
along γ. What is the rate of change of w with respect to t at the point t 2?
To use (16.48), we calculate
(16.55) ∇f yzI· xzJ · xyK
dX
dt
2tI·
1
t
J · K
so that
(16.56)
dw
dt
´∇fµ ¡ dX
dt
2tyz·
xz
t
· xy 3t2
´lnt · 1µ
since x t2, y lnt and z t. At t 2, we get dz dt 12´ln2 · 1µ.
Example 16.17 Let X´tµ costI · sintJ parametrize the unit circle, and let f´x yµ x2 · 2xy. Find
the maximum value of f on the unit circle.
The function z f´X´tµµ has a maximum when dz dt 0. We calculate:
(16.57) ∇f ´2x· 2yµI· 2xJ 2´ćost · sintµI· 2costJ
(16.58)
dX
dt
sintI· costJ
so that
(16.59)
dz
dt
´∇fµ ¡ dX
dt
2´ćost · sintµ´ sintµ · 2cos2
t
To solve dz dt 0 we use double angle formulas:
(16.60) 2´ćost · sintµ´ sintµ · 2cos2
t 2cost sint · 2ćos2
t sin2
tµ sin´2tµ · 2cos´2tµ
which is zero when tan´2tµ 2, or t 31 7Æ 211 7Æ. The corresponding values of x cost y sint are
x ¦ 526 y ¦ 851. Calculating the values of z at these points gives the maximum 1.172.
For a function w f´x y zµ of three variables defined near the point X0 : ´x0 y0 z0µ, let w0
f´x0 y0 z0µ. The equation w w0 is the level surface S of w at ´x0 y0 z0µ. For f differentiable at a point

X0, the fact that f can be approximated by a linear function implies that the surface S looks more and
more like a plane, the closer we look. This plane, given by the equation d f´X0µ 0, is the tangent plane
to S at X0. We now note that the gradient of f is the normal to this surface, and points in the direction of
maximum increase of f.
Proposition 16.5 Let f be a function differentiable in a neighborhood of the point X0.
a) ∇f´X0µ points in the direction of maximum increase of the function f at X0.
b) ∇f´X0µ is the normal to the tangent plane of the level set of f through X0.
To show a), start with a unit vector U. From (16.50) we have
(16.61) DU´X0µ ∇f ¡U ∇f cosβ
where β is the angle between ∇f and U (since U 1). This takes its greatest value when cosβ 0,
that is U ∇f. To show b), let V be a vector on the tangent plane. By definition, d f´X0µ´Vµ 0, so
(16.62) ∇f´X0µ ¡V d f´X0µ´Vµ 0
Thus ∇f´X0µ is orthogonal to every vector in the tangent plane, so can be taken to be its normal. Now, a
point X lies in the tangent plane if and only if the vector X X0 lies on the tangent plane, or
(16.63) ∇f´X0µ ¡´X X0µ 0
which is thus the equation of the tangent plane.
Example 16.18 Let f´x yµ x3 · 3x2y2 · 2y. Find the equation of the line tangent to the curve
f´x yµ 9 at (2,-1).
The above discussion for three dimensions holds just as well in two dimensions. Thus, by proposition
16.6, the normal to the tangent line to the curve is ∇f.We calculate ∇f ´3x2 · 6xy2µI · ´6x2y · 2µJ;
which at x 2 y 1 is the vector 24I 22J. Now, the equation of the tangent line is given by (16.63),
where X0 2I J is the vector to the point (2,-1):
(16.64) ´24I 22Jµ ¡´´x 2µI· ý· 1µJµ 0 or 24´x 2µ 22ý· 1µ 0
which simplifies to 24x 22y 70.
Example 16.19 Let f´x y zµ xyz. Find the gradient of f. Find the equation of the tangent plane to
the level surface f´x y zµ 2 at the point X0 : ´1 2 1µ.
We calculate:
(16.65) ∇f
∂ f
∂x
I·
∂ f
∂y
J ·
∂ f
∂z
K yzI· xzJ · xyK
At X0, ∇f 2I· J · 2K, so the equation of the tangent plane is ∇f ¡´X X0µ 0:
(16.66) 2´x 1µ · ý 2µ · 2´z 1µ 0 or 2x· y· 2z 5
Example 16.20 Let w x · xy yz2. Find the equation of the tangent plane to the surface w 2 at
(3,1,2).
We calculate ∇w ´1 · yµI· ´x z2µJ 2zyK. At the given point (3,1,2), ∇w 2I J 6K. This
is the normal to the tangent plane, at X0 3I· J · 2K, so the equation of that plane is
(16.67) ∇w¡´X X0µ 2´x 3µ ý 1µ 6´z 2µ 0

Ü16.2 Gradients and Vector methods 245
or 2x y 6z· 7 0.
Example 16.21 Let S be the sphere x2 · y2 · z2 a2 a 0. Show that at any point X on the sphere,
the vector X is orthogonal to the sphere.
Let w x2
· y2
· z2
, so that S is the level set w a2
. Then ∇w is normal to S at X xI· yJ · zK. But
(16.68) ∇w 2xI· 2yJ · 2zK 2X
Example 16.22 Let S1 be the sphere x2 · y2 · z2 4 and S2 the cylinder x2 · y2 1. Let X
xI· yJ · zK be a point on the curve γ of intersection of the surfaces S1 and S2. Find a vector tangent to
γ at X.
Let w1 x2 · y2 · z2 w2 x2 · y2, so that γ is the intersection of the level sets w1 4 and w2 1.
Then ∇w1 and ∇w2 are both orthogonal to the tangent to γ, so ∇w1 ¢∇w2 points in the direction of the
tangent to γ. We calculate:
(16.69) ∇w1 ¢∇w2 2´xI· yJ · zKµ ¢2´xI· yJµ 4´ yzI· zxJµ
In the above, we have considered a surface as a graph or as a level set of a function. Surfaces
can also be given parametrically. Let u and v be the variables in a region R of the plane, and let
Xú vµ xú vµI · yú vµJ · zú vµK be a vector-valued function on R. Then the set of values of
Xú vµ, as ú vµ ranges over R describes a surface in space.
Example 16.23 Consider the function Xú vµ ú vµI · ú · vµJ · uvK defined in ú vµ space. In
coordinates, this is given by the equations
(16.70) x u v y u · v z uv
We can solve for u and v in terms of x and y;
(16.71) u
x· y
2
v
x· y
2
;
putting these in the formula for z we have
(16.72) z uv
x· y
2
x· y
2
x2 · y2
4
so the surface is the hyperbolic paraboloid 4z y2 x2.
Now, in general it may not be so easy (or simple) to realize a parametric surface as a level set;
however, we can use the parametric equations to , for example, find the tangent plane to the surface at a
point.
Proposition 16.6 Let Xú vµ xú vµI · yú vµJ · zú vµK be a vector-valued function defined on a
region in R-space. Define
(16.73) Xu
∂x
∂u
I·
∂y
∂u
J ·
∂z
∂u
K

(16.74) Xv
∂x
∂v
I·
∂y
∂v
J ·
∂z
∂v
K
a) The vector Xu ¢Xv is normal to the surface.
b) If w f´x y zµ is a function defined near the surface, we can consider it as a function of u and v by
writing w f´xú vµ yú vµ zú vµµ. Then
(16.75)
∂w
∂u
∇w¡Xu
∂w
∂v
∇w¡Xv
To see this, fix a point ú0 v0µ. If we set v v0 and let u vary, we get the curve C given parameterically
by
(16.76) Xúµ xú v0µI· yú v0µJ · zú v0µK
The tangent vector to this curve is Xu, and since the curve lies in the surface, its tangent vector lies in the
tangent plane. Similarly, considering the curve u u0, we see that the vector Xv also lies in the tangent
plane. Thus Xu ¢Xv is normal to the tangent plane. Part b) follows directly from the chain rule, applied
to the curves u u0 and v v0.
Example 16.24 Find the equation of the tangent plane to the surface of example 23 at the point (-2,4,3).
From (16.71), this point corresponds to the values u 1 v 3. Now, we differentiate the function
defining the surface, obtaining
(16.77) Xu I· J · vK Xv I· J · uK
The values at u 1 v 3 are Xu I· J · 3K Xv I· J · K. Thus, a normal to the tangent plane
is N Xu ¢Xv 2I 4J · 2K, and the equation of the tangent plane is
(16.78) 2´x· 2µ 4ý 4µ · 2´z 3µ 0 or z x· 2y 3
Example 16.25 Consider the surface given parametrically by
(16.79) Xú vµ 3cosucosvI· 4cosusinvJ · 5sinuK
Find the normal to the tangent plane ant the point corresponding to u π 3 v π 6.
Differentiate:
(16.80) Xu 3sinucosvI 4sinusinvJ · 5cosuK
(16.81) Xv 3cosusinvI· 4cosucosvJ
Evaluating at the given point, we have
(16.82) Xu 3
Ô
3
2
Ô
3
2
I 4
Ô
3
2
1
2
J · 5
1
2
K 9
4
I
Ô
3J ·
5
2
K
(16.83) Xv 3
1
2
1
2
I· 4
1
2
Ô
3
2
J 3
4
I·
Ô
3J
A normal to the plane is N Xu ¢Xv ´5 2µ
Ô
3I· ´15 8µJ · 3
Ô
3K.

Ü16.3 Theoretical considerations 247
Ü16.3. Theoretical considerations
In order to make the intuitive concept of linear approximation, as used above, more precise we start with
the idea of closeness in the space itself. We measure the “nearness” of two points by the length of the
line segment joining the points. Thus, in vectorial terms, the distance between X and X0 is X X0 ,
that is, the square root of the sum of the squares of the components. We define limits in terms of this
distance.
Definition 16.4 The ball of radius c centered at X0 (denoted B´X0 cµ) is the set of all points of distance
less than c from X0. A neighborhood of X0 is any set which contains some ball centered at X0.
Definition 16.5 Suppose that f is a function defined in a neighborhood of X0. We say that
(16.84) lim
X X0
f´Xµ L
if we can insure that f´Xµ L can be made as small as we please by taking X close enough to X0. We
say that f is continuous at X0 if
(16.85) lim
X X0
f´Xµ f´X0µ
Just as in one variable, we are assured that all functions which can be expressed by polynomials in the
coordinates are continuous.
Definition 16.6 A linear function is a function of the form L´Xµ C¡X, for some vector C. In
coordinates we have L´x y zµ ax·by·cz, where we have written C aI·bJ·cK and X xI·yJ·
zK. Its level surface through X0 is the plane a´x x0µ · bý y0µ · c´z z0µ 0, or C¡´X X0µ 0.
Now we define differentiability at X0 of a function f: that it can be well- approximated by a linear
function. This is the direct generalization of the definition of the derivative in one dimension.
Definition 16.7 Suppose that f is a function defined in a neighborhood of X0. We say that f is differ-
entiable at X0 if there is a linear function L such that
(16.86) lim
X X0
f´Xµ f´X0µ L´X X0µ
X X0
0
In this case, we call L the differential of f at X0, denoted d f´X0µ. We can write L ¡´X X0µ ∇f ¡
´X X0µ, where ∇f is the gradient of f.
Now, as we have seen, the calculation of differentials amounts to calculating partial derivatives. To
see this in terms of the above definition, let’s look at the situation in two variables, writing X xI· yJ.
Suppose that f is differentiable at X0, and its differential there is L´x x0 y y0µ a´x x0µ·bý y0µ.
First we see what happens to equation (16.86) along the line y y0. Then X X0 ´x x0µI, and we
get
(16.87) lim
x x0
¬
¬
f´x y0µ f´x0 y0µ a´x x0µ
x x0
¬
¬
¬
¬
f´x y0µ f´x0 y0
x x0
µ a
¬
¬ 0

or
(16.88)
∂ f
∂x
lim
x x0
f´x y0µ f´x0 y0µ
x x0
a
In the same way, we see that b ∂ f ∂y.
Now we turn to an argument for the chain rule in two dimensions.
The Chain Rule. Let w f´x yµ be a differentiable function defined in a region R. Suppose that γ is
a differentiable curve in R given parametrically by X X´tµ, with t 0 corresponding to X0. Then,
considering w f´X´tµµ as a function of t along γ, we have
(16.89)
dw
dt
´∇fµ ¡ dX
dt
evaluated at X0. We start with the definition of differentiability. Let
(16.90) η´tµ f´X´tµµ f´X0µ L¡´X´tµ X0µ
where we have written L for the gradient of w evaluated at X0. By (16.86),
(16.91) lim
X´tµ X0
η´tµ
X´tµ X0
0
Now, by continuity X´tµ X0 as t 0, and thus
(16.92) lim
t 0
¬
¬
η´tµ
t
¬
¬ lim
X´tµ X0
η´tµ
X´tµ X0
lim
t 0
X´tµ X0
t
0
by (16.91), and the assumption of differentiability of X´tµ, which assure that the second limit on the right
exists. Now, by the definition of η:
(16.93) f´X´tµµ f´X0µ L¡´X´tµ X0µ · η´tµ
Divide by t, and take the limit as t 0:
(16.94)
dw
dt
lim
t 0
f´X´tµµ f´X0µ
t
lim
t 0
L¡ X´tµ X0
t
· lim
t 0
η´tµ
t
L¡ dX
dt
Ü16.4. Optimization
Now we turn to the technique for finding maxima and minima of a function z f´x yµ of two variables.
Definition 16.8 If f´x0 y0µ f´x yµ is at least as large as its value at all nearby points, we say that f
has a local maximum at ´x0 y0µ. More precisely, if, for some a 0 we have
(16.95) f´x0 y0µ f´x yµ

Ü16.4 Optimization 249
for all ´x yµ within a distance a of ´x0 y0µ, then ´x0 y0µ is a local maximum point for f. Similarly, if
instead we have
(16.96) f´x0 y0µ f´x yµ
for all ´x yµ sufficiently close to ´x0 y0µ, then ´x0 y0µ is a local minimum point for f.
The first derivative test for functions of one variable gives us the following criterion:
Proposition 16.7 Suppose that X0 is a local maximum (or minimum) for f. Then ∇f 0.
To see this, pick a vector V and consider the line given by the equation X´tµ X0 ·tV. Then f´X´tµµ
has a maximum at t 0, so
(16.97) ∇f ¡V
d
dt
f´X´tµµ
¬
¬
0
0
This can only be true for all V if ∇f 0.
Definition 16.9 If ∇f´x0 y0µ 0 we say that ´x0 y0µ is a critical point.
Thus, to find the local maxima or minima of a function in a given region, one must look among the
critical points.
Example 16.26 Find the critical points of the function f´x yµ x3 · xy· y2 x.
We calculate the components of the gradient:
(16.98)
∂ f
∂x
3x2
· y 1
∂ f
∂y
x· 2y
Now, we set these equal to zero and solve. The second equation gives x 2y; substituting that in the
first gives 12y2 · y 1 0, which has the roots
(16.99) y
1 ¦
Ô
1 · 48
24
or y
1
4
1
3
Thus, the critical points are ´ 1 2 1 4µ ´2 3 1 3µ. But now, how can we tell whether or not we have
a local maximum or a local minimum at either of these points? In fact, we may have neither; there is
a third possibility: that along certain lines through the critical point, the value is a local maximum, and
along other lines, the value is a local minimum. Such a point is a saddle point.
Example 16.27 Let z x2 y2. Then the origin is a critical point for z. Since z x2 along the line
y 0, z has a minimum at the origin on this line, but on the line x 0, we have z y2 which has a
maximum at the origin along this line.
We distinguish among these points by using the second derivative test in one variable. In order to
make clear what the criterion is, we first consider the case of a quadratic function.
Example 16.28 Let z be a quadratic function of the variables u v: z au2 · 2buv· cv2. The origin is
a critical point. By completing the square we can discover what kind of critical point it is:
(16.100) z aú2
· 2
b
a
uv·
b2
a2
v2
µ · ć b2
a
µv2
aú ·
b
a
vµ
2
·
ac b2
a
v2

Thus if both terms have positive coefficients, the origin is a minimum; if both terms have negative
coefficients, the origin is a maximum, and if the signs differ, the origin is a saddle point. We call the
expression D ac b2 the discriminant of the quadratic function defining z. Notice that if D 0, that
the coefficients of (16.102) have the same sign, and if also a 0, we have a minimum. and if a 0, a
maximum. If D 0, the coefficients have different signs and we have a saddle point.
This example leads us directly to the general criterion by applying the second derivative test along
each line through the critical point. For the function z f´x yµ, let fxx fx y fyy represent the second
partial derivatives of f.
Proposition 16.8 Suppose that ∇f´x0 y0µ 0, that is ´x0 y0µ is a critical point. Then (evaluating at
´x0 y0µ):
If D fxx fyy ´ fxyµ2 0 at a point ´x0 y0µ, the f has a saddle point there.
If D fxx fyy ´ fxyµ2 0 and fxx 0, at a point ´x0 y0µ, then f has a local minimum there.
If D fxx fyy ´ fxyµ2 0 and fxx 0, at a point ´x0 y0µ, then f has a local maximum there.
If D 0, we can conclude nothing. We note that when D 0 the second derivative along all lines
has the same sign, so we could check whether fyy is greater or less than 0 instead, if that were easier.
To see this, choose a vector V uI· vJ and consider the function fV´tµ f´X0 ·tVµ f´x0 ·tu y0 ·
tvµ. Differentiating we find, by the chain rule,
(16.101)
d
dt
fV u fx · vfy
d2
dt2
fV
d
dt
ú fx · vfyµ u
d fx
dt
· v
d fy
dt
We compute the second derivative by applying the chain rule to the functions fx fy:
(16.102)
d2
dt2
fV uú fxx · vfyxµ · vú fxy · vfyyµ u2
fxx · 2uvfxy · v2
fyy
If this is positive, then the function fV has a minimum; that is the function f has a minimum along the
line in the direction of V. If this holds for all directions V; that is, for all values of u v, then f has a local
minimum at ´x0 y0µ. But, referring back to example 16.28, this is true if D 0 fxx 0. Similarly, if
D 0 fxx 0, then f has a maximum along all lines through ´x0 y0µ, so f has a local maximum there.
However, if D and fxx have different signs, then f has a local maximum in some directions, and a local
minimum in others, so we have a saddle point.
Example 16.29 We continue with example 16.26. We found critical points at P´ 1 2 1 4µ, Q´2 3 1 3µ.
Differentiating the first partials (see (16.98)), we get
(16.103) fxx 6x fxy 1 fyy 2
Thus
(16.104) at P : D 6´ 1
2
µ´2µ 12
7 and at Q : D 6´
2
3
µ´2µ 12
7
so P is a saddle point, and since fxx 4 0, Q is a local minimum.
Example 16.30 Let f´x yµ x2 · 2y4 · xy· 4x· 2y. Find the local maxima and minima of z. Does f
have a global maximum or minimum?

First we find the critical points:
(16.105) fx 2x· y· 4 fy 8y3
· x· 2
To find the points where both are zero, we obtain x 8y3 2 from the second equation. Putting this in
the first,we get
(16.106) 2´ 8y3
2µ · y· 4 0 or 16y3
· y 0
This has the solutions y 0 ¦1 4, so the critical points are P´ 2 0µ Q´ 17 8 1 4µ, R´ 15 8 1 4µ.
We now calculate the second derivatives:
(16.107) fxx 2 fxy 1 fyy 24y2
Then D 48y2 1, which is positive at all of these points. Since fxx is everywhere positive, these are all
local minima. To determine the global minimum, we evaluate: f´Pµ 4 f´Qµ 4 0078 f´Rµ
4 0078. Thus the global minimum is -4.0078, attained at both Q and R. Everywhere else the function
has a direction in which it is increasing, so it has no global maximum.
Notice, in these problems we have to solve several equations simultaneously, and usually they are not
linear. There are no universal algorithms for solving such systems of equations, and we have to follow
our intuition. Usually the technique of substitution works (although in the above problem, with other
constants the cubic equation in (16.106) would be much more difficult). So, in general the procedure to
follow is to look at the given equations to see if, in one of the equations one of the variables can be easily
written in terms of the other. If so, substitute that expression in the other equation.
Ü16.4.1 The Method of Lagrange Multipliers
Let C be a curve in the plane, not going through the origin. Let’s find the point on C which is closest to
the origin. This amounts to finding the minimum value of f´x yµ x2 · y2 on the curve C. If C is given
parametrically by the equations x x´tµ y y´tµ, we know what to do: differentiate f´x´tµ y´tµµ and
set the derivative equal to zero. But, if the curve is given implicitly by an equation g´x yµ c, we don’t
want to solve the equation explicitly, and we don’t have to. Looking at the condition
(16.108)
d
dt
f´x´tµ y´tµ 0 as ∇f ¡ dX
dt
0
we see that the requirement is that ∇f is orthogonal to the tangent to the curve at the minimizing point.
But ∇g is orthogonal to its level set C everywhere, so at the minimizing point we have that ∇f and ∇g
are collinear; that is, they are multiples of each other. Thus, we can solve the problem by fiinding the
solution of the system
(16.109) ∇f λ∇g g´x yµ c
This gives three scalar equations in three unknowns, which, in principle, can be solved. Of course the
value of λ is not of interest, but is useful as an auxiliary to finding the values of x y.
Example 16.31 Find the point on the line 3x 2y 1 which is closest to the point (4,7).

Given the constraint g´x yµ 3x 2y 1, we want to minimize f´x yµ ´x 4µ2 · ý 7µ2. The
gradients are
(16.110) ∇f 2´x 4µI· 2ý 7µJ and ∇g 3I 2J
These gradients are collinear at the minimizing point, so we have to solve the equations
(16.111) 2´x 4µ 3λ 2ý 7µ 2λ and 3x 2y 1
We can eliminate λ from the first two equations:
(16.112) 4´x 4µ 6λ 6ý 7µ so that 4x 6y 26
Now we have simultaneous linear equations in x and y which we can solve, getting the point (16,47/2).
We note that the Lagrangian equations (16.109) just say that the line from this point to (4,7) has to be
orthogonal to the given line; something we knew from geometry.
Example 16.32 Find the maximum value of f´x yµ xy on the ellipse x2 · 4y2 1.
Let g´x yµ x2 · 4y2. We calculate the gradients: ∇f yI · xJ and ∇g 2xI · 8yJ. At the point
on the ellipse at which we have the maximum, we have ∇f orthogonal to the tangent to the ellipse, so is
collinear with ∇g. Thus we have the equation ∇f λ∇g for some λ. This gives the scalar equations
(16.113) y 2λx x 8λy x2
· 8y2
1
We can eliminate λ by dividing the first equation by the second:
(16.114)
y
x
2λx
8λy
x
4y
giving x2
4y2
Substituting that in the last equation gives 4y2 · 4y2 1, so that y ¦1 ´2
Ô
2µ. Then
(16.115) x2
4y2 4
8
so that x ¦ 1
Ô
2
The possible values of f´x yµ xy at these points are ¦1 4, so the maximum value of f is 1/4, and its
minimum is 1 4.
The parameter λ, called the Lagrange multiplier, serves the purpose of finding a relation between
x and y which is a consequence of the optimization. The value of λ is not important, but in some cases it
may make the problem easier to first determine λ.
To summarize: given the problem: minimize (or maximize) a function f´x yµ subject to a constraint
g´x yµ c. We observe that the chain rule tells us that, at the optimizing point, ∇f is orthogonal to the
tangent to the level set of g. But so is ∇g, so we must have ∇f λ∇g for some λ. Solve this equation
in conjunction with g´x yµ c to find the point. This method (of Lagrange multipliers) works in three
dimensions as well.
Proposition 16.9 Suppose that w f´x y zµ is a differentiable function, and we wish to find its maxima
and minima subject to a constraint g´x y zµ c. At an optimizing point P there is a λ such that
(16.116) ∇f´Pµ λ∇g´Pµ g´x y zµ c

These equations give a system of four equations in four unknowns which, in typical circumstances,
have only a finite number of solutions. The maximum (minimum) of the function must occur at one of
these points.
To see why this is true, we follow the two dimensional argument. Let S be the level surface g´x y zµ c.
Let C be a curve through P lying in the surface S. Then f is optimized along C, so that the derivative
of f along the curve is zero at P. But this just says that ∇f´Pµ is orthogonal to the tangent to the curve.
Since every vector in the tangent plane to S is the tangent vector to such a curve, ∇f´Pµ is orthogonal to
the tangent plane to S. But so is ∇g´Pµ, so ∇f´Pµ and ∇g´Pµ must be colinear.
Example 16.33 Find the point on the plane 2x· 3y· z 1 closest to the point ´1 1 0µ.
Here the constraint is g´x y zµ 2x · 3y · z 1 and the function to be minimized is f´x y zµ
´x 1µ2 · ý· 1µ2 · z2. Taking the gradients and introducing the Lagrange multipier, we are led to the
equations
(16.117) 2´x 1µ 2λ 2ý· 1µ 3λ 2z λ 2x· 3y· z 1
We use the first three equations to express the variables in terms of λ, and then use the last to solve for
λ:
(16.118) x λ · 1 y
3λ 2
2
z
λ
2
so that
(16.119) 2´λ · 1µ · 3
3λ 2
2
·
λ
2
1
This gives λ 1 7. Substituting into equations 16.118), we find the desired point to be (1/7, -11/14, 1/14).
Example 16.34 Farmer Brown wishes to enclose a rectangular coop of 1000 square feet. He will build
three sides of brick, costing $25 per linear foot, and the fourth of chain link fence, at $ 15 per linear foot.
What should the dimensions be to minimize the cost?
Let x and y be the dimensions of the coop, where x represents the sides, both of which are to be of
brick. The constraint is g´x yµ xy 1000, and the cost function is C 25´2x· yµ · 15y. We have
∇C 50I· 40J, and ∇g yI· xJ. The equations to solve are:
(16.120) 50 λy 40 λx xy 1000
so
(16.121) 1000 xy ´
50
λ
µ´
40
λ
µ
or λ2 ´50µ´40µ 1000 2, giving λ
Ô
2. Then x 40
Ô
2 20
Ô
2 y 50 λ 25
Ô
2.
Many problems involve finding the maximum or minimum of a function of many variables subject
to many constraints The technique of Lagrange multipliers works in this general context, but - of course
- is much more difficult to employ. To give a sense of the general procedure, we state the proposition in
the case of a function of three variables with two constraints.

Proposition 16.10 To find the extreme values of a function f´x y zµ subject to two constraints (say along
a curve), g´x y zµ c h´x y zµ d, we have to solve the five equations in the five unknowns x y z λ µ:
(16.122) ∇f´Pµ λ∇g´Pµ · µ∇h g´x y zµ c h´x y zµ d

diff. of functions

More Related Content

What's hot (18)

Viewers also liked (20)

Similar to diff. of functions (20)

More from Elias Dinsa (7)

diff. of functions