Arkansas Tech University

MATH 2934: Calculus III

Dr. Marcel B. Finan
17 Second Order Partial Derivatives
Just as we had higher order derivatives with functions of one variable we will
also have higher order derivatives of functions of more than one variable.
Consider the case of a function of two variables z = f(x, y), since both of the
rst order partial derivatives are also functions of x and y we could in turn
dierentiate each with respect to x or y. This means that for the case of a
function of two variables there will be a total of four possible second order
derivatives. The second order partial derivatives are
(f
x
)
x
=f
xx
=

x

f
x

=

2
f
x
2
(f
x
)
y
=f
xy
=

y

f
x

=

2
f
yx
(f
y
)
x
=f
yx
=

x

f
y

=

2
f
xy
(f
y
)
y
=f
yy
=

y

f
y

=

2
f
y
2
The partial derivatives f
xy
and f
yx
are called mixed partial derivatives.
Example 17.1
Find f
xx
, f
yy
, f
xy
and f
yx
given that f(x, y) = sin xy.
Solution.
We have
f
x
(x, y) =y cos xy
f
xx
(x, y) =y
2
sin xy
f
xy
(x, y) =cos xy xy sin xy
f
y
(x, y) =x cos xy
f
yy
(x, y) =x
2
sin xy
f
yx
(x, y) =cos xy xy sin xy
Example 17.2
Find f
xx
, f
yy
, f
xy
and f
yx
given that f(x, y) = x
3
+ 2xy.
1
Solution.
We have
f
x
(x, y) =3x
2
+ 2y
f
xx
(x, y) =6x
f
xy
(x, y) =2
f
y
(x, y) =2x
f
yy
(x, y) =0
f
yx
(x, y) =2
Example 17.3
Use the contour diagram shown in Figure 17.1 to decide the sign (positive,
negative, or zero) of each of the following partial derivatives at the point P.
Assume the x and yaxes are in the usual positions.
(a) f
x
(P) (b) f
y
(P) (c) f
xx
(P) (d) f
yy
(P) (e) f
xy
(P).
Figure 17.1
Solution.
Suppose that P = (a, b). Then (a) f
x
(P) = f
x
(a, b)
f(a+h,b)f(a,b)
h
< 0 since
f decreases as we go to the right.
(b) f
y
(P) = f
x
(a, b)
f(a,b+h)f(a,b)
h
> 0 since f increases as we go up.
(c) f
xx
(P) = f
xx
(a, b)
fx(a+h,b)fx(a,b)
h
> 0 : Since the level curves are
further apart to the right, f
x
is changing from a negative number with
larger magnitude to a negative number with smaller magnitude so that
f
x
(a + h, b) > f
x
(a, b), that is, f
x
is increasing as we move to the right.
(d) f
yy
(P) = f
yy
(a, b)
fy(a,b+h)fy(a,b)
h
> 0 : Since the level curves are closer
as we move up, f
y
is changing from a smaller positive number to a larger
2
positive number, that is, f
y
is increasing as we move up.
(e) f
xy
(P) = f
xy
(a, b)
fx(a,b+h)fx(a,b)
h
< 0 : As we move up f
x
changes from
a negative number with smaller magnitude to a negative number with larger
magnitude, that is, f
x
is decreasing as we move up
Observe that in the rst two examples of this section the mixed partials
f
xy
and f
yx
are equal, i.e. f
xy
= f
yx
. This is a general result as given by the
following theorem.
Theorem 17.1
If the mixed partial derivatives f
xy
and f
yx
are continuous at a point (a, b)
then
f
xy
(a, b) = f
yx
(a, b).
Proof.
For h = 0 small we dene the function
F(h) = [f(a + h, b + h) f(a + h, b)] [f(a, b + h) f(a, b)].
If we let g(x) = f(x, b + h) f(x, b) then the previous equality can be
expressed in terms of g as follows
F(h) = g(a + h) g(a).
By the Mean Value Theorem, there is a number c between a and a +h such
that
g(a + h) g(a) = g

(c)h = h[f
x
(c, b + h) f
x
(c, b)].
Applying the Mean Value Theorem again, this time to f
x
we get a number
d between b and b + h such that
f
x
(c, b + h) f
x
(c, b) = f
xy
(c, d)h.
Combining these equations, we obtain
F(h) = h
2
f
xy
(c, d).
If h 0 then (c, d) (a, b), so the continuity of f
xy
at (a, b) gives
lim
h0
F(h)
h
2
= lim
(c,d)(a,b)
f
xy
(c, d) = f
xy
(a, b).
3
Similarly, by writing
F(h) = [f(a + h, b + h) f(a, b + h)] [f(a + h, b) f(a, b)]
and using the Mean Value Theorem twice and the continuity of f
yx
at (a, b)
we obtain
lim
h0
F(h)
h
2
= lim
(c,d)(a,b)
f
yx
(c, d) = f
yx
(a, b).
It follows that f
xy
(a, b) = f
yx
(a, b) as desired
Quadratic Approximation: Taylor Polynomials for Functions of
Two Variables
In Section 13, we saw how to approximate f(x, y) by a linear function, i.e.
its local linearization. This approximation can be improved by using second
order partial derivatives to construct a quadratic Taylor approximation for
the function.
Recall that for functions of a single variable, we can approximate the function
f(x) using the Taylor expansion:
f(x) = f(a) +
f

(a)
1!
(x a) +
f

(a)
2!
(x a)
2
+ +
f
(n)
(a)
n!
(x a)
n
+
for x near a. In particular the rst order Taylor approximation gives:
f(x) f(a) + f

(a)(x a),
for x near a, which is equivalent to the tangent line to f at the point a. The
second order approximation:
f(x) f(a) + f

(a)(x a) +
f

(a)
2
(x a)
2
for x near a which gives an improved approximation to f near a when com-
pared to the rst order or linear approximation.
For functions of two variables we already discussed the linear approximation
or local linearization of f :
f(x, y) f(a, b) + f
x
(a, b)(x a) + f
y
(a, b)(y b)
which is a tangent plane to the surface z = f(x, y) at the point (a, b, f(a, b)).
This linearization is also called the rst degree Taylor polynomial of f
4
at (a, b).
Just as in the single variable case, we can use the second order partial deriva-
tives to create an improved second-order approximation for f at (a, b) using
a polynomial of degree 2 in x and y around (a, b), i.e. an expression of the
form
p(x, y) = a
0
+a
1
(xa) +a
2
(y b) +a
3
(xa)(y b) +a
4
(xa)
2
+a
5
(y b)
2
.
A simple algebra similar to the single variable case (see Section 61) leads to
f(x, y) f(a, b) + f
x
(a, b)(x a) + f
y
(a, b)(y b) + f
xy
(a, b)(x a)(y b)
+
1
2
f
xx
(a, b)(x a)
2
+
1
2
f
yy
(a, b)(y b)
2
.
The right-hand polynomial is called the second degree Taylor polynomial
of f at (a, b).
Example 17.4
Find the second order Taylor approximation for f(x, y) =
1
xy
at the point
(1, 2).
Solution.
First we must compute the partial derivatives and evaluate them at (1, 2):
f(x, y) =
1
xy
, f(1, 2) =
1
2
f
x
(x, y) =
1
x
2
y
, f
x
(1, 2) =
1
2
f
y
(x, y) =
1
xy
2
, f
y
(1, 2) =
1
4
f
xx
(x, y) =
2
x
3
y
, f
xx
(1, 2) =1
f
xy
(x, y) =
1
x
2
y
2
, f
xy
(1, 2) =
1
4
f
yy
(x, y) =
2
xy
3
, f
yy
(1, 2) =
1
4
So the quadratic Taylor polynomial for f near (1, 2) is given by
f(x, y)
1
2

x 1
2

y 2
4
+
(x 1)
2
2
+
(x 1)(y 2)
4
+
(y 2)
2
8
5