•

,
•
9. SLOPE FIELPS ANn EULER TANGENTS
•
p
•

If you are using a pre-reform textbook. there is a good chance that these two topics do not appear or
are treated in a cursory manner. The fact is. these areas could not be meaningfully addressed before
the graphing calculator because the work to produce them is so tedious. With technology. they
. .'
provide streng support to both the field of differential equations and the notion of linear
appro xi mation. Slope fields can be introduced when students first encounter the general
antiderivative of a function to reinforce the idea that the solution is a family of functions.

There are some very good calculator programs that can be used with these topics; however. some
care needs to be taken in using them. The program for slope fields will generate a flow field solution
for a differential equation in x and/or y. However. when the window is notjudiciously chosen. the
results can be horrendous. especially with exponential functions. It is helpful to try to predict the
size of slopes and to adjust the window so that they do not become too large or too small. For
instance, if dy/dx = xy. the window [-2.2] x [-2.2] gives a nice slope field with about 10 intervals for
both x and y; however, if we change the window to [-10.10] x [-10.10]. the slope field looks nearly
vertical. inasmuch as the slopes at the extremities have absolute value near 100. Similarly. one
needs to be careful in deciding the number of x and y values selected. For instance. if dy/dx = x/y
and the number of y values selected is 10. then it is very likely that the program will attempt to
evaluate the slope when y = O. I have found it useful to choose a number of values such as 11 or 13
to avoid having the slope evaluated at a window value of zero.

The programs that produce Euler tangent values and graph Euler tangent solutions are very helpful.
especially the latter. When using the Euler tangent graphic program. Ell/erg. try using a differential
equation whose solution can be found analytically so that they can see how making L1x small
provides for closer solution approximations and how propagated error grows as the number of
. values increases. [Note: Since this programs turns off functions. it will be necessary to go into the
Y= stack to tum on the solution to the D.E.]

None of these pn:;>grall1s. however. is likely to be a direct help to students on the AP exam. There
,
. has been much speculation about how the items would be tested; we discovered at least one answer
to this speculation in the 1998 Be exam. where students ~ere asked to sketch a slope field and to
compute a few Euler tangent values with a given set of conditions. Students must show their work.
so it will be necessary to know that ~y '" dy/dx'~x and how to apply it. In the multiple choice area.
it is still likely that students will be aSked to do something such as matching a slope fie'ld with a
differential equation or finding an approximation to f(xO+~x) given an initial condition (xo.f(Xo)) and
a value for ~x.

, A final note. Slope fields and Euler tangents are accessible enough as a topic that it would not
surprise me if they eventually end up in the AB syllabus. Therefore. it is probably not a bad idea to
integrate this subject matter as, time allows in an AB course.

JJENSEN APCALC
*" ,",' ,

SI,OPE FIEI,PS
"

-
- .
,
A slope field is a graphical display which shows the flow of tangent lines to the family of solution •

curves of a differential equation. Often, this flow diagram provides valuable information about the
nature of the solution curve, i.e., whether it is polynomial, exponential. circular, uigonomerric, etc. . ")
Slope fields can be.consllUcted. over a region .of the plane by direct substitution into the differential -
equation, or they can be generated using the graphing calculator and an appropriate program.

There are generally two types of problems that involve slope fields. In the first, we are given a
differential equation and asked to produce its slope field diagram. In the second, we are given a
•
slope field and asked to match it to a paJ.ticular differential equation which has those characteristics .

CONSTRUCTING A SLOPE FIELD.

When constructing slope fields, it is helpful to first create a table of values for a specific intelval of x
and y. For example, if we know that dy/dx = xy/4 for x in [-2,2] and yin [-2,2], we can consuuct a
grid to evaluate the slopes at each integer pair of values. Complete the grid of slope values for this
differential equation: .
x values
1. -2 ·1 o 1 2
•
2 -( .. J:.2.. 0
1 -\;::1-. -(.. 6
.!' l .

y values o () 0 0
- 1 '-r.1- l-,.-
'-f 0
•
. . -2 t ....
1,
..
" • " ' ,0 1,. 6 .
, •
i
"
.. '

Once the grid is complete, the slope segments can be drawn tlrrough the lattice points of a graph.
Draw the slope field for the values obtained above: .

2.
\ \
--. • •

\
'Y"'~
..., .' • • .. • •

/ ..--- ..
~' ."
~ (' , • •

./ ·
/ "" / -.' • •

3. What kind of function seems to be pictured by the slope field you constructed?

,
••

" 0"

JJENSEN APCALC

•
 •

, READING A SLOPE FIELD.

-- When reading a slope field, it is important to look for clues the slope segments give us about the
•
behavior of the differential equation and, by extension, its family of solutions.
It is possible to read a slope field one segment at a time: however, this can be exhausting for very
large fields. It is probably easier in many cases to spot trends in the slope field that tell us something
about how x and y are related in the differential equation. Here are some approaches you can use:

• Examine slope field segments along vertical lines. If the segments along each vertical line
have the same slope. then the differential equation does not depend on y, because, as y varies, the
slope does not.

• Examine slope field segments along horizontal lines. If the segments along each horizontal
line have the same slope. then the differential equation does not depend on x.

• Examine slope field segments in the first quadrant. If the segments have positive slope, then
there are likely 110 negatives in the expression of the differential equation. If the slopes become
larger as x gets larger. then dy/dx varies directly with x; likewise for y. Otherwise. we can
detelmine that the slope is inversely related to one or both vaIiables.

• If the slope field evinces a curve which looks familiar, check by differentiating that curve to
see if its slope field fits the graphical d(i(d.

NOTE: There are occasional anomalies in the appeaI'ance of slope fields. due to the way they aI'e
generated on a calculator. Small discrepancies in slope can usually be dismissed.

Consider the following slope field:

y' / ...• ,. .. ,. / '
,. / '
I
/ / . ,. ...
, /
•
•
•
. .. ...
,

' ."
'/ / /' /
/ /'
..
/ /'
'

/ / • / I.
•

/ / .
i
.. . . ,.
/ /
I / . ,. . • / /.
x [-

,4. What can you deduce from reading the slope segments? _ _ _ _ _ _ _ _ _ _ _ __

5. , Which of these is most likely the differential equation:

(A) dy/dx = .5xy (B) dy/dx = x 2/Y (C) dy/dx = .5x 2

(3 )
-

JJENSEN APe ALe

 -
• •
•

Here are some more slope fields to practice on. In each, match the slope field with its differential
•
" .•
equation.

6. Which of the following differential

equations has the,solution slope field
pictured at right?

(A) dy/dx = .5y

(B) dy/dx = .2x/y

(C) dy/dx = xy

(D) dy/dx = x + y

(E) dy/dx = l/x

7. Which of the following differential

equations has the solution slope field
pictured at light?

(A) dy/dx = x2

(B) dy/dx = y/x

.
, '
\
• • (C) dy/dx =-'1 ' ,J

(D) dy/dx = -x/y

•

(E) dy/dx = x2 + y2

8. Which of the following differential

\
equations has the solution slope field •

pictured at right?

(A) dy/dx = x + Y .'

.".
...... ...... .".

(B)dy/dx=x-y

(C) dy/dx = x2

(D) dy/dx = 2y

(E) dy/dx = y/x

-- , '\!
••

HENSEN APCALC
 •
\.., ,
EULER TANGENTS
--
•
Imagine trying to measure a curve using a rigid ruler. Because the contours of the curve cannot be
followed by the IUler. you would have to make small interval measurements as an estimate. This
process underlies a calculus procedure known as Euler Tanjients.

Let dy/dx = f(x .y). that is. a differe ntial equation in x and/or y. and let (xO. yO) be an in itia l
condition of its solution. Further. let's suppose that the differential equation cannot be solved by our
known method of separating variables. Then. we can only estimate points on the solution curve
using the evidence provided.

Illustration: dy/dx = x + y and (xO. yO) = (1.1)

From previous work with differentials. we know that. when flx is "sufficiently
small" fly:::: dy/dx • Llx, which is to say that the curve is approximated by its
tangent line over a small interval. Let's choose flx = .1: since dy/dx = x + y.
at (1,1). dy/dx = 2 and fly "" 2(.1) = .2. This brings us to an estimate of a
nearby point on the solution curve, (1.1. 1.2). Call this (xl. yl).
Now we can repeat the process. At (1.1, 1.2). dy/dx = 2.3 and fly "" (2.3 )(.1)
= .23. Thus. our next point is (x2. y2) = (1.2. 1.43).
As we continue. the y-value estimates become more complicated: however. the x-
values continue to increment by .1 each time.

1. Use the method of Euler tangents to complete the table below for this example.
Give estimates accW'ate to three decimal places.

x 1 . 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1. 9 2.0

,
Y 1 1.2 1.43 \, ,,
\,

\
\
\
\ \
\
\

. . .,,
\
• \
• ' 2:3
\
\,
dy/dx 2 2.63 , ,
,
\
\, ,,
\ •
"
\ \
\
fly .2 .23 .263
,
•
\
\
\ ,,,
\
\

\
•
, . \
• \
2. Given that dy/dx = 2y with initial condition (xO. yO) = (OJ). use Euler
tangents to estimate the solution using LlX = .2 for five steps. i.e .• from x = 0
\
to x = I.

x 0 .2 .4 .6 .8 1.0

y
j ( '{
t> "
1'\ . "',\ "\
'\ '\

1-, '''.,
dy/dx
'" '\,
.
"',~ , ,
",

~"-
•

fly
.. V " ~

,
(5)

HE NSEN APCALC
 ••
(a) Solve the differential equation in example 2. Use the initial value to resolve the •

constant. '
•
•
'>

.
..
'.

.; ,
\

"

a. _________________________

(b) In the table below, compare the values of y obtained in example 2 with the
values of your solution in part (a). Compute the difference in each case.

x 0 .2 A .6 .8 l.0
•

Euler y I I L(
y
""\ "-\
. •"-..
Actual v I t.b Ii q1., ", \\ ' \,

"\ \.
Difference 0 v 0 '1 '2_ .~ "•,...,
"

(c) In the windows below, plot: (I) the curve you found in pan (a) and,
(2) the points you found in pan (b). The window dimensions are [0.1] x [0,8].

. .
. , .. .-

\
4. For each of the following situations. indi<.:ute whether the Euler tangent approximation
is less than. greater than. or equal to the u<.:tual solution of the differential equation for
positive values of ~x . .

(a) The solution curve is concave up. a.

------------------
(b) The solution curve is concave down. b. __________________

(c) dy/dx = k. for k a constant. c. ______________________

5. In general. what can you do to improve the accuracy of an Euler tangent solution of a
differential equation?

(6 )

JJENSEN APCALC
•
. POSTSCRIPT ON EULER TANGENTS .
/
•
• The Fundamental Theorem with Euler Tangents .
..
Given a function f. find an approximation to f with the
initial condi tion (xO. f(xO)).

For small ~x. t:.y/t:.x == f(xo)

t:.y =:: f(xO) 6x

f(x l) == f(xO) + f(xo) t:.x

f(x2) == f(x 1) + f(x 1) 6x

= f(xO) + f(xO) t:.x + f(x 1) t:.x

•• •

f(xn) =:: f(xO) + f(xO) 6x + f(x 1) t:.x + ... + f( -1) ~x

By the left endpoint method, the area bounded by f from xO to xn can be

approximated as:

A =:: f(xO) ~x + f( x 1) ~x + ... + f( xn-l) ~x

•
But, f(xn) =:: f(xO) + f(xO) t:.x + rex 1) t:.x + ... + f(xn-l) t:.x
, .
" .. . . . .
or.
•
, f(xn) - f(xo) =:: f(xO) t:.x + f(x 1) Ax + ... + f(xll-l) t:.x

Thus. A = f(xn) - f(xO) as t:.x->O. which is the FTC!

.'

lJENSEN Arc-ALe