Derivative of Arcsecant (Why The Absolute Value)

The Inverse Secant Function Three Times
Have a look at the plots of the cosine and secant functions given below. It should be clear
that in order to define an inverse for secant we must restrict its domain (just as for the sine,
cosine and tangent functions). What is less clear than for the other functions is just how to restrict
the domain.
1. Arcsecant Using { }
2
[0, ]

for the Domain of Secant.
It seems to me, and to many other people, that the most natural (or obvious) choice is to
restrict x to the interval
2
[0, ] { }
. Having done this,

sec( ) x is now one-to-one and thus has an inverse. The inverse is defined in the usual way:
{ }
1
2
sec sec [0, ] , ( , 1] [1, ) y x y x x y

See the graph below. It shows
1
sec y x
together with the lines 2 y ,

1 and 1 x x .
To find the derivative of the inverse secant we proceed via implicit differentiation:
Let
1
sec y x
so that sec y x and it follows that

2
sec
sec tan 1
1 1
sec tan
1
d d
y x
dx dx
y y y
y
y y
x x
To see where the absolute value sign comes from, consider the triangle below. The sides
are determined by the relationship sec
1
x
y x . But you must keep in mind what is y in fact. It
is the number (or angle in radians)
1
sec x
and as such represents a value in the interval

{ }
2
[0, ]

which depends on the choice of x. If ( , 1] x then
( )
2
, y

(i.e. a quadrant II angle) and sosec tan y y will be
positive since both tangent and secant are negative in QII. If
[1, ) x then
2
[0, ) y

(i.e. a quadrant I angle) and so sec tan y y
is again positive. It follows that 0 y > for all x in the domain of
1
sec x
and to ensure this we take the absolute value of x in the derivative.

Sample Calculation: One important effect of the domain choice for the arcsecant
function is in the calculation of some definite
integrals. Consider the calculation
1
2
2
1 x dx
which
requires trig substitution for its calculation. We will
use the substitution { }
2
sec for [0, ]- x

. To
change the limits of integration requires the inverse
secant function:
1
2
( 2) (sec 2) sec ( 2)
3
x

_

,
and
( )
1
( 1) sec 1 sec ( 1) x
. The choices for are because must be a

quadrant two angle because of the domain { }
2
[0, ]

.
The following calculations then yield:
( )
( )
1
2 2
2 2 / 3 2 / 3
2 / 3
2 / 3
3
1 tan sec tan tan sec tan
1
sec sec sec tan ln sec tan
2
1
3 ln 2 3 1.07
2
x dx d d
d

+ 1
]
+

.
You should make sure that you can fill in the significant number of missing steps in the
calculation. Also, Maple gives the answer as
( )
1
3 ln 2 3
2
+ and you should verify that the
two answers are the same.
Important: the choice
2
tan tan is made because of the fact that the variable
is in quadrant II and hence tan 0 < . If you rework the example using
2
tan tan then you
will get the answer
( )
1
3 ln 2 3 1.07
2
+ + . This cannot be right as we recognize that the
integral must be positive since the original function is positive.
2. Arcsecant using
3
2 2
[0, ) [ , )

for the Domain of Secant.
With this new choice for the domain of sec y x (again making it one-to-one) the
formula for the derivative changes. The same approach as before (using the same triangle) yields
that if
1
sec y x
then
2
1 1
sec tan
1
y
y y
x x

. To see why the absolute value sign has

disappeared consider the angle (number)
1
sec y x
. If ( , 1] x then y is will be a QIII

angle (number) and so sec tan y y will be negative, as will
2
1
1 x x
. If [1, ) x then y will be
a QI angle (number) and so, again, sec tan y y and
2
1
1 x x
will agree.
Clearly, the graph of this version of
1
sec y x
must be different. It is decreasing on

( , 1] x and
1
3
lim sec
2 x
x

while
1
1
limsec
x
x
. On [1, ) x , the graph is increasing

starting at (1,0) and satisfying
1
limsec
2 x
x

.
Sample Calculation Revisited We can now try
1
2
2
1 x dx
again using this second

definition of
1
sec y x
. Under the same substitution approach the limits of integration will

change as follows:
1
2 2
2 4 / 3 4 / 3
1 tan sec tan tan sec tan x dx d d

. Note that
2
tan tan since now is a QIII angle and tangent is positive there. For practice, you
should work through the details and make sure that you end up with the same final result as
earlier.
3. Yet Another way to Define Arcsecant
Let ( ) sec f x x for
2
[0, ] { } x

and
1
1
( ) cos g x
x

,
, 0 x . Recall: f and g are
inverses of each other if and only if ( ( )) and ( ( ))
g f
f g x x x D g f x x x D .
1
1
1 1
( ( )) sec cos
1
cos cos
1 1
if [ 1,1] (which it is)
1
f g x
x
x
x
x
x
_ _

_ _ , ,

, ,

( )
1 1
1
( ( )) cos cos cos
sec
if [0, ] (which it is)
g f x x
x
x x

_

,

.
It follows that
1 1
1
sec cos x
x

_

,
.
This makes deriving the derivative quite easy. Regard:
( )
1 1
2
2
2
2 2 2
2
2
1 1 1
sec cos
1
1
1 1 1
1 | | 1
1
| |
d d
x
dx dx x x
x
x
x x x x
x
x
x

_ _

, ,

Derivative of Arcsecant (Why The Absolute Value)

Uploaded by

Copyright:

Available Formats

Derivative of Arcsecant (Why The Absolute Value)

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Derivative of Arcsecant (Why The Absolute Value)

Uploaded by

Copyright:

Available Formats

The Inverse Secant Function Three Times

. Having done this,

together with the lines 2 y ,

so that sec y x and it follows that

and as such represents a value in the interval

and to ensure this we take the absolute value of x in the derivative.

. The choices for are because must be a

. To see why the absolute value sign has

. If ( , 1] x then y is will be a QIII

must be different. It is decreasing on

. On [1, ) x , the graph is increasing

again using this second

. Under the same substitution approach the limits of integration will

You might also like