A Reader on Functions

by Zain Syed Ahmed

November 22, 2022

1 Introduction
What is a function?

A function is a pattern or “mapping” that connects its input (or “argument”)

to a single output. Often a function is written as f or some other letter, symbol,
or word. If x is the input and f is the function (pattern), f (x) is the output. If
we want the variable y to take on the output, we can write:
y = f (x) (1)
Often a function is described by a table of values. Here is an example:

Table 1: Table of values for a sample function.

x y = f (x)
0 -1
1 1
2 1
3 0
4 2
5 4

Here is an example of a pattern that is not a function:

There is more than one output for x = 2 and x = 5. While this not a function,
it is still a relation. A relation maps one set of values or inputs to another set
of values or outputs. A function is a relation where there is only one output for
each input.

A function can also be expressed in terms of algebra or other functions. For

example,
f (x) = 2x2 − 4x (2)
We can then generate the set of coordinates that follows the relationship y =
f (x). We can create a short table of values that represents some of these coor-
dinates:

1
 Table 2: Table of values for a sample relation that is not a function.
x y
0 -1
1 1
2 1
2 2
3 0
4 2
5 4
5 7
5 8

Table 3: Table of values for a function with an algebraic expression.

x y = f (x)
-4 48
-3 30
-2 16
-1 6
0 0
1 -2
2 0
3 6
4 16
5 30
6 48

Let us now consider a different algebraic expression: —f(x)— = x. This is

clearly not a function, since there is a positive and negative value of f (x) for
each x (other than 0). In fact, it would be improper to use this notation and
suggest that the relation is a function.

One benefit of functions is that if you put in an allowable input (not all√ in-
puts are allowable, i.e. you cannot put a negative number into f (x) = x),
you receive a single output. This is especially useful when a function can be
expressed algebraically, since all you need to do is plug in the input number.
Even when there is a function like g(x) = sin x that cannot be expressed alge-
braically (at least not with a finite number of terms), if it is continuous, it can
often be approximated by a polynomial.

You can think of a function like a industrial grade machine. If you put the
same thing in, the same thing comes out. Now, sometimes you have to put mul-
tiple things in a machine and multiple things come out. Sometimes you have

2
to put multiple things in and one thing comes out. Sometimes you have to put
one thing in and multiple things come out. All of these formats can be true for
a function. The key is that the result (output) is consistent for the same input.

Say you have a machine that is tells you the colour of a sample of paper. The
machine is used at a paint store so that one can then order an exact colour of
paint. The output is just a six digit alphanumeric code on a screen: “E74B07”.
Two different samples could have the same colour, but if you put the same sam-
ple into the machine, without damaging the machine, clouding the sensor (with
dust, water, oil from one’s hand, etc.), or changing the colour of the sample
(whether that is through spillage, long exposure to sunlight, or other), you will
always receive the same alphanumeric code. Essentially the scanner creates an
electrical response, and this response is stored by the computer. The computer
assigns a particular alphanumeric code, either based on a fixed procedure or
based on a table, just like we might use a table of values of f (x) to determine
f (3). So the same colour will never be linked with more than one alphanumeric
code. If it was not this way, we could not depend on a computer. The benefit
of a function is that there is the consistent output.

This is where the idea of the vertical line test comes in. If a horizontal line
can go through more than one point on a curve, it means that the output for
a few different inputs is the same, and this is fine. However, if a vertical line
can go through more than one point, there is more than one output. When
you make a table of values, you are writing down the output for each input.
You cannot write more than one possibility for each input. Even if you have
a function that works on probabilities, it still only has one output–it is just
that the output is the probability for each possible outcome, and that output
of probabilities will be the same every time for it to be a function.

Note that if a function has multiple variables as its input, all of the variables
have to be the same for the input to be the same. For right now, you will deal
with functions of single variables with single variable outputs.

2 Parent Functions and Transformations

One thing we have been doing is building functions from basic “parent” func-
tions. The idea is that we understand the basic pattern from the parent function,
and we then transform the pattern using reflections, stretches, and shifts.

Let us look at the following function.

1
f (x) = (3)
x
Let us graph y = f (x), i.e.
1
y= (4)
x

3
We make a table of values. We try to determine the lines that cannot be crossed
by calculating the values we cannot use or obtain. What if x = 0? We end up
dividing by zero. This means we cannot cross x = 0. So we make sure we
have coordinates in our table of values that use x-values close to zero to observe
the trend. How about y = 0? This would require x to be infinite (positive or
negative), so we cannot cross y = 0. So we make sure we have coordinates in
our table of values that yield y-values close to zero to observe the trend. We
try to see if there are any other values that the function cannot handle (like
crashing a computer program). Once we are satisfied with our table of values,
we plot the coordinates on the graph.

https://www.wolframalpha.com/input?key=&i=y+%3D+1%2Fx

Once we develop experience working with a function and understand its trends,
we build intuition on the pattern between x and y, and we can graph with little
help from a table of values. We may just need to check the exact y-value of a
few points as we are going along in our sketch.

Say we want to graph the following:

1
y= +2 (5)
x
However, if we already know how to graph y = x1 , we can graph it, and then
shift it. We will have the same trend, except all the y-values will be greater by
2. So we can shift all the points up by 2. Now y could be 0 (i.e. x1 = −2, but y
cannot be 2 (i.e. x1 ̸= 0. We could write this as:

y = f (x) + 2 (6)

Say instead we want to graph the following:

1
y= (7)
x−3
Again, we could use a table of values. However, like in the previous case, we
already know how to graph y = x1 , so we can graph it and then shift it. To
arrive at the same trend, the x-values will need to be 3 units greater than in
the graph of y = x1 , i.e. the parent function. So we can shift all the points to
the right by 3. Now x − 3 cannot be zero, so x can be zero, but x cannot be 3.
We could write this as:
y = f (x − 3) (8)
Let us now look at a different function:

g(x) = sin x (9)

We can figure out the sine ratio of any angle from trigonometry. We can even
think of there being a table of values stored in one’s calculator (although there

4
are ways to approximate a sine ratio using algebra). If we graph y = g(x), we
see an oscillating function. The period is 360◦ (or 2π radians), and the y-value
goes between -1 and 1.

https://www.wolframalpha.com/input?key=&i=y+%3D+sin+x

Now let us graph the following.

y = 5 sin x (10)

We could create an entirely new table of values, but essentially we are taking
the y column in the original table of values and multiplying it by 5. We are
stretching the function by a factor of 5; it will oscillate vertically between -5
and 5 rather than -1 and 1, but the period will still be 360◦ (2π radians). We
can write this as:
y = 5g(x) (11)
Now let us graph a different type of stretch.

y = sin 4x (12)

Again, we could create an entirely new table of values, but essentially we are
taking the x column in the original table of values and multiplying it by 41 . Why
1
4 ? We are multiplying the x-values (angles) by 4 before the sine ratio is applied
to them, and so each x-value will be 14 of the original x-value that corresponds
with the same y-value. This means we are stretching the function by a factor of
1
4 (or compressing the function by a factor of 4); it will still oscillate vertically
between -1 and 1, but the period will be 90◦ ( π2 radians, 41 of the original period).
We can write this as:
y = g(4x) (13)
Let us look at one more function. Note that this function will come up in the
section on inverses. √
q(x) = x (14)
If we graph y = q(x), we will see that the curve goes from (0, 0) up and then
curving to the right (i.e. a sideways half-parabola). The curve is entirely in the
top right quadrant of the Cartesian plane.

https://www.wolframalpha.com/input?key=&i=y+%3D+sqrt%28x%29

What if we wanted to draw out the same curve, reflected about the x-axis?
It would be a half-parabola curving down and right in the bottom right quad-
rant of the Cartesian plane. We essentially want to multiply all the y-values by
-1. Just like y = 5g(x) multiplies the outputs of the original g by 5, y = −q(x)
means multiplying all the y-values of q by -1.

y = −q(x) (15)

5
 √
y=− x (16)
https://www.wolframalpha.com/input?key=&i=y+%3D+-sqrt%28x%29

Now what if instead we wanted a half-parabola curving up and right in the

top right quadrant of the Cartesian plane? We essentially want to multiply all
the x-values in the original curve by -1. Just like y = g(4x) multiplies the inputs
to g by 14 , y = q(−x) means multiplying all the x-values of q by −11
, which is -1.

y = q(−x) (17)
√
y = −x (18)
https://www.wolframalpha.com/input?key=&i=y+%3D+sqrt%28-x%29

As you may have guessed, for a general function f (x), we can combine vertical
reflections and stretches in a, i.e. y = af (x). We can also combine horizontal
reflections and stretches in b, i.e. y = f (bx). The magnitude of a and b are the
stretches, and a negative sign signifies a reflection.

For some functions, stretches and reflections do not have unique effects. For
example, if g(x) = sin x,
g(−x) = −g(x)
since
sin(−x) = − sin x (19)
This is not true for all functions. Another interesting example is in the case of
horizontal and vertical stretches of parabolas, as is described in the following
section. There are also cases where certain transformations have no effect. For
example, if p(x) = cos x,
p(−x) = p(x) (20)
since
cos(−x) = cos x (21)
The same is true for horizontal reflections in vertical parabolas, as is shown in
the next section.

I am going to set up a word problem (and analogy) that illustrates the basic
transformations of functions. The word problem does not illustrate reflections,
but as explained above, reflections are dictated by the same variables that dic-
tate stretches.

We have a high school that is raising money to help supplement the diet of
children in Malawi with vitamin A. The supplement is not made in a lab–rather
it is 50g of carrots. We have a function f (x) that maps the amount of money the
school raises per month to the 50g units of carrots delivered to the children in
Malawi. You may wonder why this is not a simple linear (proportional) relation-
ship. In short, if there is enough money, rather than solely purchasing carrots

6
from farmers, farming equipment can be purchased and leased to the farmers,
and eventually land can be purchased directly to be farmed. This means the
amount of carrots that can be purchased with each dollar goes up if there is
enough money. A table of values for y = f (x) is shown as follows.

Table 4: Table of values for the relationship between the money spent and the
units of carrots delivered.
x y = f (x)
100 3000
1000 30000
5000 90000
20000 240000
100000 2500000

Now, we will apply some transformations. Some of those happen on the money
(input) end, and the others will happen on the carrot (output) end. On the
money end, let us assume that the school donates through a website of an or-
ganisation. Let us assume that the organisation charges a flat administrative
fee h for each donation through the website. This means that effectively the
money going towards the endeavour is x − h.

y = f (x − h) (22)

If we are graphing money on the x-axis and 50g units of carrots per month on
the y-axis, the fee shifts the curve h units to the right, since for the each amount
of carrots, we need the same amount of money plus h to cover the fee.

Table 5: Table of values for the relationship between the money spent and the
units of carrots delivered, accounting for the flat fee on the donation.
x y = f (x − h)
100 + h 3000
1000 + h 30000
5000 + h 90000
20000 + h 240000
100000 + h 2500000

Now, let us say that the federal government decides to match amount of money
donated, calculated after the fee is processed. The government will not just
match the amount exactly–it will match so that the final (total) amount donates
is b times the amount we donate (again, calculated after the fee is processed).
So if we pay x, we donate (x − h) after the fee, and this will mean a total
donation of b(x − h). (If the government was matching dollar for dollar, b would
be 2.) So after the fee is subtracted, we need to raise 1b times the amount of
money we want sent, since the government match will lead to the final amount

7
we want. The x-values, after h is subtracted, will be compressed by a factor
of b/stretched by a factor of 1b . This means they will be multiplied by 1b . The
y-values in each line in the table will remain the same.

y = f (b(x − h)) (23)

Table 6: Table of values for the relationship between the money spent and the
units of carrots delivered, accounting for the flat fee on the donation and the
government matching.
x y = f (b(x − h))
100
b + h 3000
1000
b + h 30000
5000
b +h 90000
20000
b + h 240000
100000
b +h 2500000

Now, let us consider logistics on the carrot end. Let us say that the University of
Malawi discovers a special organic fertilizer mix that increases the carrot yield
of every carrot field by a factor of a. This means that for the same amount
of money being donated, the amount of carrots will be increased. This change
is not a change in the final amount of money going into the endeavour; it is a
change in the amount of carrots coming out. For this reason, the y-values in
the table will be multiplied by a, while their associated x-values will remain the
same.
y = af (b(x − h)) (24)

Table 7: Table of values for the relationship between the money spent and
the units of carrots delivered, accounting for the flat fee on the donation and
the government matching, and accounting for the increased yield due to the
fertilizer.
x y = af (b(x − h))
100
b + h 3000a
1000
b + h 30000a
5000
b +h 90000a
20000
b + h 240000a
100000
b +h 2500000a

Finally, the school forms a friendship with a very kind farmer who agrees to
donate k 50g units of carrots to children in the area. His land’s ecology is very
sensitive, and so he is unable to use the fertilizer described above; this donation

8
of carrots is a fixed amount per month, unaffected by a. For this reason, a fixed
amount k is added to every y-value, and the x-values are unaffected.
y = af (b(x − h)) + k (25)

Table 8: Table of values for the relationship between the money spent and the
units of carrots delivered, accounting for the flat fee on the donation and the
government matching, and accounting for the increased yield due to the fertilizer
and the friendly fixed donation.
x y = af (b(x − h)) + k
100
b + h 3000a + k
1000
b + h 30000a +k
5000
b + h 90000a +k
20000
b + h 240000a +k
100000
b + h 2500000a +k

If you substitute the coordinates of any of these tables into the equation gov-
erning the respective table (indicated in the expression for y in each table), you
will see that f itself remains the same; it still follows the pattern in Table 4.

Now do not forget that h and k can also be negative, which would lead to
shifts to the left and down respectively. Rather than writing (x + h) or −k, we
allow h and k to be negative, and the algebra works out. A change in h and k
can lead to a shift of the domain and range respectively.

Similarly, a and b can be less than 1 in absolute value, and they can be negative
too. As described above, if a horizontal stretch is negative, it leads to a horizon-
tal reflection (reflection about the y-axis, i.e. x = 0), and if a vertical stretch is
negative, it leads to a vertical reflection (reflection about the x-axis, i.e. y = 0).
If |b| < 1, since the x-values are being stretched by 1b , they are increased. If
|a| < 1, since the y-values are being stretched by a, they are decreased. As
long as a and b are not zero, the algebra for the stretches works out. A change
in b and a can lead to a stretch or reflection of the domain and range respectively.

There is one more important point I should mention. Often folks mention why
x and y are so asymmetric in Equation 25. Perhaps they are not:
y − k = af (b(x − h)) (26)
Now it makes more sense algebraically that h leads to larger x-values and k
leads to larger y-values: x and y have to increase by h and k respectively to
make up for the decrement in their respective parts of the equation.
y−k
= f (b(x − h)) (27)
a

9
Had we instead written
1
y= f (b(x − h)) + k
R
and rearranged this to write

R(y − k) = (b(x − h))

1 1
R now takes on a similar role to b. Since we use a rather than R or a in front
of f , y is stretched by a while x is stretched by 1b .

3 Plotting Transformations of Functions

There are several ways to plot transformations of functions. It really depends
on the information you are given and your understanding of the particular func-
tion. We will focus on graphing in Cartesian coordinates in the Cartesian plane
(i.e. using (x, y))).

We will look at linear transformations, i.e. transformations of the form in

Equation 25:
y = af (b(x − h)) + k
where f is the parent function and a, b, h, and k are all constants. We will not
look at graphing other transformations, such as

y = f (x2 )

(In this particular case, I would recommend thinking of it like a function of a

function, and I would recommend using a table of values to feed the x2 values
into f .)

First we will consider situations where you are given a function f as a drawing
or table of values, and you have to graph y = af (b(x − h)) + k. In other words,
you are given a graph of y = f (x), or you are given a table of values for x
and y that represents f (x) (the pattern f mapping x to y). Either way, you are
going to apply stretches to the key points, then shift those points. What are the
key points? If you are given a graph. The key points include all points whose
coordinates are clear as well as all the points that illustrate the graph’s trend
or mark its trend. For example, if you have a circular arc that joins with a line
segment or a line segment that joins with a steeper line segment, the points of
intersection are key points. If there is a point where the graph suddenly starts
to level off, it is a key point. If you are given a table of values, all of the coordi-
nates are key points. To apply stretches, you multiply all of the x-values of the
key points by 1b and all of the y-values of the key points by a. You do these two
multiplications at the same time while plotting a corresponding point for each
key point. Every key point (xk , yk ) becomes ( xbk , ayk ). You have now plotted:

y = af (bx)

10
Now we apply the shifts. Remember the format in Equation 25, and calculate
h and k based on that. So if you are graphing y = f (x + 4), h = −4. Once
you have calculated h and k, take the points from your graph of y = af (bx), i.e.
not the graph or table of values of the parent function, and shift them right h
units and up k units.

You may ask why we do the horizontal shift after the horizontal stretch. When
we make the shift, we are not graphing x − h. We are graphing the new x that
is required so that the x − h that goes into the horizontal stretch is the same as
the previous value of x (which was already stretched by b).

Say instead you are graphing from an equation. It could be f (x) = x1 or

g(x) = x2 , and you are trying to graph y = 4f (x + 1) − 3 or y = −g(−x + 2) + 5.
There are three main approaches. The first approach is graphing the parent
function, then transforming it as above. This is especially efficient when you
are familiar with the parent function. For example, if you are familiar with
f (x) = x1 , you know that the function passes through (1, 1) in the quadrant
where both x and y are positive, and it passes through (−1, −1) in the quadrant
where both x and y are negative. Furthermore, in each of those two quadrants,
the curve extends to approach but not cross both asymptotes: x = 0 and y = 0
(the y-axis and the x-axis respectively). Knowing these facts, it is quite easy to
sketch y = f (x). If you are unfamiliar with f , this may not come as easily, and
you might need a table of values. Likewise, if you are familiar with g(x) = x2 ,
you might think in your head, “9-4-1-0-1-4-9,” which correspond with the y-
values for x = −3 to x = 3. In either case, if you plot the parent function,
drawing the transformed function in two steps (first y = af (bx), then the shift)
is relatively easy. Note that if you have asymptotes, shift the horizontal asymp-
totes (i.e. y = ) up by k, shift the vertical asymptotes (i.e. x = ) right by h,
and shift any other straight-line asymptotes by shifting all points on the line
right by h and up by k (in practice, you graph a straight line using only two
points, so this is not much work).

The second approach is writing a table of values for the transformed function.
First you write out:
y = a [...] + k

Next, on the dots, write out the function, except wherever you see x, write
(...). So for the example of graphing y = ag(b(x − h)) + k for g(x) = x2 , we
wouldwrite:
y = a (...)2 + k

Finally, we write b(x − h) wherever we see the new dots from the very last
step. Now you have the equation of the transformed function, and you can
write out a table of values in a “brute force” way. As you make your table of
values, plot each point, so that you can verify you have enough points to capture
the trend.

11
The third approach depends on the particular function. You take the second
approach, but before you make the table of values, you simplify as necessary to
make the table of values easier. This can also help in understanding the trend.
We will illustrate with the following example:
1
f (x) =
x
1
y = 3f ( x + 1) + 1
2
Before we start, we have to put it in the form in Equation 25. This means
factoring out b in the input to the function:
1
y = 3f ( (x + 2)) + 1
2
We write:
y = 3(...) + 1
This could have been combined with the next step, where we put in the function,
leaving room where x was:
1
y = 3( )+1
(...)
Finally, we put b(x − h) wherever x was:
1
y = 3( )+1
( 12 (x + 2))

What differentiates this third approach is that we manipulate the algebra a bit
further. In this case, we bring the 12 in the denominator to the numerator:

2
y = 3( )+1
x+2
Multiply the a (a = 3) with the fraction:
6
y= +1
x+2
Now we can make a table of values. Or we could even rewrite the graph as
y = 6f (x + 2) + 1.

This leads into another idea: sometimes b can be made to be a part of a (for par-
ticular functions). This is demonstrated
√ in the following section for g(x) = x2 .
Consider the function q(x) = x. How can we rewrite y = −3q(−4x − 4) + 2?
First we factor out b to put the function in the form of Equation 25

y = −3q(−4(x + 1)) + 2

12
Next we leave space for the actual function.

y = −3(...) + 2

We put in the actual function and leave space for the input:
p
y = −3 (...) + 2
p
y = −3 −4(x + 1) + 2
From here, we could put the 3 into the square root, re-writing it as 9 = 32 inside
the square root: p
y = −3 −4(x + 1) + 2
p
y = − 9(−4)(x + 1) + 2
p
y = − −36(x + 1) + 2
More often
√ than not, we actually take the 4 out of the square root, re-writing
it as 2 = 4 on the outside:
p
y = −3 −4(x + 1) + 2
p
y = −3(2) −(x + 1) + 2
p
y = −6 −(x + 1) + 2
Notice we did not put the negative from the outside into the square root, nor
did we take the negative inside the square root out. The reasons are as follows:
If you put the negative sign in, it will become (−1)2 = 1, which will do nothing.
The square root function always yields positive values, and so the only way to
make the final function negative is to put a negative sign in front of the square
root. A negative a is a reflection about the x-axis; the values of y before the
shift become negative. You cannot take the negative sign out. We cannot cal-
culate the square root of the negative number with real numbers. The presence
of the negative sign inside the square root sign means that x − h must be neg-
ative, so that b(x − h) is positive. A negative b is a reflection about the y-axis;
the values of x before the shift become negative.
√
By the way, when working with y = x, you do not have asymptotes, in that
there is no line that the graph
√ approaches but does not ever touch. However,
you should remember that x cannot cross the x-axis (because the square
root function always yields positive numbers or zero) or the y-axis (because you
cannot take the square root of a negative number), but it does touch the
x- and y- axes, i.e. at (0, 0). For this reason, you can think of there being
‘boundaries” (not a formal term) of x = 0 and y = 0. This is worth noting
when studying transformations, since the boundaries are shifted to x = h and
y = k. Furthermore, a reflection takes the graph to the opposite side of the
corresponding boundary, as explained in the above paragraph.

13
We find boundaries when working with other functions too. y = sin x and
y = cos x do not go higher than y = 1 or lower than y = −1. y = sin2 x and
y = cos2 x do not go higher than y = 1 or lower than y = 0. These boundaries
are shifted by k.

Graphs like y = log x and y = ln x have restrictions at x = 0, since we cannot

take the logarithm of 0 or a negative number (at least not with real numbers),
and so x = 0 is an asymptote. The asymptote will be shifted by h. Graphs
like y = 10x , y = 2x , and y = ex are always positive; they have a horizontal
asymptote of y = 0. The asymptote will be shifted by k. In whatever function
you are examining, understanding the asymptotes and boundaries of the parent
function will help with graphing transformations.

Say you are not given the parent function and transformation instructions, but
instead you are given the transformed function. How can you graph this? For
example, how can you graph

y = 2x2 − 5x + 2 (28)

or
3x − 2
y= (29)
2x − 3
In the case of Equation 28, read the following section, in which it is shown that
putting the equation into vertex form (i.e. completing the square) makes it
easy to graph. The parent function is g(x) = x2 , and the required stretches and
shifts are very clear when using vertex form.

As for Equation 29, there is a lot to observe. First of all, we can easily cal-
culate both asymptotes. The vertical asymptote occurs for this function when
we divide by zero (as shown above, like in the example of y = log x, there
are other ways to have vertical asymptotes); the vertical asymptote is when
2x − 3 = 0.
2x = 3
3
x= (30)
2
As for the horizontal asymptote, consider when x becomes extremely large in
either the positive or negative directions. When this is the case, we can approx-
imate Equation 29 as:
3x
y=
2x
When x is very large in magnitude, it is clearly not zero, and so we can divide
both the numerator and denominator by x (reduce the fraction):
3
y= (31)
2

14
Note that this would not have been possible if there were terms outside the
fraction, such as in y = 3x−2
2x−3 + 1
It just so happens that the horizontal and vertical asymptotes have the same
value–this is generally not the case. Now, we can either mark the asymptotes
and plot the rest of the graph from a table of values, or we can express Equation
29 as a transformation of the parent function y = x1 . We will do the latter here:

3x − 2
y=
2x − 3
Factor out 3 from the numerator and 2 from the denominator.

3 x − 23
 
y=
2 x − 32

Add and subtract a fixed amount to the numerator in order to force the first
part of the numerator to match the denominator, in the following way:

3 x − 32 + 32 − 32
 
y=
2 x − 32

Re-write this as follows:

3 2
−
 
3 2 3
y= 1+ 3
2 x− 2
3
Distribute the 2 outside the brackets and simplify:
3 2
−

3 3 2 3
y= + 3
2 2 x− 2
9 4
−

3 3 6 6
y= + 3
2 2 x− 2
 5 
3 3 6
y= + 3
2 2 x− 2
5
Factor out the 6 from the fraction, and rearrange:

5 1 3
y= · 3 + (32)
4 (x − 2 ) 2

a = 45 , b = 1, h = 32 , and k = 32 . Again, it just so happens that both h and

k are the same; this is not usually the case.. In this form, it is clear that the
vertical asymptote is x = 32 , since that would lead to division by zero. In this
form, we find the horizontal asymptote by setting x to be as large as possible
(approaching infinite). This means that if x is only in the denominator (not in
the numerator or in front of the fraction), setting the fraction to zero will give
you the horizontal asymptote y = 32 . Alternatively, since k = 32 , and we know

15
the horizontal asymptote for the parent function is y = 0, we know that the
asymptote of the transformed function is y = 32 .

Try doing this process on your own for:

5x − 3
y= (33)
4x + 4

4 Why do we complete the square?

In the previous section, we went over the use of transformations of parent func-
tions. In this section, I will focus on quadratic functions. We know basic pattern
of the function g(x) = x2 . We write out a table of values for x from 0 to 4. We
see 0, 1, 4, 9, 16... We then do the negative side (i.e. x from 0 to -4) and find the
same pattern. We can then apply reflections, stretches, and shifts pretty easily.
The problem is that quadratic functions generally have the form Equation 34.
y = ax2 + bx + c (34)
DISCLAIMER: variables like b are used in slightly different ways when looking
at quadratic equations and functions than when looking at the general function
notation used in the previous section.

It is much more difficult to sketch out a function when it is a combination

of more than one parent function, rather than the transformation of a single
parent function. The term bx is making our lives difficult. Fortunately, we
know something special about the square of binomials (x + β). Expand the
following using FOIL, i.e. First Outside Inside Last:
(x + 1)2 = (x + 1)(x + 1)
(x + 1)2 = x2 + (x)(1) + (1)(x) + 12
(x + 1)2 = x2 + 2x + 1
Expand the following expressions using the same techniques:
(x + 2)2 = x2 + 4x + 4
(x + 3)2 = x2 + 6x + 9
(x + 0.5)2 = x2 + 0.5x + 0.5x + 0.25 = x2 + x + 1
(x + 1.5)2 = x2 + 3x + 2.25
(x + 2.5)2 = x2 + 5x + 6.25
(x + 0.25)2 = x2 + 0.5x + 0.0625
There is clearly a pattern. We expand a square of x with a shift, and we are
able to create whatever x term in the quadratic that we want.
(x + β)2 = x2 + 2βx + β 2 (35)

16
2β can be anything. Let us try to work backwards from the x term. This
method will be called “completing the square”. We will work from Equation 34:
y = ax2 + bx + c
 
2 b
y =a x + x +c
a
   
b
y = a x2 + 2 x +c
2a
   2  2 !
2 b b b
y =a x +2 x+ − +c
2a 2a 2a
   2 !  2
2 b b b
y =a x +2 x+ −a +c
2a 2a 2a
 2  2 
b b
y =a x+ −a +c
2a 4a2
2
b2

b
y =a x+ − +c (36)
2a 4a
Now we just have the parent function g(x) = x2 with a vertical stretch and/or
b b2
reflection a, a horizontal shift − 2a to the right, and a vertical shift of − 4a +c
up. Using the format of the general transformed function in Equation 25,
b b2
y = ag(x + )− +c (37)
2a 4a
2
b b
Therefore, h = − 2a , and k = − 4a + c.

Note that we could apply a horizontal stretch α (because b is already taken) to

the parent function, but this ends up being redundant:
y = g (α(x − h)) (38)
Write out the actual square:
2
y = (α(x − h)) (39)
As you can see, we can take the α outside the brackets, and when we do so, it
becomes squared.
y = α2 (x − h)2 (40)
The α2 stands where there would normally be a, and so we group this aspect
of the stretch with the vertical stretch a. Essentially, due to the geometry of
a vertical parabola, a horizontal stretch by α1 (or a horizontal compression by
α is a vertical compression of α2 . Also note that there cannot be the usual
horizontal reflection because the parent function (parabola) is symmetric about
the y-axis. This works out in the algebra because if α is negative, the negative
sign is cancelled when it is squared in Equation 40.

17
5 Composite Functions–Functions of Functions
A composite function is the function of a function. It describes the pattern that
results when the output of one function is fed into another function as an input.
Say we are looking at the functions f (x) = x2 and g(x) = sin x. Let us build a
composite function (f ◦ g).

(f ◦ g)(x) = f (g(x)) (41)

(f ◦ g)(x) = (sin x)2 (42)

2 2
For the sine function, we write (sin x) as sin x. This is done often but not
always. The input to (f ◦ g)(x) is an angle. The angle is an input to the sine
function, which spits out a value between -1 and 1. That sine ratio output is
the input to the quadratic function f , and the output of that is between 0 and
1. Thus, the output to (f ◦ g)(x) is a value between 0 and 1.

The function (f ◦ g) is different than g ◦ f , since

(g ◦ f )(x) = g(f (x)) (43)

(g ◦ f )(x) = sin(x2 ) (44)

2 2
(sin x) ̸= sin(x )
Except in very particular cases,

(f ◦ g)(x) ̸= (g ◦ f )(x) (45)

6 Inverses
When we first hear about “inverse” functions, we think, “Switch x and y.” Sure,
but this does not tell the whole story.

An inverse function f −1 maps/links the output of the original function back

to its corresponding input that went into the original function. For this reason,
we do switch the x- and y- values. We have to be careful though. If the origi-
nal function had more than one x-value that was mapped to the same y-value,
our inverse relation will not be a function. It will attempt to map at least one
x-value to more than one y-value. In other words, if the original function could
not pass a “horizontal line test”, switching the x- and y- values would make a
relation that fails the vertical line test and thus is not a function.

There are a few ways to look at graphing the inverse. The simplest way is
to write out the table of values from the original function in reverse. Another
way is considering that the inverse is a reflection about the line y = x. A vertical
reflection is a reflection about y = 0, i.e. the x-axis, and a horizontal reflection
is a reflection about x = 0, i.e. the y-axis. When graphing a reflection in the

18
Cartesian plane, it helps that the line about which you are doing the reflection
is on the graph. Draw y = f (x), and draw y = x. This way, when you draw
y = f −1 (x), you have a reference for the reflection.

A third way is to write out an algebraic expression for f (x), write y = f (x),
switch x and y, and finally isolate y, arriving at y = f −1 (x). Here is an example:

f (x) = 2x3 − 3 (46)

y = 2x3 − 3
Switch x and y:
x = 2y 3 − 3
x + 3 = 2y 3
x+3
= y3
2
r
3 x + 3
y=
2
r
3 x + 3
f −1 (x) = (47)
2
Many common parent functions would fail a “horizontal line test”. One example
is f (x) = x2 . Let us try to determine the inverse using the algebraic method
used to find Equation 47:
x = y2
When we take the square root to isolate y, we need to write ± to account for
the possibility of negative y-values.
√
y=± x (48)

This is not a function because there are two y-values for every x-value other
than x = 0. We would usually choose
√
f −1 (x) = x (49)

but we could also choose: √

f −1 (x) = − x (50)
√
Another way to think about this is that fR−1 (x)
= x is the inverse of the right
branch of f (x) = x2 , i.e.
fR (x) = x2 , x ≥ 0
√
and fL−1 (x) = − x is the inverse of the left branch of f (x) = x2 , i.e.

fL (x) = x2 , x ≤ 0

19
Let us now consider the function g(x) = sin x. Like f (x) = x2 , g(x) would
not pass the “horizontal line test”. This means we will have similar issues in
determining an inverse.
x = sin y
y = sin−1 x
g −1 (x) = sin−1 x (51)
This can also be written g −1 (x) = arcsin(x). You may notice that this function
on a calculator always gives you an answer between -90◦ and 90◦ . This can be
problematic when determining angles of a triangle using Sine Law. For example,
sin 30◦ = 0.5, but sin 150◦ = 0.5 as well. In fact, there are an infinite number
of angles whose sine ratio equals 0.5, and these are all excluded. If you look
at the graph of y = arcsin x, you will see that vertically (i.e. range) it goes
from -90◦ to 90◦ , and it is the inverse of y = sin x from -90◦ to 90◦ on the x-axis.

https://www.wolframalpha.com/input?key=&i=y+%3D+arcsin%28x%29
√
Be mindful when using an inverse function like or arcsin, since you may
need to make adjustments to arrive at the answer you seek, i.e. the correct
choice from the many inputs to the original function that gave the same output.

Remember that restrictions on the range of a function dictate domain of the

2
√ the range of y = x is y ≥ 0, which corresponds
inverse. For example,
2
with
the domain of y = x being x ≥ 0. Likewise, √ the range of y = −x is y ≤ 0,
which corresponds with the domain of y = −x being x ≤ 0. This fact plays a
role when you apply a function to its inverse or vice versa (as will be explained
shortly).

The notation for the inverse of a function appears to be similar to the nota-
tion for the products of powers in that:

(f −1 ◦ f )(x) = (f ◦ f −1 )(x) (52)

and
(f −1 ◦ f )(x) = x (53)
This is similar to how a · a−1 = a−1 · a and a · a−1 = 1.

(f ◦ f −1 ◦ f −1 ◦ f )(x) = x

(f ◦ f ◦ f −1 )(x) = f (x)
f −1 cancels out the effect of f and vice versa, but this yields the input x rather
than 1.

If we apply the function twice, we do NOT write it with a superscript of 2,

i.e. f 2 , we write (f ◦ f ). Likewise, if we apply the inverse twice, we do NOT

20
write it with a superscript of -2, i.e. f −2 ; we write (f −1 ◦ f −1 ). If anything,
2
f 2 (x) might mean (f (x)) . The square of the inverse would not be written
f −2 (x), although in principle we could write arcsin2 x (yet not sin−2 (x)) to
mean (arcsin x)2 .

In a composite function solely made up of instances of f and f −1 (any number

of each), their order does not matter, except when we run into the aforemen-
2
tioned issue
√ with restrictions. Here is an example. Say we write f (x) = x and
−1
f (x) = x. Without indicating restrictions on the domain of f , we can arrive
at strange results: p
(f −1 ◦ f )(−7) = (−7)2
(f −1 ◦ f )(−7) = 7 (54)
This goes against Equation 53. How about the reverse order?
√
(f ◦ f −1 )(−7) = ( −7)2 (55)

We cannot take the square root of a negative number, at least not at this grade
level. So not only does this go against Equation 53, but it also yields a different
result than Equation 54, which goes against Equation 52.

Here is another example. Say we write g(x) = sin x and g −1 (x) = arcsin x.
Without indicating restrictions, this is what can happen:

(g −1 ◦ g)(150◦ ) = arcsin(sin(150◦ ))

(g −1 ◦ g)(150◦ ) = arcsin(0.5)
(g −1 ◦ g)(150◦ ) = 30◦
(g ◦ g −1 )(150◦ ) = sin(arcsin(150◦ ))
arcsin(150◦ ) does not even make sense, since arcsin can only take in a value
between -1 and 1 (i.e. a value that would correspond with a sine ratio). Even if
◦
the angle was not written in degrees, 5π
6 (= 150 ) is outside the domain of -1 to 1.

Consider the general case of calculating Equation 47. How does one calculate
the inverse of a transformed function as in Equation 25?

y = af (b(x − h)) + k

Switch x and y.
x = af (b(y − h)) + k
x − k = af (b(y − h))
1
(x − k) = f (b(y − h))
a

21
Take f −1 of both sides:
 
1
f −1 (x − k) = b(y − h)
a
 
1 −1 1
f (x − k) = y − h
b a
 
1 1
y = f −1 (x − k) + h (56)
b a
As you can see, Equation 56 corresponds with Equation 25. The horizontal shift
and vertical shift switch. The horizontal stretch and vertical stretch switch, and
their reciprocals are taken.

22