last edited April 30, 2016

6.2 Modular Arithmetic

Every reader is familiar with arithmetic from the time they are three or four
years old. It is the study of numbers and various ways in which we can combine
them, such as through addition and subtraction, multiplication and division.
Since even before they were in grade school, every reader knew that adding 2
and 2 together gives us 4, and can make that calculation now without almost
any thinking. And even if the answer is not immediately obvious, every college
student (at least in Penn), knows how to add together much larger numbers,
such as 4,378,123 and 5,621,877. This is classical arithmetic, and it turns up in
countless applications in our everyday lives.
The reader is also likely familiar with another kind of arithmetic, even if we
don’t always think of it as such. If it is 4 o’clock now, what will the time be
in 25 hours? If we didn’t know from watches and clocks, we would probably
have answered 29 o’clock. But we are familiar with watches, clocks, and the
standard conventions of time-keeping, and so every reader would probably have
answered the answer with 5 o’clock. How can we add 25 to 4 and end up with
5? The reason is that in this system 25 o’clock is the same as 1 o’clock, 26 is
the same as 2, and so forth. In many time-keeping systems, we don’t even use
numbers larger than 12, and instead use a.m. and p.m. (from the Latin ante
meridiem and post meridiem) to denote the earlier and latter halves of a 24-hour
period. Such systems, that “wrap around” after hitting some limit, are called
modular arithmetic systems, and play an important role both in theoretical
and applied mathematics.
Modular arithmetic motivates many questions that don’t arise when study-
ing classic arithmetic. For example, in classic arithmetic, adding a positive
number a to another number b always produces a number larger than b. In
modular arithmetic this is not always so. For example, if it is now 4 o’clock and
we “add” 23 hours, the time will then be 3 o’clock, which doesn’t appear to be
larger than 4 o’clock. In fact, it is no longer clear whether it makes sense at all
to discuss “larger” and “smaller” in such systems.
Here is another question. Suppose it is now 2 o’clock and we wait for 1 hour
and then write down the time. We then wait another hour and mark the time,
and repeat this until we eventually mark 2 o’clock again, at which point we
stop. It is clear that when we stop, we will have marked down every hour. If we
do the same thing but instead wait 2 or 3 hours in between each marking there
will be certain hours which we never mark, such as 7 o’clock. But if we wait
5 hours between each marking, then we will eventually mark every hour. This
raises the question, for which waiting intervals between marks can we ensure
that we will eventually mark every hour?
While this particular example may seem contrived, it should motivate us
think, if even momentarily, about modular arithmetic systems and the ways in
which they are similar to and di↵erent from the classical arithmetic with which
we are familiar. The next several sections will investigate these systems which
have a finite number of numbers, and in which numbers “wrap around” after
going too high.

72
 last edited April 30, 2016

The central definition in studying modular arithmetic systems establishes a

relationship between pairs of numbers with respect to a special number m called
the modulus:
Definition 25. Two integers a and b are congruent modulo m if they di↵er
by an integer multiple of m, i.e., b a = km for some k 2 Z. This equivalence
is written a ⌘ b (mod m).
Although this definition looks somewhat technical, the idea is very simple.
For some fixed integer m, two numbers are roughly the same if they di↵er by
multiples of m. In a sense, this definition generalizes previous discussions of
odd and even numbers. In previous sections, we proved theorems such as the
square of an even number is even and the square of odd number is odd. As
far as even and odds numbers go, and as far as these theorems are concerned,
there is no di↵erence between 17 and 2073, as both are odd and behave the
same under squaring. In a similar manner, in modular arithmetic, there is no
di↵erence between a pair of numbers that di↵er by the modulus m, which could
be 2 or could be 15,485,863. In arithmetic mod 7, for example, there is no
di↵erence between 1, 8, and 15, as they all di↵er from one another by multiples
of 7. Likewise, 22, 701 and -6 also di↵er from all of these numbers by multiples
of 7, and are hence congruent.
Example 1. Every number is congruent to itself for any modulus; that is, a ⌘ a
(mod m) for any a, m 2 Z. The reason for this is that a a = 0, which is a
multiple of m, since 0 = 0 ⇥ m for any m. It might seem a bit silly, but is a
consequence of the way in which we defined congruence.
Example 2. Every number is congruent to any other number mod 1; that is,
a ⌘ b (mod 1) for any a, b 2 Z. The reason for this is that b a, is a multiple
of 1 for any a and b. Again, this might seem a bit silly, but is a consequence of
the way in which we defined congruence.
Example 3. Any even numbers are congruent to one another mod 2; likewise,
any odd numbers are congruent to one another mod 2. For example, we have
12 ⌘ 3132 (mod 2) and 7 ⌘ 19 (mod 3). This is because any pair of even
numbers di↵er from one another by a multiple of 2. Likewise, any pair of odd
numbers di↵er from one another by a multiple of 2.
Example 4. The numbers 31 and 46 are congruent mod 3 because they di↵er
by a multiple of 3. We can write this as 31 ⌘ 46 (mod 3). Since the di↵erence
between 31 and 46 is 15, then these numbers also di↵er by a multiple of 5; i.e.,
31 ⌘ 46 (mod 5).
Example 5. By the definition of congruence, every pair of integers a and b are
congruent mod 1, since any pair of integers di↵er by a multiple of 1. In symbols,
for all integers a and b, we have a ⌘ b (mod 1).
Example 6. In general it is not true that a ⌘ a (mod m), unless m = 2 or
else a is a multiple of 2. For example, it is not true that 7 ⌘ 7 (mod 3), since
the di↵erence between 7 and -7 is 14, which is not a multiple of 3.

73
 last edited April 30, 2016

Rules of Modular Arithmetic

After considering the basic definition of modular arithmetic, we next consider
some of its basic properties. It turns out that modular arithmetic follows many
of the same rules of classical arithmetic, thus making it very easy to work with.
In order to highlight what is going on, we try to compare and contrast modular
arithmetic to classical arithmetic.
Suppose we have two numbers a and b:

a = 5
b = 8.

We all know that in classical arithmetic we can combine these equations to

obtain:

a + b = 5 + 8 = 13.

More generally, if we have

a = c
b = d,

then we can combine them in many di↵erent ways, to obtain:

a+b = c + d,
a b = c d,
a⇥b = c ⇥ d.

Pause to think about this statement, and make sure it aligns with what you
know. Of course these are only several ways of combining these equations, and
every reader can think of several others. All of the above are “rules” of classical
arithmetic. What we would like to do now is consider whether similar rules
apply to modular arithmetic as well.
Suppose we have the following two congruence relations:

a ⌘ b (mod m)
c ⌘ d (mod m).

Are we able to combine these to obtain

a+b ⌘ c+d (mod m),

a b ⌘ c d (mod m),
a⇥b ⌘ c⇥d (mod m)?

That is, do the rules that govern how we can combine equations in classical
arithmetic also govern the ways in which we combine statements in modular
arithmetic? In what follows we prove that indeed many of the rules do carry
over – the rules of modular arithmetic will be familiar to us.

74
 last edited April 30, 2016

Addition
The first rule we consider is that associated with addition. Suppose we have
two congruence relations: a ⌘ b (mod m) and c ⌘ d (mod m). In other words,
a and b are congruent and c and d are congruent, both mod m. We can add the
left sides of these congruent relations, add the right sides, and the results will
again be congruent. In symbols,
Theorem 15.
If a ⌘ b (mod m) and
c ⌘ d (mod m), then
a+c ⌘ b+d (mod m).

Proving this result involves nothing more than applying the definition of
congruence and some basic algebraic manipulation.
Proof. By the definition of congruence (Definition 25) we know that a and b
di↵er by some multiple of m, i.e.,
b a = km (64)
for some k 2 Z. Likewise we know that c and d also di↵er by some multiple of
m, i.e.,
d c = jm (65)
for some j 2 Z. Note that we use j instead of k since the multiple of m by which
c and d di↵er might be di↵erent from the multiple by which a and b di↵er. Next
we add these two equations together:
(b a) + (d c) = km + jm. (66)
We can rewrite this equation as
(b + d) (a + c) = (j + k)m. (67)
By the definition of congruence modulo m, this is the same as saying that a + c
is congruent to b+d modulo m, since a+c and b+d di↵er by an integer multiple
(j + k) of m. In symbols, we have:
a+c⌘b+d (mod m), (68)
as desired.
A similar proof can be used to show that if a ⌘ b (mod m) and c ⌘ d
(mod m), then a c ⌘ b d (mod m).
These two results allow us to treat all numbers that are congruent modulo
m as identical when adding and subtracting numbers. If we know that a ⌘ 3
(mod 7) and b ⌘ 4 (mod 7), then we can know that a + b ⌘ 7 ⌘ 0 (mod 7).
This is true whether a is 10 or 703, and whether b is 7004, 10000, or 7,000,004.
What a and b actually are does not matter if we only want to determine whether
a + b is congruent to 0 or not.

75
 last edited April 30, 2016

Multiplication
After understanding how addition and subtraction work in modular arithmetic,
we turn our attention to understanding multiplication. In classical arithmetic,
if a = 2 and b = 5, then of course a ⇥ b = 2 ⇥ 5 = 10. Does a similar relationship
also hold in modular arithmetic? In particular, if we know that a ⌘ 2 (mod m)
and b ⌘ 5 (mod m), do we know that a ⇥ b ⌘ 2 ⇥ 5 (mod m)?
The following theorem answers this question affirmatively.
Theorem 16.

If a ⌘ b (mod m) and
c ⌘ d (mod m), then
a⇥c ⌘ b⇥d (mod m).

Proof. By the definition of congruence we know that a and b di↵er by a multiple

of m, as do c and d:

b a = jm
d c = km

for some j, k 2 Z. Note that we use distinct multiples j and k for the two
equations, since a and b might di↵er by one multiple of m, and c and d might
di↵er by another multiple of m.
To prove the desired result, we rearrange the equations:

b = jm + a
d = km + c

We multiply both sides by each other to obtain

bd = (jm + a)(km + c)
= jkm2 + jmc + kma + ac
= (jkm + jc + ka)m + ac.

We then subtract ac from both sides to obtain

bd ac = (jkm + jc + ka)m.

Since (jkm + jc + ka)m is an integer multiple of m, then ac and bd di↵er by an

integer multiple of m, and so by definition are congruent mod m.
Example 1. If we know that a ⌘ 3 (mod 7) and we know that b ⌘ 4 (mod 7),
then we can determine that ab ⌘ 12 ⌘ 5 (mod 7). This is true whether a is
10, 703, or 7,000,003 and whether b is 7004 or 10000. In any of these cases, the
product ab will be congruent to 5 modulo 7.

76
 last edited April 30, 2016

Example 2. How can we simplify 20 ⇥ 21 in arithmetic modulo 19? We first

note that 20 ⌘ 1 (mod 19) and also that 21 ⌘ 2 (mod 19). Theorem 16 tells us
that we can combine these equations to obtain 20 ⇥ 21 ⌘ 1 ⇥ 2 ⌘ 2 (mod 19).
Example 3. Can we simplify 17753 in arithmetic modulo 9? We first note that
17 ⌘ 1 (mod 9), because 17 and -1 di↵er by a multiple of 9. Theorem 16 allows
us to then combine this congruence relation as many times as we would like. In
particular, by combining 753 copies, we obtain 17753 ⌘ ( 1)753 (mod 9). Since
( 1)n = 1 for any odd integer n, we have 17753 ⌘ 1 (mod 9). Finally, if we
would like to have a simple, positive answer, then we can add 9 to obtain a final
answer of 8.

Theorems 15 and 16 show us that we can treat all numbers that are congruent
modulo m as the same, in addition and in multiplication operations. Division
is much more complicated, and will not be discussed.

Remainders
We take a moment to draw out a connection to division with remainders, an idea
we considered briefly in Section 4.1. In particular, back in elementary school we
learned about a way of dividing integers by other integers that entirely avoids
decimals and fractions. In particular, suppose we divide 7 by 4. In third, fourth,
or fifth grade, we learned that we can write this as 1, remainder 3. That is, 4 can
1 time “into” 7, leaving over 3. As we got older, we learned that we could also
write the answer as 1.75 or 13⁄4, but we still occasionally deal with situations
in which discussing fractions would be silly. If we have 52 playing cards and 5
players, a dealer could give each player 10 cards and then be left with 2 cards.
It makes little sense to say that the dealer should give each player 10.2, or 10
and a fifth, cards.
What is the connection of modular arithmetic to division with remainders?
Suppose that we divide some integer a by another integer m. Notice that the
“remainder” is always congruent to a modulo m. For example, suppose we
divide 1031 by 19. We obtain 54, remainder 5. This tells us that 5 is congruent
to 1031 modulo 19. Likewise, since the remainder of 7381/57 is 28, we know
that 28 ⌘ 7381 (mod 57).
Why is the remainder after division always congruent to the number we are
dividing? One way to think about this is by considering how we can find a
remainder without actually doing any division. Suppose we want to know the
remainder of 11 after dividing by 3. We can subtract 3 over and over until we
obtain a number that is smaller than 3: 11, 8, 5, and eventually 2. Each time
we subtract 3, we are realizing that 3 can “go into” 11 one more time; whatever
is left at the end is the remainder. At the same time, we got from the original
number to the remainder by jumps of 3, so of course the di↵erence between 11
and 2 is divisible by 3, making 11 and 2 congruent. The same idea works for
dividing any number a with any other number m.

77
 last edited April 30, 2016

Standard Representation
We have by now seen that in arithmetic modulo m, there is no di↵erence between
writing 1, 1 + m, 1 + 2m, and so forth, at least as far as addition, subtraction,
and multiplication are concerned. For this reason, writing 4+11 ⌘ 15 (mod 13)
is “just as correct” as writing 4 + 11 ⌘ 2 (mod 13), and “just as correct” as
writing 4 + 11 ⌘ 11 (mod 13). As far as arithmetic modulo 13 is concerned,
2, 15, and -11 are exactly the same number. However, in some applications it
is convenient to agree upon a standard way to represent numbers. What is a
good way to do this? Which of {. . . , a 2m, a m, a, a + m, a + 2m, . . .} should
we consider the standard representative?
You have likely encountered a similar problem back in your days learning
about trigonometric functions. A teacher may have asked you what is the inverse
sine of 1, i.e., sin 1 ( 1). You may have correctly answered 270 . Or you
may have correctly answered 90 . In fact, any number that can be written
270 + n360 , for any integer n 2 Z, would also be equally correct. But if
each student wrote a di↵erent number on an exam, it could take a long time to
determine whether or not every answer is correct. Is 1500 a correct solution?
Is 1530 ? For this reason, we might specify that we looking for a correct answer
between 0 and 360 , or else between 180 and 180 , since there is exactly one
correct answer in each of these ranges.
In the same way, when working in arithmetic modulo 41, the numbers
{. . . , 29, 12, 53, 94, 135, . . .} are all the same, yet we might hope to specify
one of them to be the standard representation of them. Indeed, in arithmetic
modulo m, we refer to the numbers {0, 1, 2, . . . , m 1} as the standard repre-
sentations of the integers. If numbers are always represented in this standard
form, determining whether or not two numbers are congruent is as easy as look-
ing at whether the numbers are equal. Notice also that this set of numbers is
also the set of possible remainders after dividing a number by m.
Example 1. Suppose we want to know the remainder of 17 ⇥ 18 when it is
divided by 19. We can do this in two di↵erent ways. First, we can multiply
the two numbers directly and obtain 306; some calculation will show that 306 is
congruent to 2 modulo 19. Alternatively, we know that 17 ⌘ 2 (mod 19) and
18 ⌘ 1 (mod 19). Multiplying both sides we see that 17⇥18 ⌘ ( 2)⇥( 1) ⌘ 2
(mod 19).
Example 2. Suppose we want to determine the standard form of 172 in mod
19 arithmetic. One way in which we can do this is by considering the square of
17, which is 289, divide that by 19 and then take the remainder. However, since
we know that 17 ⌘ 2 (mod 19), we can multiply this congruence equation
by itself to obtain 172 ⌘ 22 ⌘ 4 (mod 19). We can easily verify that the
remainder of 289, when divided by 17, is indeed 4.
Example 3. Suppose we want to determine the standard form of 18489391312
in mod 19 arithmetic. We should first notice that in mod 19 arithmetic, 18 is
congruent to 1, and so 18489391312 ⌘ ( 1)489391312 (mod 19). It is relatively

78
 last edited April 30, 2016

easy to see that if n is odd then ( 1)n = 1, and if n is even then ( 1)n = 1.
Since 489391312 is even, 18489391312 ⌘ 1 (mod 19).

Dividing by 9
We can use the rules of modular addition and multiplication to prove a theorem
you may have once seen. Suppose we have a number, for example 2,383,623,
and want to know whether it is divisible by 9. Is there an easy way to figure this
out without doing “long division”? You may have learned the following trick:
add up the digits of the number (e..g., 2 + 3 + 8 + 3 + 6 + 2 + 3 = 27). If this
sum is divisible by 9, then so is the original number; if the sum is not divisible
by 9, then neither is the original number. Is this just a miraculous trick, or is
it something that we can prove should work?
The rules of modular addition and multiplication (Theorems 15 and 16
above) can help us prove this beautiful result. Let’s begin by proving a sim-
pler result about the remainders we get when we divide powers of 10 by 9. In
particular, the remainder is always 1.
Lemma 17. For any natural number n, we have 10n ⌘ 1 (mod 9).
Proof. Recall that if we have two congruences: a ⌘ b and c ⌘ d (mod m), then
we can combine them to form a new congruence relation: ac ⌘ bd (mod m).
Since 10 ⌘ 1 (mod 9), then we can combine the equation with itself to obtain
100 = 10 ⇥ 10 ⌘ 1 ⇥ 1 ⌘ 1 (mod 9). We can indeed combine this equation with
itself as many times as we want (e.g., n times), and therefore have 10n ⌘ 1n ⌘ 1
(mod 9) for any natural number n.
Next, let’s consider what happens when we divide numbers such as 300,
5000, and 2,000,000 by 9. What are the remainders? Theorem 16 can help us
see that the remainders are 3, 5, and 2 in these examples. To see why this is so,
notice that each of these numbers can be written as the product of an integer
and a power of 10: 300 = 3 · 102 , 5000 = 5 · 103 , and 2,000,000= 2 · 106 . This
leads us to the following theorem.
Lemma 18. For any natural numbers c and n, we have c · 10n ⌘ c (mod 9).
Proof. Recall that if we have two congruences: a ⌘ b and c ⌘ d (mod m), then
we can combine them to form a new congruence relation: ac ⌘ bd (mod m).
Since c ⌘ c and 10n ⌘ 1 (mod 9) for any n, then we can combine the equations
to obtain c · 10n ⌘ c · 1 ⌘ c (mod 9).
This now leads us to our central theorem:
Theorem 19. A number is divisible by 9 if and only if the sum of its digits
(written in base 10) is divisible by 9.
Proof. In base 10, every number can be written as a sum of ones, tens, hundreds,
thousands, and so forth. For example, 5776 = 5000+700+70+6. More generally,
we can write this as n = c0 + c1 101 + c2 102 + c3 103 + . . ., where the ci variables

79
 last edited April 30, 2016

are the numbers of ones, tens, hundreds, thousands, and so forth. According to
Lemma 18, for each of the ci we have ci · 10n ⌘ ci (mod 9). Using Theorem 15,
we can combine the congruence relations

c0 ⌘ c0 (mod 9),
c1 ⌘ c1 101 (mod 9),
2
c2 ⌘ c2 10 (mod 9),
3
c3 ⌘ c2 10 (mod 9),
...
cn ⌘ c2 10n (mod 9),

to give us

c0 + c1 101 + c2 102 + . . . cn 10n ⌘ c0 + c1 + c2 + . . . cn (mod 9) (69)

In other words, a number n is congruent to the sum of its digits in mod 9. If a

number is divisible by 9, i.e., n ⌘ 0 (mod 9), then so is the sum of its digits.

80