Topic 2: Number Theory

Topic 2: Number Theory

Part 1 – Introduction
a. Some History
b. Divisibility Tricks
c. Coprimality
d. Breaking down divisibility problems
Part 2 – Factors and Divisibility
a. Using the prime factorisation
i. Nearest cube/square
ii. Number of zeros
iii. Number of factors
b. Factors in an equality
c. Consecutive integers
Topic 2: Number Theory
Part 3 – Diophantine Equations
a. Factors in an equality (revisited)
b. Dealing with divisions
c. Restricting integer solutions

Part 4 – Modular Arithmetic

a. Introduction
b. Using laws of modular arithmetic
c. Useful properties of square numbers
c. Multiples and residues
d. Playing with different moduli
Topic 2: Number Theory
Part 5 – Digit Problems

a. Reasoning about last digit

b. Representing algebraically

Part 6 – Rationality

Part 7 – ‘Epilogue’
ζ
Topic 2 – Number Theory

Part 1: Introduction
What is Number Theory?
Number Theory is a field concerned with integers (and fractions), such as
the properties of primes, integer solutions to equations, or proving the
irrationality of π/e/surds.

How many zeros does

How many factors does 50! have? What is its
10001000 have? last non-zero digit?

Are there any integer

solutions to a3 + b3 = c3?

Prove that the only non-trivial

integer solutions to ab = ba is {2,4}
Who are the big wigs?

Euclid (300BC)
Better known for his work in geometry, but proved there are infinitely
many primes. Euclid’s Algorithm is used to find the Greatest Common
Divisor of two numbers.

Fermat (1601-1665)
Most famous for posing “Fermat’s Last Theorem”, i.e. That
has no integer solutions for , and when .
Also famous for Fermat’s Little Theorem (which we’ll see), and had an
interest in ‘perfect numbers’ (numbers whose factors, excluding itself,
add up to itself).

Euler (1707-1783)
Considered the founder of ‘analytic number theory’. This included
various properties regarding the distribution of prime numbers. He
proved various statements by Fermat (including proving there are no
integer solutions to ). Most famous for ‘Euler’s Number’, or ‘e’ for short
and Euler’s identity, .
Who are the big wigs?

Lagrange (1736-1813)
Proved a number of Euler’s/Fermat’s theorems, including proving that
“every number is the sum of four squares” (the Four Square Theorem).

Dirichlet (1805-1859)
Substantial work on analytic number theory. e.g. Dirichlet’s Prime Number
Theorem: “All arithmetic sequences, where the initial term and the
common difference are coprime, contain an infinite number of prime
numbers.”

Riemann (1826-1866)
The “one hit wonder” of Number Theory. His only paper in the field
“On the number of primes less than a given magnitude” looked at the
density of primes (i.e. how common) amongst integers. Led to the yet
unsolved “Riemann Hypothesis”, which attracts a $1m prize.
Who are the big wigs?

Andrew Wiles (1953-)

He broke international headlines when he proved Fermat’s Last
Theorem in 1995. Nuf’ said.
Is 1 a prime number?

Vote
 No Vote
Yes

Euclid’s Fundamental Theorem of Arithmetic, also known as the Unique Factorisation

Theorem, states that all positive integers are uniquely expressed as the product of
primes.

Assume that 1 is a prime.

Then all other numbers can be expressed as a product of primes in multiple ways: e.g. ,
but also , and , and so on.
Thus the Fundamental Theorem of Arithmetic would be violated were 1 a prime.

[Link] provides some other reasons.

(Note also that 0 is neither considered to be ‘positive’ nor ‘negative’. Thus the
‘positive integers’ start from 1)
Divisibility Tricks
How can we tell if a number is divisible by...

2 Last number is even. ?

3 Digits add up to multiple of 3. e.g: ?1692: 1+6+9+2 = 18 
4 ? 143328
Last two digits are divisible by 4. e.g.
5 Last digit is 0 or 5. ?
6 ? tests for 2 and 3).
Number is divisible by 2 and 3 (so use
7 There isn’t really any trick that would save time. You could double the last digit
and subtract it from the remaining digits, and see if the result is divisible by 7.
?
e.g: 2464 -> 246 – 8 = 238 -> 23 – 16 = 7. But you’re only removing a digit each time,
so you might as well long divide!

8 Last three digits divisible by 8.

?
9 Digits add up to multiple of 9.
?
10 Last digit 0.
?
11 When you sum odd-positioned digits and subtract even-positioned
digits, the result is divisible by 11.
?
e.g. 47949: (4 + 9 + 9) – (7 + 4) = 22 – 11 = 11, which is divisible by 11.

12 Number divisible by 3 and by 4.

?
Notation

This means “2 divides 8”.

2∨8
This means “2 does not
divide 9”.
2 ∤9
is divisible by 5.

5∨𝑛
is a factor of 15.

𝑛∨15
Notation
This is NOT a coordinate! It
means “the Greatest
Common Divisor of 15 and
10”.

( 15 , 10 ) =5
The Lowest Common
Multiple of 6 and 8.

𝐿𝐶𝑀 ( 6 , 8 )=24
True or false

If and , then is it the case that ? False True 

If and , then is it the case that ? False  True 

Take 12 for example. It’s divisible by 4 and 6, but

not by 24.

In general, if a number is divisible by a and b, then

the largest number it’s guaranteed to be divisible
by the Lowest Common Multiple of a and b.
LCM(4,6) = 12.
Coprime

If two numbers a and b share no common factors, then the

numbers are said to be coprime or relatively prime. The following
then follows:

Coprime?

2 and 3? No 
True

5 and 6? No 
True

10 and 15? No  
True
Breaking down divisibility problems

We can also say that opposite:

If we want to show a number is divisible by 15:

?
...we can show it’s divisible by 3 and 5.

But be careful. This only works if the two numbers are coprime:

If we want to show a number is divisible by 8:

...we can just show it’s divisible by 4 and 2?

No: LCM(2,4) = 4, so a number divisible by 2 and 4 is definitely

divisible by 4, but not necessarily?divisible by 8.
Breaking down divisibility problems

Key point: If we’re trying to show a number is divisible by some large number, we
can break down the problem – if the number we’re dividing by, , has factors ,
such that and and are coprime, then we show that is divisible by and divisible
by . Similarly, if and , , and are all coprime, we show it’s divisible by , and .

If we want to show a number is divisible by 24:

? 8
We can show it’s divisible by 3 and
(Note, 2 and 12 wouldn’t be allowed because they’re not coprime. That same applies for 4 and 6)

Which means we’d have to show the number has

the following properties:
1. Its last 3 digits are divisible by 8.
2. Its digits add up to? a multiple of 3.
Breaking down divisibility problems

[Hamilton 2006 Q1] Find the smallest positive integer which

consists only of 0s and 1s, and which is divisible by 12.

Answer: 11100 ?

A number divisible by 12 must be divisible by 3 and

4. If divisible by 4, the last two digits are divisible
by 4, so most digits must be 0.
If divisible be three, the number of 1s must be a
multiple of 3. For the smallest number, we have
exactly 3 ones.
Breaking down divisibility problems

Explain why and are coprime for any positive integer .

Answer:
Suppose had some factor . Then must have a remainder
?
of 1 when divided by , so is not divisible by .

The same reasoning underpins Euclid’s proof that there are infinitely many
primes. Suppose we have a list of all known primes: . Then consider one more
than their product, . This new value will always give a remainder of 1 when we
divide by any of the primes to . If it’s not divisible by any of them, either the new
number is prime, or it is a composite number whose prime factors are new
primes. Either way, we can indefinitely generate new prime numbers.
Coprime

If is odd, will and be coprime?

Answer:
? a factor of 2.
No. Because k-1 and k+1 will be contain

If is even, will and be coprime?

Answer:
Yes. If a number d divides k-1, then the remainder will be 2 when k+1
is divided by d. Thus the divisor could?only be 2, but k-1 is odd.
Therefore there can be no common factor.

These are two very useful facts that I’ve seen come up in a lot of problems.
We’ll appreciate their use more later:
1. and are coprime for any positive integer .
2. and are coprime if is even.
ζ
Topic 2 – Number Theory

Part 2: Factors and Divisibility

Using the prime factorisation
Finding the prime factorisation of a number has a number of useful
consequences.

3 2
360=2 × 3 × 5 ?

We’ll explore a number of these uses...

Using the prime factorisation
Handy Use 1: Smallest multiple that’s a square or cube number?

3 2
360=2 × 3 × 5
?
Smallest multiple of 360 that’s a perfect square = 3600

If the powers of each prime factor are even, then the number
is a square number (known also as a “perfect square”).
For example . So the smallest number we need to multiply by to
get a square is , as we’ll then have even powers.
Using the prime factorisation
Handy Use 1: Smallest multiple that’s a square or cube number?

3 2
360=2 × 3 × 5
?
Smallest multiple of 360 that’s a cube = 27000

If the powers of each prime factor are multiples of three, then

the number is a cube number.
For example . So the smallest number we need to multiply by to
get a square is .
Using the prime factorisation
Handy Use 2: Number of zeros on the end?

7 2 4
2 ×3 × 5
Q1) How many zeros does this number have on the end?
Answer: 4. 27 x 32 x 54 = 23 x 32 x (2 x 5)4
= 23 x 32 x 10?4

Q2) What’s the last non-zero digit?

Answer: Using the factors we didn’t combine to make
2-5 pairs (i.e. factors of 10),?we have 23 x 32 left. This
is 72, so the last non-zero digit is 2.
 Using the prime factorisation
Handy Use 2: Number of zeros on the end?

What is the highest power of 10 that’s a factor of:

Answer: 12
50! 50! = 50 x 49 x 48 x ... We know each prime factor of 2 and 5 gives us
a power of 10. They’ll be plenty of factors of 2 floating around, and
less 5s, so the number of 5s give us? the number of pairs. In 50 x 49 x
48 x ..., we get fives from 5, 10, 15, etc. (of which there’s 10). But we
get an additional five from multiples of 25 (of which there’s 2). So
that’s 12 factors of 10 in total.

1000! Answer: 249

Within 1000 x 999 x ... , we get prime factors of 5 from each multiple
of 5 (of which there’s 200), an additional 5 from each multiple of 25
(of which there’s 40), an additional ? 5 from each multiple of 125 (of
which there’s 8) and a final five from each multiple of 625 (of which
there’s just 1, i.e. 625 itself). That’s 249 in total.
Using the prime factorisation
Handy Use 2: Number of zeros on the end?

What is the highest power of 10 that’s a factor of:

In general,
⌊log 5 𝑛 ⌋ 𝑛
Σ 𝑘=1 ⌊ 𝑘⌋
5
gives us power of 5 that results in . So rounding this down, we get
the largest power of 5 that results?in a number less than . is
known as the ‘floor function’ and rounds anything inside it down.
So . Then if is the power of 5 we’re finding multiples of, there’s
of these multiples (after we round down).
Using the prime factorisation
[SMC 2005 Q24] For how many positive integer values of less than 50 is it
impossible to find a value of such that ends in exactly zeros?

A: 0 B: 5 C: 8

D: 9  E: 10

When is written in full, the number of zeros at the end of the number is equal to the power of 5 when
is written as the product of prime factors. We see that 24! ends in 4 zeros as 5, 10, 15 and 20 all
contribute one 5 when 24! is written as the product of prime factors, but 25! ends in 6 zeros because
25 = 5 × 5 and hence contributes two 5s. So there is no value of for which ends in 5 zeros. Similarly,
there is no value of for which ends in 11 zeros since 49! ends in 10 zeros and 50! ends in 12 zeros. The
full set of values of less than 50 for which it is impossible to find a value of such that ends in zeros is
5, 11, 17, 23, 29, 30 (since 124! ends in 28 zeros and 125! ends in 31 zeros), 36, 42, 48.
Using the prime factorisation
Handy Use 3: Number of factors?

7 4
72576=2 ×3 ×7
A factor can combine any
number of these prime factors
together. e.g. , or none of
them (giving a factor of 1).
And we can either have the 7
We can use between 0 and 7 or not in our factor. That’s 2
of the 2s to make a factor. possibilities.
That’s 8 possibilities.

Similarly, we can have So there’s factors

between 0 and 4 threes.
That’s 5 possibilities.
Using the prime factorisation
Handy Use 3: Number of factors?

𝑞 𝑟 𝑠
𝑎 ×𝑏 ×𝑐
In general, we can add 1 to each of the indices, and
multiply these together to get the number of factors.
So above, there would be factors.
Using the prime factorisation
Handy Use 3: Number of factors?

How many factors do the following have?

50? 10100?
?
so factors.
?
So 1012 factors
= 10201 factors.

200? 20032003? (Note: 2003 is prime)

?
so factors. This is already prime-
factorised,?so there’s
2004 factors.
Using the prime factorisation

Question: How many multiples of 2013 have 2013 factors?

A: 0 B: 1 C: 3

D: 6  E: Infinitely
many

Hint:
Use the ‘number of factors’ theorem backwards: If there are 2013 factors, what
could the powers be in the prime factorisation?
Solution: Firstly note that any multiple of 2013 must have at least powers of 3, 11
and 61 in its prime factorisation (with powers at least 1). If there are 2013 factors,
then the product of one more than each of the powers in the prime factorisation
is 2013. e.g. We could have , since . There’s ways we could arrange these three Int Kangaroo
powers, which all give multiples of 2013. Our multiple of 2013 can’t introduce any
new factors in its prime factorisation, because the number of factors 2013 only Pink
has three prime factors, and thus can’t be split into more than three indices.
Grey
Factors in an equality
We can reason about factors on each side of an equality.

What do we know about and ?

3 𝑛=8 𝑘
Answer:
If the LHS is divisible by 3, then so must the RHS.
And since 8 is not divisible?by 3, then k must be. By
a similar argument, n must be divisible by 8.
Factors in an equality
In general, if we know some property of a number, it
can sometimes help to replace that number with an
expression that represents that property.

This skill becomes hugely important when

considering integer solutions for equations.

is even: Let for some integer

is odd: Let ?
is a multiple of 9: Let ?
only has prime factors of 3: Let ?
is an odd square number: If and is odd, must
also be odd. So?
Factors in an equality

Question: Show that has no integer solution

for .

Answer:
Since the LHS only has prime factors of 2, then so must
the RHS. Therefore let for some integer .
Then and equating indices,?
. But the RHS is divisible by 3 while the LHS is not,
leading to a contradiction.
Factors in an equality

Question: If , then what can we say about and ?

(Recall: and are coprime)

Answer:
If k and k+1 are coprime, they share no factors, so the prime factors
on the LHS must be partitioned into two, depending whether they
belong to k or k+1. In n2, each prime factor appears twice, so they
must both belong to either k or k+1 (but can’t be in both). So far,
both k and k+1 will both be square,? because each prime factor comes
in twos. This just leaves the 3, which is either a factor of k or k+1.
Therefore, one of k and k+1 is three times a square, and the other a
square.
(An interesting side point: Finding possible n is quite difficult. Using a
spreadsheet, the only valid n I found up to 10,000 were 2, 28, 390 and 5432.)
Divisibility with consecutive integers

Every other integer is divisible by 2.

1 2 3 4 5 6 7 8
Every third integer is divisible by 3. 1 2 3 4 5 6 7 8
Every fourth integer is divisible by 4. 1 2 3 4 5 6 7 8

An ‘obvious’ fact that can aid us in solving less

than obvious problems!
Divisibility with consecutive integers

[SMC 2003 Q13] Which of the following is divisible by 3

for every whole number x?

A:  B:  C: 

D:  E: 

Since , is always the product of three consecutive whole numbers

when is a whole number. As one of these must be a multiple of 3,
will be divisible by 3.
Alternatively, substituting 2 for in the expressions in B,C and E and
substituting 3 for in the expression in results in D numbers which
are not divisible by 3.
(Note: We’ll revisit this problem later when we cover modulo
arithmetic!)
 Divisibility with consecutive integers
[BMO 2005/06 Q1] Let n be an integer greater Use what you know!
than 6. Prove that if and are both prime, then If and are both

is divisible by 720. prime, I can establish
properties about ’s
divisibility.

 720 has a factor of 5.

Solution: As and are prime, must be divisible by 2 (since ). What expression can
I form that we know
Thus is divisible by , as and both are. will be divisible by 5?
One of , and must be divisible by 3, but since and are
prime, must be divisible by 3. Therefore must be divisible by
9, as is.
One of , , , and are divisible by 5. and can’t be as they’re
prime. Therefore is a multiple of 5. We now need to
?
somehow relate this to . If is divisible by 5, then is, and is
because is clearly divisible by 5. Therefore is divisible by 5.
Thus, is divisible by .
ζ
Topic 2 – Number Theory

Part 3: Diophantine Equations

What is a Diophantine Equation?
An equation for which we’re looking for integer solutions.
Some well-known examples:

When n=2, solutions known as

𝑥 𝑛+ 𝑦 𝑛=𝑧 𝑛 Pythagorean triples. No solutions
when n>2 (by Fermat’s Last Theorem).

3 𝑥+ 4 𝑦 =24 Linear Diophantine Equation.

Erdos-Staus Conjecture states that 4/n

can be expressed as the sum of three
unit fractions (unproven).

Pell’s Equation. Historical interest because it

2 2
𝑥 −𝑛 𝑦 =1 could be used to find approximations to square
roots. e.g. If solutions found for x2 – 2y2 = 1, x/y
gives an approximation for √2
Factors in an equality
To reason about factors in an equality, it often helps to get it into a form where
each side is a product of expressions/values.

Example: How many positive integer solutions for the

following?

( 𝑥 − 6)( 𝑦 −10)=15
Answer: 6. Possible (x,y) pairs are (7, 25), (9, 15),
?
(11, 13), (21, 11), (3, 5), (1, 7)

The RHS is 15, so the multiplication on the LHS must be

, , , , , , etc. So for the first of these for example, and , so and .
Make sure you don’t forget negative factors.
Forming a Diophantine Equation
You should try to form an equation where you can reason about factors in this
way!

[Hamilton 2011 Q3] A particular four-digit number is such that:

(a) The sum of and 74 is a square; and
(b) The difference between and 15 is also a square.
What is the number ?

Step 1: Represent algebraically: Step 3: Reason about factors

2 Conveniently 89 is prime, and since is

𝑁 =74=𝑞 ? greater than , then and .
Solving these simultaneous equations
Step 2: Combine equations in some gives us and . ?
useful way. Using one of the original equations:
“Perhaps if I subtract the second from
the first, then I’ll get rid of , and have
the difference of two ? squares on the
RHS!” 𝟖𝟗=( 𝒒 +𝒓 ) ( 𝒒 − 𝒓 )
Forming a Diophantine Equation
Aim to factorise your equation.

[BMO 2011/12 Q1] Find all positive values of for which is

a (perfect) square.
Hint: Perhaps complete the square?

Solution: .
for some integer .

89 is prime. And since ,

and . ?
Using the latter,
So substituting into the first, .
, so .
For problems involving a square number, the ‘difference of two
squares’ is a handy factorisation tool!
Forming a Diophantine Equation
There’s a variety of different strategies to factorise a Diophantine Equation.

[BMO 2012/13 Q4] Find all positive integers such that and are
both perfect squares.
Based on the strategy on the previous question, we might have tried one
equation subtracting the other to get the difference of two squares:

But this is a bad strategy, because unlike before, we haven’t eliminated the
variable on the LHS, and thus the above equation isn’t particularly useful.
How could we deal with just 2 variables?

Make the subject of each to get:

? ( 5𝑘+2𝑞)( 5𝑘−2𝑞 )
Manipulating a Diophantine Equation
Aim to factorise your equation.

[Maclaurin 2008 Q3] Show that the following equation

has no integer solutions:

Questions of this form are quite common, particularly in the Senior Maths
Challenge/Olympiad. And the approach is always quite similar...

Step 1: It’s usually a good strategy in algebra to get rid of fractions: so multiply
through by the dominators.

11 𝑥+11 𝑦=5
? 𝑥𝑦
Manipulating a Diophantine Equation
Aim to factorise your equation.

11 𝑥+11 𝑦=5 𝑥𝑦
Step 2: Try to get the equation in the form

This is a bit on the fiddly side but becomes easier with practice.
Note that
Similarly

So initially put the equation in the form

Looking at the form above, it would seem to help to multiply by the
coefficient of (i.e. 5), giving
This allows us to factorise as (5x – 11)(5y – 11) – 121 = 0.
The “-121” is because we want to ‘cancel out’ the +121 the results
from the expansion of (5x – 11)(5y – 11).

So
Manipulating a Diophantine Equation
Aim to factorise your equation.

( 𝟓 𝒙 −𝟏𝟏 ) (𝟓 𝒚 −𝟏𝟏 )=𝟏𝟐𝟏

Step 3: Now consider possible factor pairs of the RHS as before.

Since the RHS is , then the left hand brackets must be or or or , etc. (don’t
forget the negative values!)

If , then is not an integer.

If , then is not an integer.
If , then , but , where is not an integer.
(And for the remaining three cases, there is no pair of positive integer
solutions for and )
 Manipulating a Diophantine Equation
Let’s practice! Put in the form
Use the 4 -5 and -7 (-5) x (-7)
swap
7 5 from
positions.
+ =4 𝟒 𝒙𝒚 − 𝟓 𝒙 −𝟕 𝒚 =𝟎
𝑥 𝑦 (𝟒 𝒙 – 𝟕)(𝟒 𝒚 – 𝟓)=𝟑𝟓
1 1 𝒙𝒚 − 𝒙?− 𝒚 =𝟎 ( 𝒙 − 𝟏 ) ( 𝒚?−𝟏 )=𝟏
+ =1
𝑥 𝑦
3 3
+ =2 𝟐 𝒙𝒚 −𝟑 ?𝒙 − 𝟑 𝒚=𝟎 ( 𝟐 𝒙 −𝟑 )( 𝟐?𝒚 −𝟑 ) =𝟗
𝑥 𝑦
(Source: SMC)

1 2 3 𝟑 𝒙𝒚 −𝟑𝟖 𝒙?−𝟏𝟗 𝒚 =𝟎 ( 𝟑 𝒙 −𝟏𝟗 ) (𝟑 𝒚?−𝟑𝟖 )=𝟕𝟐𝟐

+ =
𝑥 𝑦 19
In general, this technique is helpful whenever we have a mixture Now for each of these, try
of variables both individually and as their product, e.g. , and , to find integer solutions
and we wish to factorise to aid us in some way.. for and ! (if any)
Dealing with divisions

Suppose you are determining possible values of a variable in

a division, aim to get the variable in the denominator only.

Example: How many positive integer solutions for given

that the following is also an integer:
𝑛
100 − 𝑛
We can rewrite this as: (Alternatively, you could
100 − ( 100 − 𝑛 ) 100 have used algebraic long
= −1 division, or made the
substitution )
100 − 𝑛 100
?
− 𝑛
Now is just in the denominator. We can see that whenever divides 100, the fraction
yields an integer. This gives 99, 98, 96, 95, 89, 79, 75, 50
Dealing with divisions
In a division, sometimes we can analyse how we can modify the dividend so that
it becomes divisible by the divisor.

[SMC 2005 Q21] What is the sum of the values of for which both and are
integers?

A: -8 B: -4 C: 0

D: 4 E: 8 

Note that is divisible by . Thus:

. So must divide 8.

The possible values of are −8, −4, −2, −1, 1, 2, 4, 8, so is −7, −3, −1, 0, 2, 3,
5, 9. The sum of these values is 8.
(Note that the sum of the 8 values of is clearly 0, so the sum of the 8
values of is 8.)
Dealing with surd expressions
For Diophantine Equations involving surds, remember that the contents of the
surd must be a square number. Square each side of the equation.

[SMC 2000 Q24] How many pairs of positive integers

satisfy the equation: .

A: 0 B: 1 C: 2

D: 17
 E: 

Squaring both sides:

If is an integer, then must be an integer. This will be the case when for any
positive integer (except , because then would be 0). But there’s infinitely
many choices for , thus there’s infinitely many solutions for .
Restricting integer solutions
When you have to find all integer solutions to some equation, there’s usually
some way to round down your search.

[Cayley 2011 Q5] Solve the equation , where and are positive integers.

Answer: , and . ?

Hint: What do we know about the RHS of the equation?

What do this then tell us about and ?

, therefore . And since is positive, then dividing both sides by gives us . This means we only need to
try 1, 2, 3, 4 and 5!

If we sub in b = 1, we get 4a = 9, for which there’s no integer solution.

Continuing with possible b, we eventually find all our solutions.

In general, look out for things that are squared, as we know their value must be at
least 0 (nonnegative).
Diophantine Equation Summary

Try to get whatever equation you have as a product on each side, so

1 that you can reason about the factors. e.g.

You can occasionally use the difference of two squares to factorise. e.g:
2

To factorise, you might need to think backwards to determine what could

3 expand to get the terms you have. e.g. If you have , and in your
expression, then would expand to give all 3 of these.
In some contexts you can complete the square.

Once factorised, you need to consider possibilities for the factors on each
4 side. Don’t forget negative factors.
You can use number theory knowledge to round down what factors could be.
5 e.g. If you have , then prime factors in come in pairs.
e.g. If you have two factors that are consecutive, they are coprime and thus
share no factors.
ζ
Topic 2 – Number Theory

Part 4: Modular Arithmetic

What the devil is it?

On a digital clock, were we to

specify the hour as “27”, what we’d
actually mean is 3 in the morning.

These hours are the same in

“modulo 24 arithmetic”, i.e. our
numbers are limited to 0 to 23,
after which they loop back round.

27  3 (mod 24)

We’d say “27 is congruent to 3

modulo/mod 24”
What the devil is it?
Numbers in modulo arithmetic are all equivalent to numbers
in the range 0 to , where they then repeat.

0, 1, 2, 3, 4, 5, 6, 7, ...  0, 1, 2, 0, 1, 2, 0, 1, ... (mod 3)

This operator usually means ‘equivalent’, and in

this context more specifically means ‘congruent’.

We can use modulo arithmetic to represent the remainder

(also known as the residue) when we divide by some number.
What the devil is it?

How would we represent ?

𝑥 ≡ 0 ( 𝑚𝑜𝑑
? 3)

How would we represent “ is one less than a

multiple of 5”.

𝑥 ≡ 4 ( 𝑚𝑜𝑑 5 )
?
Or we could even use . We’ll see why that
might be useful later.
Properties of Modular Arithmetic

Addition works just as if it was a normal equality.

If 4  1(mod 3) then 4 + 5  1 + 5 (mod 3)

Multiplication also works.

If 4  1(mod 3) then 8  2(mod 3)

Exponentiation also works (this one we’ll use a lot!).

If 5  2(mod 3) then 5k  2k (mod 3) for any k

Quickfire Examples

Given that
Then ?

Given that
Then ?

Given that
Then ?

Given that
Then ?
Properties of Modular Arithmetic
Multiplication also works.
If , then
e.g. If , then

But is the converse always true?

i.e. If , then is ?
If not, can you think of a counterexample?
No. For example, note that , but when we divide each number by 4.
However, it IS true when the number we’re dividing by is coprime to the
modulo, i.e. .
?
e.g. . i.e. .
But 5 and 7 are coprime, so
Properties of Modular Arithmetic
Another common misconception (according to a BMO veteran) is that if:

) and
then:

This is not in general true!

I’ll leave it as an exercise to find a counterexample…
Using Laws of Modular Arithmetic
Often, it helps to consider all the possible residues.

Question: Show that the arithmetic sequence 2, 5, 8, 11, ...

does not contain a square number.

Let’s use modulo-3 arithmetic:

The given sequence:

2, 5, 8, 11, ...  2, 2, 2, 2, ... (mod 3)

The natural numbers:

0, 1, 2, 3, 4, 5, ...  0, 1, 2, 0, 1, 2, ... (mod 3)
Then by the laws of modulo arithmetic: ?
02, 12, 22, 32, 42, 52, ...  02, 12, 22, 02, 12, 22, ... (mod 3)
 0, 1, 1, 0, 1, 1, ... (mod 3)

We can see therefore that the square numbers only give a remainder of 0 or 1
when divided by 3, so we never see any of the numbers on the sequence.
Using laws of Modular Arithmetic
[SMC 2005 Q14] A square number is divided by 6.
Which of the following could not be the remainder?

A: 0 B: 1 C: 2 

D: 3 E: 4 

When divided by 6, a whole number leaves remainder 0, 1, 2, 3, 4 or 5. So the

possible remainders when a square number is divided by 6 are the
remainders when 0, 1, 4, 9, 16 and 25 are divided by 6. These are 0, 1, 4, 3, 4
and 1 respectively, so a square number cannot leave remainder 2 (or
remainder 5) when divided by 6..
Using laws of Modular Arithmetic

Problem Revisited!
Which of the following is divisible by 3 for every whole
number ? (Now answer using modular arithmetic)

A:  B:  C: 

D:  E: 

If for the natural numbers. then:

Then
i.e. For all numbers of , gives us a remainder of 0 when
dividing by 3.
Using laws of Modular Arithmetic

Source: Frosty Special

22
A square chessboard of
sides (for any ) is tiled
with L-shapes, each of 3
squares, such that tiles
22 don’t overlap.
Show that you will always
have 1 square on the
chessboard left untiled.

Solution: We’re finding the remainder when we

divide by 3. ?
. So .
Using laws of Modular Arithmetic
[BMO 1999/2000 Q2] Show that, for every positive integer ,
is divisible by 2000.
Hint: 2000 = 24 x 53, thus the only two coprime factors are 16 and 125.

Solution: If 121  9 (mod 16), then 121n  9n (mod 16). Similarly 25  9 (mod 16)
means that 25n  9n (mod 16). Conveniently, since the second is subtracted,
we’re left with 0 (mod 16) so far. 1900n  12n (mod 16) and (-4)n  12n (mod 16),
where with the latter we’ve just added 16 to make the remainder positive. These
? meaning that the expression is
again cancel, so overall we have 0 (mod 16),
divisible by 16.

Can use the same principle to show it’s divisible by 125.

Useful properties of square numbers
We’ve so far seen that it can sometimes be useful to consider the possible residues
of a square number to eliminate possibilities (as we’ll see for an upcoming
example).

There’s other handy properties to add to our ‘toolkit’:

Prove that if that if a square number is Prove that if a square number is odd, then
even, then it’s divisible by 4. it’s one more than a multiple of 8.
Answer: Answer:
(Method 1) All powers in the prime Note first that if a square is odd, then is odd since
factorisation of a square number are even, so odd × odd = odd.
if a factor of 2 appears (which it does because
the square is even), it must appear at least (Method 1) We need to show one less than a square
twice, so the square is divisible by 4. is divisible by 8.
. Both and are even. But one must be divisible by 4.
?
(Method 2) If is even, then must be even ?
So we get a factor of 4 from one and 2 from the
since even × even = even. other, thus it is divisible by 8.
Let . Then , which is clearly divisible by 4.
(Method 2) If is odd then let .
. One of and is even, so is divisible by 8.
Another problem revisited…

Question: If , then what can we say about and ?

(Recall: and are coprime)

We previously established that either is a square and is three

times a square, or vice versa. We can eliminate one of these cases
using modular arithmetic.

Case 1: and If is a multiple of 3, then has a

residue of 2 modulo 3. But, we
earlier saw square numbers can only
?
have residues of 0 or 1 modulo 3.
This contradicts that is a square.
We’ve eliminated this as a case.

Key Point: Modular Arithmetic can be useful to reason about what numbers can and can’t be.
Multiples and Residues

Suppose we’re working in modulo 7 arithmetic, and that we

start with a number 3, and find successive multiples:

3, 6, 9, 12, 15, 18, 21

≡ 3, 6, 2, 5, 1, 4, 0 (mod 7)
Notice that we get all possible remainders/residues. Under what
conditions do you think this happens?

When the modulo (in this case 7) and the difference

(in this case 3) are coprime. ?
We say we have a complete residue system if we have all residues.

We can see that because the last residue is 0, this number will be divisible by 7.
i.e. Every 7th number will be divisible by 7 under the above conditions.
Multiples and Residues
[BMO 2004/05 Q4] Determine the least possible value of the largest
term in an arithmetic progression of seven distinct primes..
Hint: If a is the first value and d is the difference, what properties must d have to
avoid being divisible by something?

Solution: 907
If our first number is prime, it’s clear that if the difference WASN’T a multiple of 2, then
every other number would be even. In terms of the theory on the last slide, we know we will
see all possible residues (i.e. 0 and 1) in modulo-2 arithmetic if the number we’re finding
multiples of, i.e. d, is not divisible by 2. Those with residues 0 will be divisible by 2 (unless it
is 2 itself) and thus not prime.

?
The same applies with 3 and 5 (the next two primes) so d must be divisible by these to avoid
residues of 0 every 3 and 5 numbers respectively.
7 however is more interesting. In modulo-7 arithmetic, the first number a could be 7 – while
divisible by 7, it’s clearly prime. This means that a needn’t be a multiple of 7 since it’s
possible we won’t see a residue of 0 again until 7 numbers later in
the list (i.e. beyond the end of our list!). BMO
So let’s make a = 7, and d a multiple of 2 x 3 x 5 = 30. After trying a Round 2
few multiples of 30, we’ll find that d = 150. So the last number is
a + (n-1)d = 7 + (6x150) = 907 Round 1
Multiples and Residues

A bit of extra context for this problem:

In the introduction, we saw Dirichlet’s Prime Number Theorem: “All arithmetic

sequences, where the initial term and the common difference are coprime,
contain an infinite number of prime numbers.”
3 and 4 are coprime, so sequence
3, 7, 11, 15, 19, ... will contain infinitely many prime
14, 16, 18, 20, 22, ... numbers.

But 14 and 2 are not coprime.

As recently as 2004, it was proven that the sequence of prime numbers contains an
arbitrarily long arithmetic progression. i.e. We can find an arithmetic sequence of any
length. (This is now known as the Green-Tao Theorem)
e.g. 3, 5, 7 and 47, 53, 59 are prime arithmetic sequences of length 3.

The theorem however only proves their existence; it doesn’t provide a method to find
a sequence of a given length. The longest sequence found so far is of length 26.
Dealing with remainders
If divided by gives a remainder of , then is divisible by .

For example, consider that 53 divided by 10 gives a remainder of 3.

Then obviously 53 – 3 = 50 is divisible by 10.

[Kangaroo Pink 2012 Q7] When 144 is divided by the positive

integer , the remainder is 11. When 220 is divided by the positive
integer , the remainder is also 11. What is the value of ?

A: 11
 B: 15
 C: 17


D: 19 E: 38

By our above rule, divides 144 – 11 = 133 and 220 – 11 = 209.

133 = 19 x 7 and 209 = 19 x 11
So both are divisible by 19.
Negative remainders

Sometimes it can be more convenient to put our remainder as a

negative number for purposes of manipulation.
For example, if the remainder when we divide a number by 3 is 2, then we
could also say this remainder is -1 because they are congruent.
By laws of modular arithmetic, 2n ≡ (-1)n (mod 3). We can more easily see the
remainder oscillates between -1 (i.e. 2) and 1 as n increases.

2n + 3n ≡ (-1)
? n (mod 3)

3 ≡ -2? (mod 5)
7 ≡ -3? (mod 10)
Playing with different moduli

An extremely useful method is to consider your equation in

different moduli to see if we can discover anything about the
variables.

Question: Is ever a perfect square? [Source OEIS]

Hint: See what you find modulo 3 and modulo 5.

Properties of discovered in modulo 3:

. But all squares are 0 or 1 modulo 3, so must be even or the remainder will be 2. So
let . ?
Using this information, we now we have .

Properties of n discovered in modulo 5:

Since our number has to be square, consider possible residues modulo-5: these are 0, 1
and 4. This doesn’t include 2 or 3 (i.e. -2). ?
We have therefore shown can never be a perfect square.
Playing with different moduli

Show that has no integer solutions.

(Hint: try using mod 4)

Since , we have .
Similarly .

So then considering all possibilities,

However, . ?

Thus there can be no integer solutions.

Putting everything together
The following was a particularly badly answered BMO problem. But we can
systematically reason through each step using the tips we’ve seen – no magic
required!

Question: Let be an integer. Show that, if is an integer, then it is a perfect

square.

First note that the question says IF [..] is an integer, THEN it is a

1 square. We need to start with the assumption, and reason
towards the conclusion – don’t be tempted to prove the opposite.

If is an integer, what can we assert about ?

2
a) It is a perfect square, since the square root has to be an integer.
b) It is odd. ?

3 What equation could we therefore write that would model this?

?
Putting everything together
The following was a particularly badly answered BMO problem. But we can
systematically reason through each step using the tips we’ve seen – no magic
required!

Question: Let be an integer. Show that, if is an integer, then it is a perfect

square.

To reason about factors, we know it’s generally a good idea to put an equation
4 in the form where we have the product of expressions on each side.
So rearrange

?
5 Use this to reason about the factors (Hint: We’ve seen this example before!)

• and are coprime.

• We earlier determined that one of and is the square, and the other 3
times a square. ?
• We also earlier determined that if was a multiple of 3, then by
modular arithmetic, couldn’t be a square. Therefore is a square, and is
three times a square.
Putting everything together
The following was a particularly badly answered BMO problem. But we can
systematically reason through each step using the tips we’ve seen – no magic
required!

Question: Let be an integer. Show that, if is an integer, then it is a perfect

square.

When we’ve used an expression to represent a restriction on a number, we

6 ought to substitute it into the original expression. Use

7 We earlier found that is a perfect square so what can we conclude?

A square times a square is a square, since , so is a perfect square.

?
Fermat’s Little Theorem
Not to be confused with Fermat’s Last Theorem!

If is prime, and is any integer (such that is not a multiple of ), then:

𝑝− 1
𝑎 ≡1 ( 𝑚𝑜𝑑 𝑝 )

EXAMPLES:

Fermat’s Little Theorem is a special case of Euler’s Theorem, which makes use of
something called Euler’s Totient Function. It’s not difficult, and worth looking up.
Fermat’s Little Theorem

Show that .
We’re trying to show that:

i.e.

By Fermat’s Little Theorem:

?

So:
Modular Arithmetic Summary
When working in modulo-k arithmetic, all integers that give the same
1 remainder when divided by k are equivalent/‘congruent’.

In many problems, it’s useful to consider the possible residues of

2 square numbers and cube numbers, for example to contradict the
other side of an equation.

If x divided by y gives a remainder of z, then x – z is divisible by y.

3 Use this in problems which specify the remainders for certain divisions.

Experimenting with different modulo can reveal information about your

4 variables, particular for problems involving squared/cubed numbers.

If working in modulo-p arithmetic where p is prime, then we see all the

5 possible residues for each p numbers in an arithmetic sequence, unless
the common difference is a multiple of p.
ζ
Topic 2 – Number Theory

Part 5: Digit Problems

Reasoning about last digits

When we want to find the last digit of some expression, we

can do our arithmetic modulo:

10
?
Using Laws of Modular Arithmetic

Prove that the last digit of a square number can never be 2.

If a square number DID end with a 2, then expressing this in
modular arithmetic:

Since , then . ?
So square numbers cannot end in 2 or 3 or 7 or 8.
Reasoning about last digits

3 1000

31 32 33 34 31000

3 9 7 1 1 (mod 10)

27  7 (mod 10), i.e. we This is a very useful trick!

only ever need to keep If a  1 (mod n), then ak  1k  1 (mod n)
the last digit when we’re So if 34  1 (mod 10), then (34)250  31000  1 (mod 10).
working modulo-10 A strategy to find the last digit in general of ab
arithmetic. therefore is to try and get to 1 by incrementally raising
the power, at which point we can multiply the power
by anything we like!
Reasoning about last digits
[SMC]: The value of 12004 + 32004 + 52004 + 72004 + 92004 is calculated using a
powerful computer.
What is the units digit of the correct answer?

A: 9  B: 7 C: 5

D: 3 E: 1 

The last digit of 34 is 1, as is the last digit of 74 and the last

digit of 92. So the last digit of (34)501, that is of 32004, is 1.
Similarly, the last digit of (74)501, that is of 72004, is 1 and the
last digit of (92)1002, that is of 92004, is 1. Furthermore, 12004 = 1
and the last digit of 52004 is 5. So the units digit of the
expression is 1 + 1 + 5 + 1 + 1, that is 9.
Reasoning about last digits

Question: Find the last non-zero digit of 50! Source: Team SMC

We could use find the complete prime factorisation of 50!.

50! = 247 x 322 x 512 x 78 x 114 x 133 x 172 x 192 x 232 x 29 x 31 x 37 x 41 x 43 x 47.
To find the number of twos for example, we look for multiples of 2 up to 50
(there’s 25), then get bonus 2s from multiples of 4 (there’s 12), then bonus 2s
from multiples of 8 (6) then 16 (3) and the extra 2 from the 32, giving 47 in
total.
We eliminate the 5s to get rid of the zeros ? on the end of 50!, and thus must get
rid of 12 twos as well, leaving 35 twos.
At this point, we can use modulo-10 arithmetic to find the last digit quickly,
which we can do without a calculator because at any point we only ever need
to keep the last digit:
235 x 322 x 78 x 114 x 133 x 172 x 192 x 232 x 29 x 31 x 37 x 41 x 43 x 47
≡ 8 x 9 x 1 x 1 x 7 x 9 x 1 x 9 x 9 x 1 x 7 x 1 x 3 x 7 ≡ 2 (mod 10)
Representing digit problems algebraically

Suppose we have a 2-digit number “”.

Q1: What range of values can each variable have?

: 1 to
? 9 : 0 to
? 9
It couldn’t be 0 otherwise we’d have
a 1-digit number.

Q2: How could we represent the value () of the digit using and ?
e.g. If and , we want

𝑛=10 𝑎+𝑏
? Similarly, a 3-digit number “” could be
represented as
Representing digit problems algebraically
[Hamilton 2005 Q4] An ‘unfortunate’ number is a Use what you know!
positive integer which is equal to 13 times the sum of its I can represent the
digits. Find all ‘unfortunate’ numbers.  digits algebraically
and form an
equation.

? 195
Answer: 117, 156, 
I know each of my
digits can be
Let’s try 2-digit numbers first. Algebraically: between 1 and 9
(and 0 if not the first
10a + b = 13(a + b)
digit)
So 3a + 12b = 0. But this gives us no solutions because one of a or b would
have to be negative.

Now try 3-digit numbers:

100a + 10b + c = 13(a + b + c)
This simplifies to 29a = b + 4c
Suppose a = 1. Then if b=1, c=7, giving 117 as a solution.
We also get a=1, b=5, c=6 and a=1, b=9, c=5.
If a=2 or greater, then the LHS is at least 58. But b + 4c can never be big
enough, because at most b=c=9, so b+4c = 45.

Now try 4-digit numbers:

We get 329a + 29b = c + 4d after simplification. But when a is at its
lowest, i.e. a=1, and b=0, the c+4d can clearly never be big enough.
ζ
Topic 5 – Number Theory

Part 6: Rationality and Miscellaneous

Irrationality of
You may well have seen a proof before for the irrationality of 2. Recall that a rational
number is one that can be expressed as a fraction.
Aristotle’s Proof Something I just thought of...

Assume that √2 is rational. Then it can be

Use a proof by contradiction: Let’s reason about the factors on each
side of the equation .
expressed as a fraction in its simplest form , We know that the powers in the prime
where and are coprime (if they weren’t factorisation of a square number need to
coprime, we’d be able to simplify the fraction. be even. So for each of and , it can either
not contain a 2, or its 2s come in pairs.
Then squaring both sides: Either way, we have an even number of
2s on the LHS of the equation, and an
Then is even, and so is even. odd number on the RHS due to the extra
Therefore let . 2.
Thus the equation has no integer
solutions, i.e. a square number cannot be
so twice another square number.
Therefore is also even. Then and share a
common factor of 2, contradicting that is in its
simplest form.
A rationality BMO problem
Question: Let be a set of rational numbers with the following properties:
1) is an element of
2) If is an element of , then both is an element of and is an element of
Prove that contains all rational numbers in the interval .

What would the structure of our proof look like?:

1. Start with some rational number in the interval .
2. Show somehow that we can use either of the statements in (2) until we eventually get to
?
a half (satisfying (1)), and thus we can always find some chain starting from to get to any
rational number in the interval.
Solution:

irrationality of √2 proof), then and is

We could show that if x is some rational number (for some coprime and , as with the

But we get nicer expressions if we do it backwards: if , then . Similarly, if , then

This means we can subtract the numerator from the denominator, and reciprocate the
fraction if it’s above 1, and still have a value in?the set .
e.g. 5/7  5/2 2/5  2/3  2/1  1/2 (Continued on next slide)
A rationality BMO problem
[BMO 2004/05 Q5] Let be a set of rational numbers with the following properties:
1) is an element of
2) If is an element of , then both is an element of and is an element of
Prove that contains all rational numbers in the interval .

e.g. 5/7  5/2 2/5  2/3  2/1  1/2

All that remains therefore is to prove that we can make sure a chain from any to
eventually get to .

Informally, we could argue that as the numerator or denominator can always

decrease in each step, then one of them will reach 1. If the denominator reaches 1
first and we have , we know is in the set. If we have , then we can always use our
rule to reduce by 1 each time until we reach .

A more formal proof could use a proof by contradiction, found here:

[Link] BMO
Round 2
Round 1
ζ
Topic 2 – Number Theory

Part 7: ‘Epilogue’
The rest of these slides don’t explore any theory that is likely to be use in
any Maths Challenges/Olympiads or university interviews, but explore an
interesting area of Number Theory...
Let’s finish with something light...
Analytic Number Theory!

= ‘Mathematical analysis’ + Number Theory

Using differentiation, integration, Properties of integers.
limits, and usually considering real
and complex numbers.

That’s interesting: we’re using analysis, which concerns real and complex numbers,
to reason about the integers.
Let’s finish with something light...
Analytic Number Theory!
There’s broadly two types of problem studied in this field:

× Those involving multiplication + Those involving addition

...which includes reasoning about factors. e.g. The yet unproven Goldbach
Usually concerns properties of prime Conjecture: “every even integer is the
numbers. sum of two primes”.

Let’s have a tiny bit of a look at prime numbers...

Distribution of primes

Prime Number Theorem:

The probability that a randomly chosen large number is prime is
approximately 1 in
Since the graph of ln(N) always increases but gradually slows down, this suggests
(as we might expect) that primes gradually become more spread out for larger
numbers, but that the gap between prime numbers gradually levels off.

P(10,000 is prime) =
So around this number we’d ‘expect’ roughly 1 in 10 numbers to be prime.

P(1 billion is prime) =

So around this number we’d ‘expect’ roughly 1 in 20 numbers to be prime.
Counting primes

𝜋 ( 𝑥 ) The ‘prime-counting function’, i.e. the number of primes up to

and including x.

So , because there are 4 prime numbers (2, 3, 5, 7) up to 10.

(Note that the symbol is being used a function here, not as the constant you know and love!)

Could we use the Prime Number Theorem to come up with an estimate for ?

Consider 100 people who have been asked to come to your party. If each person has
a 0.3 chance of coming to your party, you’d expect people to come.
But more generally, if each person had different probabilities of coming to your
birthday, you could add the probabilities to get an estimate for the total coming.

Similarly, if we added up the probability of each number of prime up to , we’d get an

estimate of the number of primes up to . So:
Counting primes

But since ln(x) is a continuous function, we may as well use integration instead,
finding the area under the graph:

The function on the RHS is known as the “logarithmic integral”, written

But if we consider the graph of ln(x), and note that as x becomes large, the gradient
of ln(x) becomes 0, and thus we could come up with a looser but easier to calculate
approximation that assumes we use the same probability ln(x) for all numbers up to
x (rather than calculate ln(k) for each k up to x as before).
Then, given the probability is constant, then going back to our party analogy, we can
just multiply this constant probability by the number of people to get the estimate
attendance, i.e. Multiply 1/ln(x) by x to get an estimate number of primes:
 Counting primes

The graph indicates how accurate these two estimates area compared to the
true number of primes .

1.1 means we’ve

overestimated
by 10%

We can see that this estimate is 99%

accurate once we consider the number
of primes up to about 100,000
The th prime?

There’s currently no formula to generate the th prime.

But we can use the approximation seen earlier.

If there are prime numbers up to , this suggests that the () th prime number is
roughly .
That means that the th prime number will be roughly

Example:
The actual 100,000th prime number is just under 1.3 million
And 1.15 million.

This percentage error is reduced as the number becomes larger.

The probability two numbers are coprime?

To solve this problem, let’s first consider the Riemann Zeta Function
(which these resources are named after!)

So for example:

Which curiously comes to (and yes, here means 3.14…)

Euler proved that such as sum is related to a product involving primes:

For example:
The probability two numbers are coprime?

How then is this related to the probability of two numbers being

coprime?

What’s the probability an integer is divisible by 4?

1
4?
What’s the probability that two numbers are 1
divisible by some number p? ?2
𝑝
What’s the probability that neither of two 1
numbers is divisible by a number p?
1 −? 2
𝑝
To consider whether two numbers are coprime, we need to test whether each possible prime
is a factor of both.
We need not test whether they’re both divisible by non-primes, because if for example both
numbers are divisible by 8, we would have already earlier found that they are divisible by 2. It
also ensures we have independence: the probability of a number being divisible by 2 is not
affected by the probability of being divisible by 3, but if a number is divisible by 2 say, then this
increases the chance it’s divisible by 4 (from 0.25 to 0.5).
The probability two numbers are coprime?

Then by considering all possible primes p, the probability is:

The RHS looks familiar! We saw that the product of such expressions
involving primes was the same as the Riemann Zeta Function.

So the probability is ζ(2)-1, which is (π2/6)-1

6
¿ 2
𝜋

I find this remarkable, that , usually associated with circles, would

arise in a probability involving coprimality!