Department of Mathematics, Ranade Bhavan, Kalina

Santacruz(East), Mumbai-400098.

PROJECT NAME :
p-ADIC INTEGERS

SUBMITTED BY :
1. Anjali Singh
2. Kainat Shaikh
3. Mariya Sadaf Shaikh
4. Nasra Momin
5. Sadia Khatoon
6. Safeera Shaikh
7. Shubham Jadhav
8. Suresh Mishra

UNDER THE GUIDANCE OF :

DR. ANURADHA S. GARGE

1
 STATEMENT TO BE INCORPORATED BY THE
CANDIDATES IN THE THESIS AS REQUIRED
UNDER REGULATION FOR THE M.SC DEGREE
As required by the University Regulation No. R.1972/ R.2190 we wish
to state that the work embodied in this thesis titled “p-ADIC INTEGERS
” is expository in nature and does not contain any new results. The entire
work is carried out under the guidance of Dr. Anuradha S. Garge at the
Department of Mathematics, University of Mumbai. This work has not been
submitted for any other degree of this or any other University. Whenever
references have been made to previous works of others, it has been clearly
indicated as such and included in the Bibliography.

PLACE : Vidyanagari, Mumbai.

DATE :

Signature of Students
Anjali Singh
Kainat Shaikh
Mariya Sadaf Shaikh
Nasra Momin
Sadia Khatoon
Safeera Shaikh
Shubham Jadhav
Suresh Mishra

2
 Certificate

This is to certify that, Anjali Singh, Kainat Shaikh, Mariya Sadaf

Shaikh, Nasra Momin, Sadia Khatoon, Safeera Shaikh, Shubham
Jadhav, Suresh Mishra have worked under the supervision of Dr. Anu-
radha S. Garge for M.Sc project titled “p-adic Integers”.

Guide Head of Department

Dr. Anuradha S. Garge Prof. Rajendra M. Pawale

External Examiner

Date : / / 2022

3
 Acknowledgement

We express our heartfelt indebtedness and owe a deep sense of gratitude

to our professor and the guide Dr. Anuradha S. Garge, Department of
Mathematics, University of Mumbai, for her sincere guidance and inspiration
in completing this project. We would like to express our thanks to her for
giving us a great opportunity to do this wonderful project on the topic “p-
adic Integers”, which also helped us in doing a lot of research and we came
to know about so many new things.
We are extremely thankful to Prof. Rajendra M. Pawale, Head of
the Department of Mathematics, University of Mumbai, for his kindness and
encouragement for pursuing us for higher studies and other faculty members
for their co-ordination and cooperation.
We also thanks to all our friends and family members for supporting us
throughout this endeavor.

DATE :
PLACE : MUMBAI

Department of Mathematics,
University of Mumbai,
Kalina, Mumbai-400098.

4
Contents

1 Introduction 6

2 Absolute Value On The Field Q 7

3 Valuation On The Field Q 8

4 p-adic Absolute Value Of Q 12

5 Distance On The Field Q 15

6 p-adic Distance On The Field Q 15

7 Geometry in Qp 18

8 Completions 22

9 p-adic Integers 33

10 Solving Congruences Modulo pn 39

11 Hensel’s Lemma 44

12 Hensel’s Lemma, Second Form 47

5
1 Introduction
Over the last century, p-adic number and p-adic analysis have come to play a
central role in modern number theory. This importance comes from the fact
that they afford a natural and powerful language for talking about congru-
ences between integers and allow the use of methods borrowed from calculus
and analysis for studying such problems. More recently, p-adic number is
shown up in other areas of mathematics and even in physics. The study
of p-adic number is attractive because it blends together so many part of
mathematics.
The p-adic numbers were first introduced by the German mathematician
Kurt Hensel in 1897 (though they are fore-shadowed in the work of his prede-
cessor E. Kummer). It seems that Hensel’s main motivation was the analogy
between the ring of integers Z, together with its field of fractions Q. The
p-adic numbers were then generalized to valuations by Hungarian mathe-
matician József Kürschák in 1913. More formally, for a given prime p, the
field Qp of p-adic number is completion of the rational numbers with respect
to p-adic metric. The field Qp is also given a topology derived from a metric,
which is itself derived from the p-adic order. This metric space is complete
in the sense that every Cauchy sequence converges to a point in Qp . This is
what allows development of calculus on Qp and it is the interaction of this
analytic and algebraic structure that gives the p-adic numbers system their
power and utility.
The p in “p-adic” is a variable and may be replaced with a prime (yield-
ing, for instance, “the 2-adic numbers”) or another expression representing a
prime number. The “adic” of “p-adic” comes from the ending in words such
as dyadic or triadic. The p-adic numbers are most simply a field extension of
Q, the rational numbers which can be formulated in two ways using either an-
alytic or algebraic methods. The p-adic integers are the p-adic numbers with
non-negative valuation. Every integer is a p-adic integer (including zero).
The p-adic numbers are useful because they provide another toolset for
solving problems, one which is sometimes easier to work with than the real
numbers. Suppose we have a physical or any other system and we make
measurments. To describe results of the measurments, we can always use
rationals. According to the Ostrowski’s theorem there are only two kinds
of completions of the rationals. They give real or p-adic Qp numbers fields,
where p is any prime number with corresponding p-adic norm | x |p which is

6
non-archimedean. Also, p-adic numbers have application in number theory,
where it has a significant role in diophantine geometry and diophantine ap-
proximation, some application required the development of functional anal-
ysis, algebra and more. One example is Hensel’s lemma for finding roots of
the polynomial.
To build a mathematical model of the system we will use real or p-adic
numbers or both, depending on the properties of system. We shall start with
introducing the absolute value on the field Q and then valuation on the field
Q. We then move on to p-adic absolute value of Q and some properties. Next
we state the Ostrowiski’s Theorem. Then we define distance on the field Q:
the p-adic distance etc. Then we will look at how the p-adic geometry differs
from Euclidean geometry. Next, we discuss the completion using the concept
of topology to get a sense of p-adic spaces. We then move to p-adic integers.
Finally, we will look at Hensel’s lemma.

The main idea will be to introduce a different absolute value function on

the field of rational numbers. This will give us a different way to measure
distances. Once we have that, we will use it to construct the p-adic numbers.
To get the p-adic numbers, we need to start with the field Q of rational
numbers.

2 Absolute Value On The Field Q

Definition : An absolute value on Q is a function | | : Q → R+ that satisfies
the following conditions: for all x, y ∈ Q,

1. | x |≥ 0;

2. | x | = 0 if and only if x = 0;

3. | xy | = | x || y |;

4. | x + y | ≤ | x | + | y |;

5. | x + y | ≤ max{| x |, | y |}.

The absolute value is non-archimedean if it satisfies the condition 1 to 4;

otherwise, we will say that the absolute value is archimedean if it satisfies
condition 1 to 3 and 5.

7
Note that | x | is usual absolute value.
| x |= 1, if x ̸= 0 and | 0 | = 0 is trivial absolute value.

Now for defining p-absolute value, let us explore some definitions and lemma.

3 Valuation On The Field Q

a1
Definition : Fix a prime number p. Let 0 ̸= x ∈ Q. Then write x = b1
=
pl ×a
pm ×b
, a, b, a1 , b1 ∈ Z such that p ̸ | a and p ̸ | b.
This implies x = pl−m × ab = pn × ab ; n = (l − m) ∈ Z and this n = vp (x),
where vp (x) is said to be p-adic valuation for this x ∈ Q.

Remark : The p-adic valuation of any x ∈ Q, i.e., vp (x) is also defined

as, for x = ab ∈ Q, vp (x) = vp (a) − vp (b), where p ∈ Z is prime.
From above calculation we can see that valuation of p at x, i.e., vp (x), is
just multiplicity of p as divisor of x.

Examples :
(i) v3 (402) = 1, multiplicity of 3 as divisor of 402 is 1, since 402 = 31 × 134.
(ii) v3 ( 123
48
) = v3 (123) − v3 (48) = 1 − 1 = 0, multiplicity of 3 as divisor of 123
and 48 is 1, since 123 = 31 × 41, 48 = 31 × 16.
(iii) vp (1) = 0, multiplicity of any prime number as divisor of 1 is 0, since
1 = p0 × 1.

Note that vp (0) = +∞, since any arbitrarily high power of p divides zero.

We observe that for any x ∈ Q, the value of vp (x) does not depend on it’s
representation as a quotient of two integers. In other words, if ab = dc , then
vp (a) − vp (b) = vp (c) − vp (d).
We will prove this as follows:
Let x = ab = dc .
We have x = ab implies that vp (x) = vp (a) − vp (b). Similarly vp (x) =
vp (c) − vp (d). Hence vp (a) − vp (b) = vp (c) − vp (d).

6 12
Example : As 4
= 8
then v2 (6) − v2 (4) = −1 = v2 (12) − v2 (8).

8
Now we will discuss some basic properties of p-adic valuation.

Lemma : For all x and y ∈ Q, we have

1. vp (xy) = vp (x) + vp (y);

2. vp (x−1 ) = −vp (x);

3. vp (x + y) ≥ min{vp (x), vp (y)}, with equality if vp (x) ̸= vp (y),

with the obvious conventions of vp (0) = +∞.

Proof : First we prove that vp (xy) = vp (x) + vp (y).
Let any x, y ∈ Q.
Let x = pn ab and y = pm dc such that p ̸ | a, b, c, d. Then
ac
xy = pn+m ,
bd
where p ̸ | ac, bd.
By definition we get, vp (xy) = n + m and vp (x) = n and vp (y) = m.
Thus vp (xy) = n + m = vp (x) + vp (y).

Next we prove that vp (x−1 ) = −vp (x).

We know that,

0 = vp (1) = vp (xx−1 ) = vp (x) + vp (x−1 ).

This gives that vp (x−1 ) = −vp (x).

Further, we are trying to prove that vp (x + y) ≥ min{vp (x), vp (y)}.

Let x = pn ab and y = pm dc such that p ̸ | a, b, c, d. Then

vp (x) = n and vp (y) = m.

If x + y = 0 then vp (x + y) = +∞ ≥ min{vp (x), vp (y)} trivially.

We now assume that x + y ̸= 0.
If n = m,

9
Then min{vp (x), vp (y)} = n = m.
Consider,
a c a c
x + y = pn + pm = pn ( + ).
b d b d

Then p-adic valuation of x + y is,

a c
vp (x + y) = vp (pn ) + vp ( + )
b d
ad + bc
= n + vp ( )≥n
bd

because p ̸ | a, b, c, d it implies p ̸ | bd, so vp (bd) = 0.

Also p ̸ | ad, bc it doesn’t imply p ̸ | ad + bc, so vp ( ad+bc
bd
) ≥ 0.
Hence vp (x + y) ≥ min{vp (x), vp (y)}.

If n ̸= m then,
CASE I: If n < m,
Consider,
a c a c a c
x + y = pn + pm = pn + pm+n−n = pn ( + pm−n ).
b d b d b d

We get that vp (x + y) = vp (pn ) + vp ( ab + pm−n dc )

ad + pm−n bc
= vp (pn ) + vp ( ) = vp (pn ) + vp (ad + pm−n bc) − vp (bd).
bd

As we have, p ̸ | a, b, c, d. Hence p ̸ | ad, bc, bd.

This implies that p ̸ | ad + pm−n bc,

i.e., p is not the divisor of ad + pm−n bc and bd.

Hence vp (ad + pm−n bc) = 0 and vp (bd) = 0.

Therefore vp (x + y) = vp (pn ) = n.

Also we have n < m, implies min{vp (x), vp (y)} = vp (x) = n.

10
Thus we get, vp (x + y) = min{vp (x), vp (y)}.

CASE II: If n > m,

Then we have min{vp (x), vp (y)} = vp (y).
Consider,
a c a c a c
x + y = pn + pm = pn+m−m + pm = pm (pn−m + ).
b d b d b d

Then vp (x + y) = vp (pm ) + vp (pn−m ab + dc )

pn−m ad + bc
= vp (pm ) + vp ( ) = vp (pm ) + vp (pn−m ad + bc) − vp (bd).
bd

As we have, p ̸ | a, b, c, d. Hence p ̸ | ad, bc, bd.

This gives that p ̸ | pn−m ad + bc,

i.e., p is not the divisor of pn−m ad + bc and bd.

Hence vp (pn−m ad + bc) = 0 and vp (bd) = 0.

Therefore vp (x + y) = vp (pm ) = m.

Thus we get, vp (x + y) = m = vp (y) = min{vp (x), vp (y)}.

Hence from both the cases 1 and 2 we get vp (x + y) = min{vp (x), vp (y)}.

Thus we get equality if n ̸= m, otherwise vp (x + y) ≥ min{vp (x), vp (y)}.

11
4 p-adic Absolute Value Of Q
Definition : For any x ∈ Q, we define p-adic absolute value of x by

| x |p = p−vp (x) ,

where p ∈ Z is prime.

Examples :

1. For x = 0, | 0 |p = p−vp (0) = p−∞ = 1

p∞
= 1
∞
= 0.

2. | 35 |7 = 7−v7 (35) = 7−1 = 17 .

56
3. | 56
|
12 7
= 7−v7 ( 12 ) = 7−v7 (56)+v7 (12) = 7−1+0 = 7−1 = 17 .
360
4. | 360
|
93555 7
= 7−v7 ( 93555 ) = 7−v7 (360)+v7 (93555) = 7−0+1 = 7.

Lemma : The function | |p : Q → R+ has the properties:

(a) | x |p ≥ 0;

(b) | x |p = 0 if and only if x = 0;

(c) | xy |p = | x |p | y |p ;

(d) | x + y |p ≤ max{| x |p , | y |p }, with equality if | x |p ̸= | y |p .

Proof : Fix a prime number p ∈ Z. Let x, y, z ∈ Q are arbitrary non-zero

rational numbers.

(a) For all 0 ̸= x ∈ Q we have | x |p = p−vp (x) .

Since p ∈ Z is prime, p > 0 and hence p−vp (x) > 0.
Hence | x |p > 0.

(b) Note that | x |p = 0 if and only if p−vp (x) = 0

if and only if vp (x) = ∞
if and only if x = 0.

12
(c) | xy |p = p−vp (xy)
= p−(vp (x)+vp (y))
= p−vp (x) p−vp (y)
=| x |p | y |p .

(d) If | x |p = | y |p ,
then max{| x |p , | y |p } = | x |p = | y |p .
As | x |p = | y |p .
By p-adic absolute value, p−vp (x) = p−vp (y) .
This gives vp (x) = vp (y).
Since by property 2 of valuation, vp (x + y) ≥ vp (x).
So −vp (x + y) ≤ −vp (x).
This gives p−vp (x+y) ≤ p−vp (x) .
We get that | x + y |p ≤ | x |p .
Hence | x + y |p ≤ max{| x |p , | y |p }.

If | x |p ̸= | y |p ,

CASE I: | x |p > | y |p .
Then max{| x |p , | y |p } = | x |p .
As | x |p > | y |p .
We have p−vp (x) > p−vp (y) , which gives −vp (x) > −vp (y).
We can write vp (x) < vp (y).
Then vp (x) = min{vp (x), vp (y)} = vp (x + y).
We get −vp (x + y) = −vp (x).
This implies p−vp (x+y) = p−vp (x) .
Hence | x + y |p = | x |p = max{| x |p , | y |p }.

CASE II: | x |p < | y |p .

Then max{| x |p , | y |p } = | y |p .
As | x |p < | y |p , we get p−vp (x) < p−vp (y) .
This gives −vp (x) < −vp (y),

13
 i.e., vp (x) > vp (y).
Then vp (y) = min{vp (x), vp (y)} = vp (x + y),
i.e., −vp (x + y) = −vp (y).
This implies p−vp (x+y) = p−vp (y) .
Hence | x + y |p = | y |p = max{| x |p , | y |p }.

From both cases we get that | x + y |p = max{| x |p , | y |p }.

Hence we have | x + y |p ≤ max{| x |p , | y |p } and this is equal when
| x |p = | y |p .
Thus we get that the p-adic absolute value on Q is properly constructed
so we can conclude following:

Proposition : The p-adic absolute value is non-archimedean absolute value

on Q.
Proof : Proof follows from previous lemma.

Now we state the theorem that classifies all non-trivial absolute values
defined on the field of the rational numbers Q.

Ostrowski’s Theorem : [Alexander Ostrowski (1916)] Every non-trivial

absolute value on the rational numbers Q is equivalent to either the usual
absolute value | |∞ or a p-adic absolute value | |p , for some prime p.
The whole idea of an absolute value is that it provides us with a notion
of “size”. In other words, once we have an absolute value, we can use it to
measure distances between numbers, to define a metric on our field Q.

14
5 Distance On The Field Q
Definition : Let Q be a field of rationals numbers and | | an absolute value
on Q. We define the distance d(x, y) between two elements x, y ∈ Q by

d(x, y) = | x − y | .

The function d(x, y) is called the metric induced by absolute value.

Now we see the general properties for a metric d(x, y) as follows, for any
x, y, z ∈ Q:

(a) d(x, y) ≥ 0;

(b) d(x, y) = 0 if and only if x = y;

(c) d(x, y) = d(y, x) (Symmetric Property);

(d) d(x, z) ≤ d(x, y) + d(y, z) (Triangle Inequality).

Here the last inequality is called the triangle inequality, since it expresses the
usual fact that the sum of the lengths of two sides of triangle is bigger than
the length of the other sides.
A set on which the metric is defined is called the ‘metric space’.

Lemma : | x | is non-archimedean if and only if for any x, y, z ∈ Q, we

have
d(x, y) ≤ max{d(x, z), d(z, y)}.
Now we have a new concept of the distance between two rationals numbers
which is different from the usual absolute value. We define it in the next
section.

6 p-adic Distance On The Field Q

Definition : Let Q be a field of rationals numbers, p be a prime and | |p
be a p-adic absolute value on Q. We define the distance d(x, y) between two
elements x, y ∈ Q by
d(x, y) =| x − y |p ,

15
for some fixed prime p.

Examples : Take p = 7, (i) d(3, 2) = | 3 − 2 |7 = | 1 |7 = | 70 × 1 |7 = 7−0 = 1;

(ii) d(28814, 2) = | 28814 − 2 |7 = | 28812 |7 = | 74 × 13 |7 = 7−4 = 1/74 .
In above example 1 > 1/74 so we can conclude that 28814 and 2 are
“closer together” than 3 and 2 are!

The above metric d satisfies the following properties:

Proposition : For all x, y, z ∈ Q, we have

(a) d(x, y) ≥ 0;
(b) d(x, y) = 0 if and only if x = y;
(c) d(x, y) = d(y, x) (Symmetric Property);
(d) d(x, z) ≤ d(x, y) + d(y, z) (Triangle Inequality).
Proof : Let x, y, z ∈ Q.
(a) Since x and y ∈ Q then x − y ∈ Q.
Let x − y = pn ab , where p ̸ | a, p ̸ |b.
Then vp (x − y) = n.
Consider d(x, y) = | x − y |p = p−vp (x−y) = p−n > 0, as p ∈ Z is prime.

(b) Note that d(x, y) = 0 if and only if | x − y |p = 0

if and only if p−vp (x−y) = 0
if and only if vp (x − y) = ∞
if and only if x − y = 0
if and only if x = y.

(c) Since x and y ∈ Q, x − y ∈ Q.

Let x − y = pn ab , where p ̸ | a, p ̸ | b.
Then vp (x−y) = n. We can also write y −x = −pn ab , where p ̸ | a, p ̸ |b.
Then vp (y − x) = n = vp (x − y).
Consider d(x, y) = | x − y |p = p−vp (x−y) = p−n = p−vp (y−x) =| y − x |p
= d(y, x). Thus d(x, y) = d(y, x).

16
 (d) Consider d(x, z) = | x − z |p

= | x + y − y − z |p

= | (x − y) + (y − z) |p
≤ max{| x − y |p , | y − z |p }
≤ | x − y |p + | y − z |p
= d(x, y) + d(y, z).

Hence we get d(x, z) ≤ d(x, y) + d(y, z).

Proposition : If d(x, z) ≤ max{d(x, y), d(y, z)} then the metric defined
by d is non-archimedean.
Proof : Let x, y, z ∈ Q.
Consider d(x, z) = | x − z |p

= | x + y − y − z |p

= | (x − y) + (y − z) |p
≤ max{| x − y |p , | y − z |p }
= max{d(x, y), d(y, z)}.
Thus,
d(x, z) ≤ max{d(x, y), d(y, z)}.

This inequality is known as the “ultrametric inequality”, and a metric for

which it is true is sometimes called an “ultrametric”.
A space with an ultrametric is called an “ultrametric space”.

17
7 Geometry in Qp
Once we have a way to measure distances, we can do the geometry. So to do
the geometry in Qp we start with three following remarks:

Remarks :
(a) A point is an element of Qp .

(b) A triangle is three distinct points a, b and c in Qp . We will denote a

triangle by △abc.

(c) A triangle abc has sides of lengths given by d(a, b), d(b, c) and d(c, a).
Now we will see a geometry different from the euclidean geometry in
which all triangles are isosceles.

Theorem : In Qp , all triangles are isosceles.

Proof : Let △abc be a triangle. Let a, b and c be distinct points in Qp .
Then d(a, b), d(b, c) and d(a, c) are the lengths of the sides of the triangle
determined by those points.
If any two of d(a, b), d(b, c) and d(a, c) are equal, then we have nothing to
prove.
Without loss of generality, suppose d(a, b) ̸= d(b, c).
Then | a − b |p ̸= | b − c |p .
Here we note that | a − c |p = | (a − b) + (b − c) |p .
By using (d) property of p-adic absolute value of Qp , i.e.,
If | x |p ̸=| y |p then | x + y |p = max{| x |p , | y |p }, we get,

| a − c |p = max{| (a − b) |p , | (b − c) |p },

i.e., we get either | a − c |p = | (a − b) |p or | a − c |p = | (b − c) |p .

This implies either d(a, c) = d(a, b) or d(a, c) = d(b, c),

i.e., we get that, the length of atleast two sides of the triangle is always equal
in Qp . Thus all triangles are isosceles in Qp .

18
Example : Let Q5 with | |5 .
Let x = 200, y = 185, z = 85 ∈ Q5 be vertices of a triangle.
By 5-adic metric we have,
d(x, y) =| x − y |5 = | 200 − 185 |5 = | 15 |5 = 1/5
and
d(y, z) =| y − z |5 = | 185 − 85 |5 = | 100 |5 = 1/25.
It implies that d(x, y) ̸= d(y, z).
Now consider,
d(x, z) =| x − z |5 = | x − z + y − y |5 = | (x − y) + (y − z) |5
= | (200 − 185) + (185 − 85) |5 = | 15 + 100 |5 .
Since d(x, y) ̸= d(y, z), therefore by using property (d) of p-adic absolute
value of Qp ,
i.e., if | x |p ̸=| y |p then | x + y |p = max{| x |p , | y |p }, we get,
d(x, z) =| 15 + 100 |5 = max{| 15 |5 , | 100 |5 } = max{1/5, 1/25} = 1/5.
Hence we get, d(x, y) = d(x, z).
Therefore △xyz is isosceles triangle.

Theorem : If a triangle is not equilateral, the unequal side has the largest
valuation and hence becomes the shortest side of the triangle.
Proof : Let △abc be a triangle. Let a, b and c be distinct points in Qp .
Then d(a, b), d(b, c) and d(a, c) are the lengths of the sides of the triangle
determined by those points.
Since △abc is not equilateral, without loss of generality, let d(a, b) = d(b, c) ̸=
d(a, c).
Then | a − b |p = | b − c |p ̸= | a − c |p ,
i.e., p−vp (a−b) = p−vp (b−c) ̸= p−vp (a−c) .
We get vp (a − b) = vp (b − c) ̸= vp (a − c).
Then vp (a − c) = vp (a − c + b − b) = vp ((a − b) + (b − c))
≥ min{vp (a − b), vp (b − c)}, by property (3) of valuation on the field Q.
Since vp (a − c) ̸= vp (a − b) = vp (b − c) and the triangle is not equilateral, we
have vp (a − c) > vp (a − b).
Hence the unequal side has largest valuation.

Now, to show that the unequal side has the shortest length.
From above, we have vp (a − c) > vp (a − b) or vp (a − c) > vp (b − c).

19
Let vp (a − c) > vp (a − b).
Then pvp (a−c) > pvp (a−b) .
This gives 1/pvp (a−c) < 1/pvp (a−b) ,
i.e., p−vp (a−c) > p−vp (a−b) ,
i.e., | a − c |p < | a − b |p .
Thus d(a, c) < d(a, b). Hence the unequal side has the shortest length.

In Euclidean geometry with the usual distance, collinearity is possible

with more than two point, but if we use the p-adic distance on Q, we see
that collinearity is impossible with more than two points.

Theorem : Given three points a, b and c ∈ Qp , d(a, c) < d(a, b) + d(b, c).
In other words, no three points in Qp are collinear.
Proof : Let △abc be a triangle. Let a, b and c be distinct points in Qp .
Then d(a, b), d(b, c) and d(a, c) are the lengths of the sides of the triangle
determined by points a, b, c.
By triangle inequality of p-adic distances on the field Q, we have

d(a, c) ≤ d(a, b) + d(b, c). (1)

We have to prove that no three points are collinear,

i.e., d(a, c) < d(a, b) + d(b, c) (since three points a, b, c are collinear if and
only if d(a, c) = d(a, b) + d(b, c)), for this we use contradiction method.
Assume that these three points a, b and c are collinear this implies

d(a, c) = d(a, b) + d(b, c).

So we have d(a, c) > d(a, b) and d(a, c) > d(b, c).

Since d(a, c) > d(a, b) then | a − c |p > | a − b |p ,

i.e., p−vp (a−c) > p−vp (a−b) i.e., vp (a − c) < vp (a − b). (2)

Similarly,
d(a, c) > d(b, c) i.e., vp (a − c) < vp (b − c). (3)

As vp (a − c) ≥ min{vp (a − b), vp (b − c)}, by property (3) of valuation on the

field Q.

20
It means that either vp (a − c) ≥ vp (a − b) or vp (a − c) ≥ vp (b − c).
So its a contradiction to equations (2) and (3).
Hence our assumption is wrong.
Therefore d(a, c) ̸= d(a, b) + d(b, c).
Thus by equation (1),
d(a, b) < d(a, c) + d(b, c).
Thus three points a, b and c are not collinear.

Example : Let a, b, c be the three points in Q3 .

Let a = 100, b = 75 and c = 93.
By 3-adic metric, d(x, y) = | x − y |3 ;
d(a, b) =| 100 − 75 |3 = | 25 |3 = 3−v3 (25) = 1;
d(a, c) =| 100 − 93 |3 = | 7 |3 = 3−v3 (7) = 1;
d(b, c) =| 75 − 93 |3 = | −18 |3 = 3−v3 (−18) = 3−2 = 1
32
.
We can see that d(a, c) ̸= d(a, b) + d(b, c).
Hence a, b, c are not collinear.

21
8 Completions
We are now ready to construct the p-adic field Qp . So first we need to recall
some important concepts from basic topology.

Definition : Let K be a field and let | | be absolute value on K.

(a) A sequence of elements (xn ) ∈ K is called a Cauchy sequence if for

every ϵ > 0 there exists N ∈ N such that | xn − xm |< ϵ whenever
n, m > N .

(b) The field K is called Complete with respect to the absolute value | | if
every Cauchy sequence of elements of K are convergent in K (i.e., it
has limit that is also in K).

The first useful thing to note is that the Cauchy sequences can be char-
acterized much more simply when the absolute value is non-archimedean.

Lemma : A sequence (xn ) of rational numbers is a Cauchy sequence with

respect to non-archimedean absolute value | |p if and only if we have

lim | xn+1 − xn |p = 0.
n→∞

We already know that Q is not complete with respect to the metric given
by the usual absolute value. Is Q complete with respect to the new distances
or not ?
The answer to this question is no. We will show this in the following lemma:

Lemma : The field Q of rational numbers is not complete with respect

to any of its nontrivial absolute values corresponding to odd prime p.
Proof : By Ostrowski’s theorem, any non-trivial absolute value on Q is ei-
ther | |∞ or | |p . In case of usual absolute value, i.e., | |∞ , we already know
Q is not complete, so we look for p-adic absolute values.
We take non-trivial absolute value to be | |p , for some odd prime p and try
to prove that Q is not complete with respect to | |p . To prove this, we need
to construct a Cauchy sequence in Q which does not have limit in Q.
Suppose p is a odd prime. Choose an integer a ∈ Z such that

• a is not a square in Q;

22
 • p does not divide a;

• a is a quadratic residue modulo p, i.e., the congruence x2 ≡ a mod p

has a solution.

Note that a exists as there are exactly half the residues modulo p and half
non-residues modulo p.
Now we can construct a Cauchy sequence (with respect to ||p ) in the following
way:

• choose x0 to be any solution of x2 ≡ a mod p, i.e., x20 ≡ a mod p where

1 ≤ x0 ≤ p − 1 and x0 ̸= 0 as x0 should be invertible.

• choose another solution x1 such that x1 ≡ x0 mod p and x21 ≡ a mod p2 .

We prove that such x1 mod p exists, as x1 ≡ x0 mod p, we have x1 =
x0 + pt. Now
x21 ≡ a mod p2
x21 = a + p2 k
(x0 + pt)2 = a + p2 k
x20 + 2px0 t + p2 t2 = a + p2 k
x20 + 2px0 t = a + p2 (k − t2 )
a + pλ + 2px0 t = a + p2 k ′ .

Thus, p(2x0 t + λ) = p2 k ′ this gives 2x0 t + λ ≡ 0 mod p.

Choose t ≡ (2x0 )−1 (−λ) mod p, value of t exist as we are in congruence
modulo p. This proves the existence of x1 .
Similarly we can choose x2 such that x2 ≡ x1 mod p2 and x22 ≡ a mod p3
and continuing like this in general we can choose xn ≡ xn−1 mod pn and
x2n ≡ a mod pn+1 .

We can see that such a sequence (xn ) does exist as the initial element x0
exists.
Now we check that the sequence obtained above is a Cauchy sequence. It is
clear from the construction that xn+1 ≡ xn mod pn+1 .
This implies that xn+1 = xn + λpn+1 . Hence

| xn+1 − xn |p = | λpn+1 |p ≤ p−(n+1) → 0, as n → ∞.

23
Hence by previous lemma we see that the sequence (xn ) is indeed a sequence.
We also know that x2n ≡ a mod pn+1 . So, x2n = a + µpn+1 and hence

| x2n − a |p = | µpn+1 |p ≤ p−(n+1) → 0, as n → ∞.

Thus we get x2n converges to a.

i.e., | x2n − a | < ϵ implies x2n = a, which is contradiction as a is not a square
in Q. Hence Q is not complete with respect to | |p , where p is odd prime.

Now for completion we emphasize p and Q. For this we need to add more
points to Q as completion of Q depends on the metric used and is based on
the notion of Cauchy sequences.

Definition : Let | |p be a non-archimedean absolute value on Q. We denote

by C, or by Cp (Q), the set of all Cauchy sequences of elements of Q.

C = Cp (Q) = {(xn ) : (xn ) is a Cauchy sequence with respect to | |p }.

Proposition : By defining

(xn ) + (yn ) = (xn + yn )

(xn )(yn ) = (xn yn ),

the set C becomes a commutative ring with unity.

Proof : Note that the addition and multiplication be commutative and as-
sociative as well as distributive, as Q forms a ring itself. Now only remains
to show that the set C contains unity. For this we take the constant Cauchy
sequence (xn ) = (1, 1, 1, ...) which is converges to 1.
Consider (xn )(yn ) = (1, 1, 1, ...)(yn ) = (yn ) = (yn )(1, 1, 1, ...) = (yn )(xn ).

Now we only need to check that the sum and product of two Cauchy se-
quences, as defined in above proposition, are also Cauchy sequence.

Let (xn ) be a Cauchy sequence in Q with respect to | |p .

By definition, for some ϵ > 0, ∃N ∈ N such that | xn − xm |p < 2ϵ , for all

24
n > N.
Simillarly, let (yn ) be a Cauchy sequence in Q with respect to | |p .
By definition, for some ϵ > 0, ∃M ∈ N such that | yn − ym |p < 2ϵ , for all
n > M.
Choose n > max{N, M }.
Consider for ϵ > 0 and n > max{N, M },

| (xn + yn ) − (xm + ym ) |p ≤ | xn − xm |p + | yn − ym |p
ϵ ϵ
< + = ϵ.
2 2
This implies | (xn + yn ) − (xm + ym ) |p < ϵ, for all n > max{N, M }.
Thus (xn + yn ) is a Cauchy sequence in Q with respect to | |p .
Further, for multiplication we have,

| xn yn − xm ym |p = | xn yn − yn xm + yn xm − xm ym |p

= | yn (xn − xm ) + xm (yn − ym ) |p
≤ | yn |p | xn − xm |p + | xm |p | yn − ym |p
ϵ ϵ
< | yn |p + | xm |p
2 2
1ϵ 1ϵ 1 ϵ ϵ 1
< + = ( + ) = ϵ < ϵ,
k2 k2 k 2 2 k

as a Cauchy sequence (xn ) and (yn ) is bounded by some 1/k, where k > 0.
This implies | xn yn − xm ym |p < ϵ, for all n > max{N, M }.
Thus (xn yn ) is also a Cauchy sequence in Q with respect to | |p .
Hence the sum and product of two Cauchy sequences is also a Cauchy se-
quence. Therefore closure property is also satisfied under addition and mul-
tiplication. Thus the set C forms a commutative ring with unity.

Now we have a set of all Cauchy sequences C which forms a commutative

ring with unity. Then, we can ask the question that “can a ring C forms a
field too ?”

Let (xn ) ∈ C, where (xn ) ̸= 0 = (0, 0, 0, ....). If any term in (xn ) equals

25
zero (for example, (1, 0, 1, 0, ...)), then (xn ).(yn ) ̸= 1 = (1, 1, 1, ...) for any
(yn ) ∈ C, also 0 is not invertible, and so (xn )−1 does not exist, for any
(xn ) ∈ C.

Hence we conclude that the ring C does not form a field, because not all
non-zero elements are invertible. Moreover, it contains zero divisors as the
product of the non-zero two sequences in C (for example (0, 1, 1, 1, ...) and
(1, 0, 0, 0, ...)) is clearly 0 = (0, 0, 0, ...).

Now we define f : Q → C by constant sequence f (x) = {x, x, x, ...} denoted

by (x). Note that f is 1-1 and f (Q) ⊂ C. So, we say there is an inclusion of
Q into C denoted Q ,→ C.

We say two Cauchy sequences are equivalent when they share the same limit
and we define the set N ⊂ C of sequences that tends to zero.
Definition : We define N ⊂ C to be the ideal
N = {(xn ) : xn → 0} = {(xn ) : lim | xn |p = 0}
n→∞

of sequences that tend to zero with respect to absolute value | |p .

Now, we will show that N is an ideal of C.

Let (xn ), (yn ) ∈ N. Then xn → 0 and yn → 0,
i.e., limn→∞ | xn |p = 0 and limn→∞ | yn |p = 0, by definition of N.
Consider,
(xn − yn ) = (xn + (−yn )) = (xn ) + (−yn ) = (xn ) − (yn ) → 0 − 0 = 0.
Hence (xn − yn ) ∈ N.
Now, for (zn ) ∈ C and (xn ) ∈ N.
Consider,
limn→∞ | xn .zn |p = limn→∞ (| xn |p . | zn |p )
= limn→∞ | xn |p . limn→∞ | zn |p = 0. limn→∞ | zn |p = 0.
Simillarly, limn→∞ | zn .xn |p = 0.
Hence (xn ).(zn ) ∈ N and (zn ).(xn ) ∈ N.
Therefore, N is an ideal of C.

Now, we will show that N is maximal ideal of C.

26
Lemma : N is a maximal ideal of C.
Proof : Let (xn ) ∈ C be a Cauchy sequence that does not tend to zero (i.e.,
does not belong to N).
Let I be the ideal generated by (xn ) and N, i.e., I = ⟨x, N⟩.
We get N ⊂ I ⊂ C, we have to prove that N is maximal ideal, i.e., I = N or
I = C.
So we want to show that I must be all of C if I ̸= N. We will do that by
showing that the unit element (1) (i.e., the constant sequence corresponding
to 1) is in I.

This is enough, because any ideal that contains the unit element must be
the whole ring.

Now, since (xn ) does not tend to zero and is a Cauchy sequence, it must
“eventually” be away from zero, that is, there must exist a number c > 0
and an integer N such that | xn | ≥ c > 0, whenever n ≥ N . Now in particu-
lar this means that xn ̸= 0 for n ≥ N , so that we may define a new sequence
(yn ) setting yn = 0 if n < N and yn = x1n if n ≥ N .

The first thing to check is that (yn ) is a Cauchy sequence. That is clear
because if n ≥ N we have
 
 1 1  | xn+1 − xn |p | xn+1 − xn |p
| yn+1 − yn |p = 
 −  = ≤ → 0,
xn+1 xn p | xn xn+1 |p c2

as (xn ) is Cauchy sequence.

It shows that (yn ) is Cauchy sequence, i.e., (yn ) ∈ C.

Now notice that, (

0, if n < N
xn y n =
1, if n ≥ N .

This means that the product sequence (xn )(yn ) consists of a finite number
of 0’s followed by an infinite string of 1’s,
i.e., (xn )(yn ) = (0, 0, · · · , 0, 1, 1, · · · ). In particular, if we subtract it from the
constant sequence (1), we get a sequence that tends to zero,
i.e., (1, 1, · · · ) − (0, 0, · · · , 0, 1, 1, · · · ) = (0, 0, · · · ).

27
In other words
(1) − (xn )(yn ) ∈ N.
It means that
(1) ∈ (xn )(yn ) + N.
This says that (1) can be written as a multiple of (xn ) plus an element of N.
Hence (1) belongs to I, as we had claimed. Thus N is maximal ideal of C.

Now by taking the quotient of the ring C by the ideal N, we make things
even nicer, i.e., taking a quotient of the ring by a maximal ideal gives us a
field. Momentarily, we define this field:

Definition: We define the field of p−adic numbers to be the quotient of

the ring C by its maximal ideal N:
Qp = C/N
and we call it the field of p-adic numbers Qp .

For any two Cauchy sequences (xn ) and (yn ), if | xn − yn |p → 0 as n → ∞

then the two Cauchy sequences are equivalent.
Then we say the set of equivalence classes of Cauchy sequences, Qp , the
p-adic numbers.

Note : The p-adic absolute value of an equivalence class x is limn→∞ | xn |p ,

where (xn ) is any sequence which represents x,
i.e., | x |p = limn→∞ | xn |p .

Now we define addition, multiplication, multiplicative inverse and addi-

tive inverse on Qp .

Let x and y be the two equivalence classes of Cauchy sequences with repre-
sentatives (xn ) and (yn ) respectively, i.e., x, y ∈ Qp .

We show that x.y to be the equivalence class, which is represented by the

Cauchy sequence (xn yn ), i.e., x.y ∈ Qp .
If we have another two Cauchy sequence (x′n ) and (yn′ ) which also represents
x and y respectively then we would have,
| x′n yn′ − xn yn |p = | x′n yn′ − x′n yn + x′n yn − xn yn |p

28
 = | x′n (yn′ − yn ) + yn (x′n − xn ) |p
≤ max{| x′n (yn′ − yn ) |p , | yn (x′n − xn ) |p }
= max{| x′n |p | yn′ − yn |p , | yn |p | x′n − xn |p }.
The limit of terms in right hand side of above expression is
| x |p limn→∞ | yn′ − yn |p = 0 and | y |p limn→∞ | x′n − xn |p = 0, (since x
and y are equivalence classes).
This implies that limn→∞ | x′n yn′ − xn yn |p = 0.
Therefore (x′n yn′ ) ∼ (xn yn ).
This implies that x.y ∈ Qp , it means that multiplication is defined on Qp .

Now we want to define multiplicative inverse on Qp .

For this we have to prove that 1/x to be the equivalence class with the
representative (1/xn ), i.e., 1/x ∈ Qp .
Let 1/x′n be another Cauchy sequence which also represent x.
Then we would have,
xn − x′n
| 1/x′n − 1/xn |p = | ′
|p = | xn − x′n |p | (x′n )−1 (xn )−1 |p
x n xn
= | xn − x′n |p | 1/(x′n xn ) |p .
Taking limit as n tends to infinity on both the side,

limn→∞ | 1/x′n − 1/xn |p = limn→∞ | xn − x′n |p limn→∞ | 1/(x′n xn ) |p ,

but limn→∞ | xn − x′n |p = 0 (as x is equivalence class).

This gives that limn→∞ | 1/x′n − 1/xn |p = 0.

Hence (1/x′n ) ∼ (1/xn ).

It implies that 1/x ∈ Qp , i.e., multiplicative inverse is defined on Qp .

Now we define x + y to be equivalence class that is represented by the

Cauchy sequence (xn + yn ) , i.e., x + y ∈ Qp .
Once again if we have another two Cauchy sequences (x′n ) and (y) which also
represents x and y respectively then we would have:
| (x′n + yn′ ) − (xn + yn ) |p = |x′n − xn + yn′ − yn |p

29
 ≤ max{| x′n − xn |p , | yn′ − yn |p }.
Then limit of right hand side of above expressions is,
limn→∞ (| x′n − xn |p ) = 0 and limn→∞ (| yn′ − yn |p ) = 0 (since x and y are
in Qp ).
This implies that limn→∞ | (x′n + yn′ ) − (xn + yn ) |p = 0.
Hence (x′n + yn′ ) ∼ (xn + yn ).
This implies that x + y ∈ Qp . This means that addition is defined on Qp .

Now define −x to be equivalence class that is represented by Cauchy

sequence (−xn ), i.e., (−x) ∈ Qp .
If we have (−x′n ) which also represents −x then we have

| (−x′n ) − (−xn ) |p = | −x′n + xn |p

=| −(x′n − xn ) |p = | x′n − xn |p .

As x is in Qp , limn→∞ | x′n − xn |p = 0.
Therefore by above expression we have limn→∞ | (−x′n ) − (−xn ) |p = 0.
Hence (−x′n ) ∼ (−xn ).
It means that −x ∈ Qp , i.e., additive inverse is also defined on Qp .

With addition, multiplication and inverses as defined above now we show

that the set of equivalence classes of Cauchy sequence Qp , is a field.

Let the Cauchy sequences (xn ), (yn ), (zn ) be representatives of x, y, z ∈ Qp .

Now let us check for the distributive law; x(y + z) is the equivalence class of,

(xn (yn + zn )) = (xn yn + xn zn )

so xy + xz is also an equivalence class. Hence x(y + z) = xy + xz, so the

distributive law holds.

Now the associative law; x + (y + z) is the equivalence class of,

(xn + (yn + zn )) = (xn + yn + zn ) = ((xn + yn ) + zn )

30
so (x + y) + z is also an equivalence class.
Hence x + (y + z) = (x + y) + z, so the associative law holds.

Now commutative law; x + y is the equivalence class of,

(xn + yn ) = (yn + xn )

so y + x is also an equivalence class.

Hence x + y = y + x, so the commutative law holds.
Thus by all above defined properties we can conclude that Qp is a field.

Notice that the two different constant sequences never differ by an element
of N (their difference is just another constant sequence). Hence we still have
an inclusion
Q ,→ Qp
by sending x ∈ Q to the equivalence class of the constant sequence (x).
To check we have indeed obtained the completion, we must now check the
remaining two requirements: that Q is dense in Qp , and that Qp is complete.

Proposition : The image of Q under the inclusion Q ,→ Qp is a dense

subset of Qp .
Proof : Given the inclusion of Q into Qp denoted Q ,→ Qp , therefore we
have f (Q) ⊂ Qp . We have to show that f (Q) is dense in Qp . That is we
have to show, intersection of any open ball in Qp with f (Q) is non-empty,
i.e., we need to show that any open ball around an element in Qp contains an
element of (the image of) Q, i.e., a constant sequence, so fix a radius r > 0.
Let λ = (xn ) ∈ Qp be a Cauchy sequence and let ϵ be a number slightly
smaller than r. By Cauchy property, for ϵ > 0, ∃ an natural number N such
that | xn − xm |p < ϵ whenever n, m ≥ N .
Let x = xN and consider the constant sequence (x).
Claim: f (Q) ∩ B(λ, r) ̸= ∅, for every λ ∈ Qp and for fix r > 0,
i.e., for (x) ∈ f (Q), (x) ∈ B(λ, r),
i.e., | λ − (x) |p < r.
Consider, | λ − (x) |p = | (xn ) − (x) |p = | (xn − x) |p = limn→∞ | xn − x |p
= limn→∞ | xn − xN |p < ϵ < r.
Thus constant sequence of (x) ∈ f (Q) belong to B(λ, r) for λ ∈ Qp and for

31
some fix radius r > 0. Hence we are done.

Now, it remains to check the completion of the field of p-adic numbers Qp .

Proposition : The field of p-adic numbers Qp is complete with respect

to | |p .
Proof : Let λ1 , λ2 , · · · , λn , · · · be a Cauchy sequence of elements of Qp (so
that each λi , for i = 1, 2, · · · , n, · · · is a Cauchy sequence of elements of Q,
taken upto equivalence).
As we know that the image of Q is dense in Qp , so we can find the ratio-
nal numbers y (1) , y (2) , · · · , y (n) , · · · such that a Cauchy sequence λn and the
constant sequence y (n) are arbitrarily close, i.e., we have
limn→∞ | λn − (y (n) ) |p = 0.
This shows that (λn − y (n) ) is zero sequence in Qp .
We can write y (n) = λn − λn + y (n) = λn − (λn − y (n) ).
As (λn ) is a Cauchy sequence in Qp and (λn − y (n) ) is zero sequence hence a
Cauchy sequence.
We get that (y (n) ) is a Cauchy sequence in Q as well as Qp ,
i.e., the rational numbers y (1) , y (2) , · · · , y (n) , · · · themselves form a Cauchy
sequence in Q. Let λ denote the element of Qp corresponding to this se-
quence, i.e., λ = y (n) + N. Now we can write λ − λn = (λ − y (n) ) + (y (n) − λn ).
By the denseness of Q in Qp , λ − y (n) converges to zero and (y (n) − λn ) also
converges to zero as it is a zero sequence. Hence λ − λn converges to zero in
Qp , i.e., limn→∞ λn = λ.
We get that a Cauchy sequence λn in Qp converges to λ ∈ Qp .
Thus Qp is complete with respect to | |p because every Cauchy sequence in
Qp is convergent.

Now we begin to explore the structure of the field Qp . Since Qp is the

field with non-archimedean valuation, we can consider the corresponding
valuation ring. The resulting ring has a name of its own:

32
9 p-adic Integers
We shall devote this section to the p-adic integers which we denote by Zp .
Definition : The ring of p-adic integers is the valuation ring

Zp = {x ∈ Qp : | x |p ≤ 1}.

Examples :
(i) For p = 2, | 3 |2 = 2−v2 (3) = 2−0 = 1. Thus 3 ∈ Z2 .
(ii) For p = 2, | 1
|
4 2
= 2−v2 (1/4) = 2−v2 (1)−(−v2 (4)) = 2−(−2) = 4 > 1.
Thus 4 ∈/ Z2 .

Remarks :

• Here we observe that all p-adic absolute values of integers Z is always

less than are equal to 1 hence all integers lies in Zp .

• Also note that Zp is closed unit ball with centre 0, therefore it is a

closed set in Qp , as every closed ball is closed set.

Now we discuss more about the ring of the p-adic integers Zp . For this
we first define local ring and then recall the following proposition from the
abstract algebra.

Local Ring : A ring R is said to be a local ring if it has exactly one

maximal ideal.

Proposition : Let R be a commutative ring with unity and if M ̸= (1)

be any ideal of commutative ring R with unity such that every x ∈ R \ M is
an unit of R. Then R is local ring and M is its maximal ideal.
Proof : We prove that M is the unique maximal ideal of R.
We first prove that M is a maximal ideal of R.
This is easy because for every x ∈ R \ M, the ideal generated by M and x
contains an unit and hence the ideal generated by M and x becomes the unit
ideal (1). Thus we proved M is maximal.
Now we prove M is unique with the above property, i.e., every element of M
is unit. Suppose there exits M ′ a maximal ideal of R with the same property
as above, i.e., every element of M ′ is unit.
CASE I: M ⊊ M ′

33
Then every x ∈ M ′ \ M is an unit.
This implies M ′ = R,
i.e., M ′ can not be a maximal ideal of R.
Thus we get M ⊂ M ′ .

CASE II: M ′ ⊊ M
This implies every element outside M ′ is a unit, hence 1 ∈ M , which is a
contradiction as M is proper ideal of R.
Thus M ′ ⊂ M .
Hence by both cases I and II we get M = M ′ ,
i.e., M is the unique maximal ideal of R.
Thus R is a local ring.

Lemma : The valuation ring Zp is local ring with unique maximal ideal
pZp = {x ∈ Zp : | x |p < 1}. We shall call pZp the valuation ideal of Zp .
Proof : We will show that pZp is an ideal.
Let x, y ∈ pZp , this implies | x |p < 1, | y |p < 1, for x, y ∈ Qp . So we get
max{| x |p , | y |p } < 1.
Therefore we get | x − y |p ≤ max{| x |p , | −y |p } = max{| x |p , | y |p } < 1,
as | −1 |p = 1.
Thus x − y ∈ pZp .
Now for r ∈ Zp , | r |p ≤ 1 and x ∈ pZp , | x |p < 1.
We consider | rx |p = | r |p | x |p ≤ 1 | x |p < 1.
This implies rx ∈ pZp .
Hence pZp is an ideal of Zp .
Further, only we have to prove that every x ∈ Zp \ pZp is an unit of Zp .
If x ∈
/ pZp but x ∈ Zp =⇒ | x |p = 1.
Thus | 1/x |p = 1.
This implies x−1 ∈ Zp , i.e., x is invertible in Zp .
Hence we get for x ∈ Zp , there exists x−1 in Zp such that x.x−1 = 1.
Since x is arbitrary, therefore we get every element in Zp outside pZp is a
unit of Zp .
Thus by using the proposition of this section, we get that Zp is local ring and
pZp is its maximal ideal.

34
Proposition : The ring Zp of p-adic integers is a local ring whose maxi-
mal ideal is the principal ideal pZp = {x ∈ Zp : | x |p < 1}. Furthermore,
i) Q ∩ Zp = Z(p) := { ab ∈ Q : p ̸ | b}.
ii) The inclusion Z ,→ Zp has dense image. In particular, given x ∈ Zp and
n ≥ 1, there exist α ∈ x ∈ Z, 0 ≤ α ≤ pn − 1, such that | x − α |≤ p−n . The
integer α with these properties is unique.
iii) For any x ∈ Zp , there exists a Cauchy sequence αn converging to x, of
the following type:

• αn ∈ Z satisfies 0 ≤ αn ≤ pn − 1;

• for every n we have αn ≡ αn−1 (mod pn−1 ).

The sequence (αn ) with these properties is unique.

Proof : Let I be the valuation ideal of the valuation ring Zp ,
i.e., I = {x ∈ Zp : | x |p < 1}.
To see that the valuation ideal is indeed generated by p.
Let x ∈ I ⊂ Zp , as I is an ideal of Zp .
This implies | x |p < 1.
By definition of p-adic absolute value of Qp ,
| x |p = p−vp (x) < 1, i.e. 1 < pvp (x) =⇒ vp (x) > 0 =⇒ vp (x) ≥ 1.
1
Thus | x |p = pvp (x)
< p1 , (as pvp (x) > p for vp (x) ≥ 1).
x
Similarly, | x
|
p p
= p−vp ( p ) = p−(vp (x)−vp (p)) = p−vp (x)+vp (p)
= p−vP (x)+1 = p−vp (x) × p = | x |p ×p < 1
p
× p = 1.
x
Thus, | |
p p
< 1.
This implies xp ∈ Zp , i.e., x ∈ pZp .
Hence I ⊂ pZp , i.e., valuation ideal is contained in pZp .
We know that pZp is maximal ideal and pZp ̸= Zp .
Hence by definition of maximal ideal, I = pZp .

We now prove: i) Q ∩ Zp = Z(p) = { ab ∈ Q : p ̸ | b}.

Let x ∈ Q ∩ Zp ,
i.e., x ∈ Q and x ∈ Zp , by definition of p-adic, x = pn a/b , p ̸ | a, b.
We can write x = a′ /b, where a′ = pn a with p ̸ | b.

35
This implies x ∈ Z(p) . Hence Q ∩ Zp ⊂ Z(p) .
Now to prove that Z(p) ⊂ Q ∩ Zp .
Let x ∈ Z(p) , this implies x = a/b ∈ Q; p ̸ | b, by definition of Zp .
a |a|p |a|p
Now | |
b p
= |b|p
= p−0
, as p ̸ | b
|a|p
= 1
=| a |p . As a ∈ Z, vp (a) ≥ 0. Hence | a |p ≤ 1.
a
Hence | |
b p
≤ 1, this implies | x |p ≤ 1.
Hence x ∈ Zp .
Therefore Z(p) ⊂ Q ∩ Zp . Hence we can conclude that Q ∩ Zp = Z(p) .

ii) The inclusion Z ,→ Zp has dense image. In particular, given x ∈ Zp

and n ≥ 1, there exist α ∈ x ∈ Z, 0 ≤ α ≤ pn − 1, such that | x − α |≤ p−n .
The integer α with these properties is unique.
We choose x ∈ Zp and n ≥ 1.
Since Q is dense in Qp , one can certainly find a/b ∈ Q which is close enough
to x so that
a
| x − | ≤ p−n ≤ 1.
b

The point is to show that we can in fact choose an integer close enough to
x. Notice that for p ̸ | b as above, we will have
a a
| | ≤ max{| x |, | x − |} ≤ 1,
b b

which says that a/b ∈ Z(p) , i.e., p ̸ | b. Now recall that from the elementary
theory of congruences, if p ̸ | b there exits a unique integer b′ ∈ Z such that
bb′ ≡ 1(modpn ) and hence
a
| − ab′ | ≤ p−n ,
b

and of course ab′ ∈ Z. Finally, we need to check that we can find an inte-
ger between zero and pn − 1, but this is clear from the connection between
congruences modulo powers of p-adic absolute value: choosing α to be the
unique integer such that
0 ≤ α ≤ pn − 1

36
and
α ≡ ab′ (modpn ),

gives | x − α | ≤ p−n .

iii) It follows directly from (ii); just use (i), we define αn = α for a se-
quence of integers n = 1, 2, 3, etc.
Further note, αn−1 ≡ ab′ (mod pn−1 ) and αn ≡ ab′ (mod pn ).
This implies αn ≡ ab′ + λpn ≡ ab′ (mod pn−1 ) ≡ αn−1 (mod pn−1 ).

This proposition says several important things. Notice, that the sequence
(iii) is the “coherent sequences”. Now we begin from item (iii) in the above
proposition: for given x ∈ Zp , we can find a special kind of a Cauchy se-
quences converging to x and this sequences has the property of “coherent
sequences”. So we go to obtain a canonical way to represent the elements of
Qp as “power series in p”.

Let x ∈ Zp and (an ) be the Cauchy sequence described in the above item
(iii) of the proposition. We write them in the base p. Let a0 = b0 for some
0 ≤ b0 ≤ p − 1. Since a1 ≡ a0 ( mod p), a1 = b0 + b1 p for some 0 ≤ b1 ≤ p − 1.
Going up the sequences, we get
a0 = b 0 0 ≤ b0 ≤ p − 1
a1 = b 0 + b 1 p 0 ≤ b1 ≤ p − 1
2
a2 = b 0 + b 1 p + b 2 p 0 ≤ b2 ≤ p − 1
2 3
a3 = b 0 + b 1 p + b 2 p + b 3 p 0 ≤ b3 ≤ p − 1
and so on. Putting all of this together, we get an infinitely long expansion

b0 + b1 p + b2 p 2 + · · · + bn p n + · · · .

Now we see the series on right side converges to x.

Lemma : Given any x ∈ Zp , the series

b0 + b1 p + b2 p 2 + · · · + bn p n + · · ·

obtained as above, converges to x.

Proof : Note that a series converges to x if the sequence of its partial sums
converges to x. Since the partial sums of the above series are exactly the

37
(an ), which we already know converge to x, by the proof of part (ii) we have
done.

From the above lemma, the ring of the p-adic integers can also be stated
as: Definition: Let p < ∞ be a prime. P We define the p-adic integers, de-
noted Zp , to be the collection of all sums: ∞ n
n=0 bn p , with bn ∈ {0, ..., p − 1}.
In other words, Zp consists of all numbers with no negative powers of p in
its expansion.

Corollary : Every x ∈ Zp can be written in the form

x = b0 + b1 p + b2 p 2 + · · · + bn p n + · · ·

with 0 ≤ bi ≤ p − 1, and this representation is unique.

Now, we need to get all of Qp . Let x ∈ Qp and | x |p = pm ≥ 1. Let

z = pm x, then | z |p =| pm |p | x |p = p−m pm = 1. So we can write x = z/pm
where z ∈ Zp . If we express z as a power series in p, then divide by pm , we
just get a power series in p where some of the powers may be negative.

38
10 Solving Congruences Modulo pn
The “p-adic numbers” we have just constructed are closely related to the
problem of solving congruences modulo powers of p. We will look at some
examples of this.
Here we consider the system of congruence x2 ≡ 2 mod 7n and we try
to show that 7-adic expansion number

x = 3 + 1 × 7 + 2 × 49 + 6 × 343 + · · ·

satisfies x2 = 2 in Q7 .
For this, we first compute the solution for x2 ≡ 2 mod 7, which is 3 and 4.
Therefore, we get x ≡ 3 mod 7 and x ≡ 4 mod 7 ≡ −3 mod 7.
Now we compute the solution for x2 ≡ 2 mod 7j , for this we lifting our
solution from modulo 7 to modulo 72 to modulo 73 , until we get to the 7j
that is our target.
Let us see this,
for j = 2, i.e.,
x2 ≡ 2 mod 72
x2 ≡ 2 mod 49.

Now to find solution, note that their reductions module 7 must be solution
for j = 1.
Hence we set x = 3 + 7k or x = 4 + 7k, for some k ∈ Z.
Now solve for k,
consider
x = 3 + 7k
(3 + 7k)2 ≡ 2 mod 72
9 + 42k ≡ 2 mod 49
7 + 42k ≡ 0 mod 49
7(1 + 6k) ≡ 0 mod 49
1 + 6k ≡ 0 mod 7
6k ≡ −1 mod 7
6k ≡ 6 mod 7

39
 36k ≡ 36 mod 7
k ≡ 1 mod 7.

We get k = 1.
Therefore x = 3 + 7k gives x = 3 + 7 = 10.
Thus,
x ≡ 10 mod 49.

For j = 3, i.e.,
x2 ≡ 2 mod 73
x2 ≡ 2 mod 343.

Now to find solution, note that their reductions module 72 must be solution
for j = 2.
Hence we set x = 10 + 49k.
Solve for k,
(10 + 49k)2 ≡ 2 mod 343
100 + 980k ≡ 2 mod 343
98 + 980k ≡ 0 mod 343
49(2 + 20k) ≡ 0 mod 343
2 + 20k ≡ 0 mod 7
20k ≡ −2 mod 7
6k ≡ 5 mod 7
36k ≡ 30 mod 7
k ≡ 2 mod 7.

We get k = 2.
Therefore x = 10 + 49k gives x = 10 + 49 × 2 = 10 + 98 = 108.
Thus,
x ≡ 108 mod 343.

40
Similarly for j = 4,
we have x = 108 + 343k and we get that

x ≡ 2166 mod 2401.

We can easily see that this process can be continued indefinitely.

Hence we get,
for j = 1, x ≡ 3 mod 7,
for j = 2, x ≡ 10 = 3 + 1 × 7 mod 49,
for j = 3, x ≡ 108 = 3 + 1 × 7 + 2 × 49 mod 343,
for j = 4, x ≡ 2166 = 3 + 1 × 7 + 2 × 49 + 6 × 343 mod 2401
and so on.
We obtain a sequence
x0 , x1 , · · · , xj , · · · ,

satisfying the condition

xj ≡ xj−1 mod 7j .
Thus 7-adic expansion of the solution x = 3 + 1 × 7 + 2 × 49 + 6 × 343 + · · · =
3 + 1 × 7 + 2 × 72 + 6 × 73 + · · · .
Thus given value of x satisfies x2 = 2 in Q7 .
Now we compute a solution to x2 ≡ 2 mod 7j+1 ( the induction step).
Here we explain the induction step:
Assume that we have a solution x0 such that x2 ≡ 2 mod 7j .
Then we look for a lift of x0 mod 7j to x1 mod 7j+1 that satisfies x21 ≡ 2 mod
7j+1 .
Here the lifting equation is

x1 = x0 + 7 j y 0 ,

(1)
where y0 is unknown.
It ensures that x0 mod 7j+1 is a lift of x0 mod 7j .
And
x21 ≡ 2 mod 7j+1
(2)
is the equation we are trying to solve.

41
For this plugging equation (1) into equation (2) gives,

2 ≡ x21 mod 7j+1

2 ≡ (x0 + 7j y0 )2 mod 7j+1

2 ≡ x20 + 72j y02 + 2 x0 y0 7j mod 7j+1
2 ≡ x20 mod 7j+1 .

We get,
x20 ≡ 2 mod 7j+1 .

Here we can see that the variable y0 has completely disappeared from the
equation so we cannot solve for it.
Hence by induction hypothesis, equation x2 ≡ 2 mod 7n has solution for all
n.

We now note the interesting fact that the 7-adic field Q7 is strictly bigger
than the field Q. Clearly Q is contained in Q7 , as we know that Q is 7-adic
expansion of Q.
Now only remain to√ show that Q7 is strictly bigger than Q.
For x2 = 2, x = ± 2 ∈ / Q,
2
i.e., x = 2 has no solution in Q.
Note that it has solution in Q7 , as we have seen above.
Hence the field Q7 is strictly bigger than Q.

As we saw above that the x2 = 2 has solution in Q7 but now we will see
that x2 = 2 has no solution in Q5 . For proving this we use contradiction.
We consider the system of congruence

x2 ≡ 2 mod 5n .

Suppose it has solution in Q5 , i.e., that solution will have 5-adic expansion,
i.e., it is of the form a0 + a1 5 + a2 52 + · · · .

Therefore by our assumption x2 ≡ 2 mod 5 has solution a0 ,

i.e., a0 must satisfy a congruence module 5.

42
For finding a0 , the possible values of a0 is 0, 1, 2, 3 and 4.
For a0 = 0, 02 = 0 ̸≡ 2 mod 5 as 5 ̸ | 2;
a0 = 1, 12 = 1 ≡
̸ 2 mod 5 as 5 ̸ | 1;
a0 = 2, 22 = 4 ≢ 2 mod 5 as 5 ̸ | 2;
2
a0 = 3, 3 =9≡ ̸ 2 mod 5 as 5 ̸ | 7;
2
and a0 = 4, 4 = 16 ≡ ̸ 2 mod 5 as 5 ̸ | 14.

Observe that any value of a0 doesn’t satisfy congruence x2 ≡ 2 mod 5. It

means that a0 is not solution,
i.e., x2 ≡ 2 mod 5 has no solution of the form x ≡ a0 mod 5.
As a0 is not solution then 5-adic expansion a0 +a1 5+a2 52 +· · · is not solution
of the system of congruence x2 ≡ 2 mod 5n .
Our assumption is wrong. Hence x2 = 2 has no solution in Q5 .

43
11 Hensel’s Lemma
In this section we discuss Hensel’s lemma, concerning the solvability of p-adic
polynomials. Let’s first restrict our attention to finding roots in Zp . With
this theorem we test whether a polynomial has root in Zp or not.

Theorem : Let F (x) = a0 + a1 x + a2 x2 + a3 x3 + ... + an xn be a polyno-

mial whose coefficients are in Zp . Suppose that there exists a p-adic integers
α1 ∈ Zp such that
F (α1 ) ≡ 0 (mod pZp )
and
F ′ (α1 ) ̸≡ 0 (mod pZp ),
where F ′ (x) is the derivative of F (x). Then there exists a p-adic integer
α ∈ Zp such that α ≡ α1 (mod pZp ) and F (α) = 0.
Proof : We need to find a p-adic integers α such that α ≡ α1 (mod pZp ),
i.e., for all n, F (α) ≡ 0 (mod pn ) and F (α) = 0.
We need to construct a sequence of integers α1 , α2 , . . . , αn , . . . such that
for all n ≥ 1, we have

(1) F (αn ) ≡ 0 (mod pn );

(2) αn ≡ αn+1 (mod pn ).

By the assumption in the theorem, α1 exists.

To find α2 , we use condition (2),
therefore α2 = α1 + b1 p for some b1 ∈ Zp .
Putting this expression into the polynomial F (x) and expanding, we get
n
X
F (α2 ) = F (α1 + b1 p) = ai (α1 + b1 p)i
i=0
n
X
= (ai α1i + iai α1i−1 b1 p + terms divisible by p2 )
i=0
n
X n
X
≡ ai α1i +( iai α1i−1 )b1 p (mod p2 )
i=0 i=0
′
≡ F (α1 ) + F (α1 )b1 p (mod p2 ).

44
To show that one can find α2 , we have to show that one can find b1 so that

F (α1 ) + F ′ (α1 )b1 p ≡ 0 (mod p2 ).

Now, we know that F (α1 ) ≡ 0 (mod p), so that F (α1 ) = px for some x.
Then the equation becomes px + F ′ (α1 )b1 p ≡ 0 (mod p2 ),
after we divide by p we obtain x + F ′ (α1 )b1 ≡ 0 (mod p).
Note that F ′ (α1 ) is not divisible by p (as F ′ (α1 ) ̸≡ 0 (mod pZp )), and hence
its inverse exists in Zp (invertible), so that we can take

b1 ≡ −x(F ′ (α1 ))−1 (mod p)

such that 0 ≤ b1 ≤ p − 1.
For this choice of b1 , we set

α2 = α1 + b1 p,

which will have the stated properties. Continuing inductively we can find
α1 , α2 , · · · , αn−1 that satisfies conditions (1) and (2). We want to find αn
that satisfies conditions (1) and (2).
So, αn = αn−1 + bn−1 pn−1 with bn ∈ Zp .
We expand
F (αn ) = F (αn−1 + bn−1 pn−1 ))
≡ F (αn−1 ) + F ′ (αn−1 )bn−1 pn−1 (mod pn ).

We know that F (αn−1 ) ≡ 0 (mod pn−1 ), by the induction hypothesis.

This implies that F (αn−1 ) = ypn−1 for some y. Then the equation becomes

F (αn ) ≡ ypn−1 + F ′ (αn−1 )bn−1 pn−1 ≡ 0(mod pn ),

i.e.,
y + F ′ (αn−1 )bn−1 ≡ 0 (mod p).

Note that, F ′ (αn−1 ) is not divisible by p.

Therefore F ′ (αn−1 ) is invertible.
Thus we get bn−1 ≡ −y(F ′ (αn−1 ))−1 (mod p) such that 0 ≤ bn−1 ≤ p − 1 for

45
this choice of bn−1 , we get αn = αn−1 + bn−1 pn−1 and F (αn ) ≡ 0 (mod pn ).
i.e., F (αn ) ≡ 0 (mod pn ) and αn ≡ αn+1 (mod pn ).

Hence by the mathematical induction we can construct a sequence of in-

tegers α1 , α2 , · · · , αn−1 , αn , · · · such that for all n ≥ 1, we have

(1) F (αn ) ≡ 0 (mod pn );

(2) αn ≡ αn+1 (mod pn ).

Examples :

(1) Let f (x) = x3 + 2x2 + 2x + 4.

Let us look for 5-adic solution. Notice that α1 = 3 is the solution to
f (α1 ) ≡ 0 (mod 5),
i.e., f (3) = 27 + 18 + 6 + 9 = 55,
i.e., f (3) ≡ 0 (mod 5).
Also f ′ (α1 ) = 3x20 + 4x0 + 2 = 41 ̸≡ 0(mod 5).
So f (x) = 0 satisfies Hensel’s lemma.
This implies that there exists a unique 5-adic numbers that solves x3 +
2x2 + 2x + 4 = 0 and x ≡ 3 (mod 5).
Therefore x = 3 + b1 5 + b2 52 + b3 53 + · · · , where bi ∈ {0, 1, 2, 3, 4}.
Now we can calculate bi ’s as follows:
Consider
b1 ≡ −x(f ′ (α1 ))−1 (mod 5),
where x = f (α5 1 ) ∈ Z.
Therefore x = 55 5
= 11 as we have f (α1 ) = f (3) = 55.
Also we have f ′ (α1 ) = 41.
Hence we get

b1 ≡ −11(41)−1 (mod 5) ≡ −1(1) (mod 5) = 4.

Similarly, b2 = 4, b3 = 4 and so on.

Hence we get x = 3 + 4 × 5 + 4 × 52 + 4 × 53 + · · · is a root of f (x) in
Z5 .

(2) Let f (x) = x2 + 7.

Then f (1) = 1 − 7 = −6 ≡ 0 (mod 3)

46
 and f ′ (1) = 2(1) = 2 ̸≡ 0 (mod 3).
So there exists a unique 3-adic integer x such that x2 + 7 = 0 and
x ≡ 1 (mod 3).
Therefore x = 1 + b1 3 + b2 32 + b3 33 + · · · , where bi ∈ {0, 1, 2}.
Here we can also find bi ’s as we have found in above example.

12 Hensel’s Lemma, Second Form

Theorem : Let F (x) ∈ Zp [X] be a polynomial with coefficients in Zp , and
assume that there exist polynomials g1 (X) and h1 (X) in Zp [X] such that

i) g1 (X) is monic (i.e., the coefficient of the highest degree term is one),

ii) g1 (X) and h1 (X) are relatively prime modulo p, and

iii) f (X) ≡ g1 (X)h1 (X) (modp) (understood coefficient-by-coefficient).

Then there exists polynomials g(X), h(X) ∈ Zp [X] such that

i) g(X) is monic,

ii) g(X) ≡ g1 (X) (mod p) and h(X) ≡ h1 (X) (mod p) and

iii) f (X) = g(X)h(X).

Proof : Let d be the degree of f (X) and m be the degree of g1 (X).

Assume that deg(h1 ) ≤ d−m (it could be less, because the top coefficient of f
could be divisible by p). We want to construct two sequences of polynomials
gn (X) and hn (X) such that

i) each gn is monic and of degree m,

ii) gn+1 ≡ gn (mod pn ) and hn+1 ≡ hn (mod pn ),

iii) f (X) ≡ gn (X)hn (X) ( mod pn ) (we take the congruences coefficient-by-
coefficient).

If we can find such sequences, we are already done, since going to the limit
gives the desired polynomials g(X) and h(X). (In other words, the coef-
ficients of g(X) will be the limits of the corresponding coefficients of the
gn (X).)

47
 We already have g1 (X) and h1 (X). Let us describe how to get g2 (X) and
h2 (X). Since the g’s are to be congruent, we must have

g2 (X) = g1 (X) + pr1 (X),

for some polynomial r1 (X) ∈ Zp [X] (since g2 (X) ≡ g1 (X) (mod p), i.e., p
divides g2 (X) − g1 (X)).
Similarly, we must have

h2 (X) = h1 (X) + ps1 (X).

To show that g2 and h2 exist, we simply have to show that it is possible to

find r1 and s1 such that the desired conditions are satisfied. For that, we
need to solve the equation

f (X) ≡ g2 (X)h2 (X) (mod p2 )

which expand to

f (X) ≡ (g1 (X) + pr1 (X))(h1 (X) + ps1 (X)) (mod p2 ).

By multiplying, we get

f (X) ≡ g1 (X)h1 (X) + pr1 (X)h1 (X) + ps1 (X)g1 (X) + p2 r1 (X)s1 (X) (mod p2 )

f (X) ≡ g1 (X)h1 (X) + pr1 (X)h1 (X) + ps1 (X)g1 (X) (mod p2 ).

Given that,
f (X) ≡ g1 (X)h1 (X) (mod p),
so that we have
f (X) − g1 (X)h1 (X) = pk1 (X),
for some k1 (X) ∈ Zp [x]. Rearranging, we get

pk1 (x) ≡ pr1 (X)h1 (X) + ps1 (X)g1 (X) (mod p2 ).

48
Dividing both sides by p, we get

k1 (X) ≡ r1 (X)h1 (X) + s1 (X)g1 (X) (mod p).

This is the equation we need to solve to determine r1 and s1 .

To do so, we use condition ii) given to us, i.e., g1 and h1 are relatively
prime modulo p.
This means that there exists a(X), b(X) ∈ Zp [X] such that

a(X)g1 (X) + b(X)h1 (X) ≡ 1 (mod p).

Now consider two polynomials

r̃1 (X) = b(X)k1 (X) and s̃1 (X) = a(X)k1 (X).

These will almost do the trick:

They clearly will make all the congruence condition true. The only problem is
that we have no control over the degree of r̃1 (X) and hence cannot guarantee
that g1 (X) + pr̃1 (X) is monic.
To remedy that, only a slight change is needed. We already know that

r̃1 (X)h1 (X) + s̃1 (X)g1 (X) ≡ k1 (X) (mod p).

Now divide r̃1 (X) by g1 (X), and let r1 (X) be the remainder

r̃1 (X) = g1 (X)q(X) + r1 (X).

We know that deg r1 (X) < deg g1 (X). But now, if we set

s1 (X) = s̃1 (X) + h1 (X)q(X),

it all works out:

r1 (X)h1 (X) + s1 (X)g1 (X) ≡
≡ (r̃1 (X) − g1 (X)q(X))h1 (X) + (s̃1 (X) + h1 (X)q(X))g1 (X)
≡ r̃1 (X)h1 (X) − g1 (X)h1 (X)q(X) + s̃1 (X)g1 (X) + g1 (X)h1 (X)q(X)

49
≡ r̃1 (X)h1 (X) + s̃1 (X)g1 (X)
≡ k1 (X) (mod p)
so that our congruence conditions are satisfied, and the fact that the degree
of r1 (X) is smaller than the degree of g1 (X) is enough to guarantee that
g1 (X) + pr1 (X) is monic, and we are done.

This shows that g2 and h2 exist. Since they are congruent to g1 and h1
modulo p, they are also relatively prime modulo p, so that there will be no
difficulty in going on to the next step.
Now we repeat the argument changing the indices and exponents to find
g3 and h3 . Inductively, it can be proved that this produces the sequence
whose convergence proves the theorem.

50
References
[1] Fernando Q. Gouvêa, p-adic Numbers: An Introduction, 2nd edition,
New York: Springer-Verlag, 1997.
[2] William Stein, Elementary Number Theory: Primes, Congruences,
and Secrets. Available at https://www.wstein.org.
[3] Brian Courthoute, Pablo Guzman, Antoine Ronk, The p-adic integers.
Available at https://www.google.com.
[4] Logan Quick, p-adic Absolute Values. Available at https://www.
math.uchicago.edu.
[5] Andrew Baker, An Introduction to p-adic Numbers and p-adic Analy-
sis, (2011) Available at https://scholar.google.co.in.
[6] Alexa Pomerantz, An Introduction to the p-adic Numbers. Available
at http://www.math.uchicago.edu.
[7] Theodor Christian Herwig, The p-adic completion of Q and Hensel’s
Lemma. Available at http://www.math.uchicago.edu.
[8] Neal Koblitz, p-adic Numbers, p-adic Analysis, and Zeta-Functions,
2nd ed., New York: Springer-Verlag, 1984. Available at https://www.
math.mcgill.ca.
[9] Catherine Crompton, “Some Geometry of the p-adic Rationals,” Rose-
Hulman Undergraduate Mathematics Journal: Vol. 8 : Iss. 1 , Article
2, (2007). Available at https://scholar.rose-hulman.edu/rhumj/
vol8/iss1/2.
[10] Charles I. Harrington, “An Introduction to the p-adic Numbers,”
Honours Theses. 992, (2011). Available at https://digitalworks.
union.edu/theses/992.
[11] Article in p-adic Numbers Ultrametric Analysis and Applica-
tions, April 2009. Available at https://www.researchgate.net/
publication/24374483.
[12] Eric W. Weisstein, “p-adic Number,” From MathWorld–A Wol-
fram Web Resource. Available at https://mathworld.wolfram.com/
p-adicNumber.html.

51