Math 3527 (Number Theory 1)

Lecture #29

Polynomial Congruences:
Polynomial Congruences Modulo m
Polynomial Congruences Modulo p n and Hensel’s Lemma
This material represents §5.1 from the course notes.
Overview

The goal of this last segment of the course is to discuss quadratic

residues (which are simply squares modulo m) and the law of
quadratic reciprocity, which is a stunning and unexpected relation
involving quadratic residues modulo primes.
We begin with some general tools for solving polynomial
congruences modulo prime powers, which essentially reduce
matters to studying congruences modulo primes.
Then we study the quadratic residues (and quadratic
nonresidues) modulo p, which leads to the Legendre symbol, a
tool that provides a convenient way of determining when a
residue class a modulo p is a square.
We then discuss quadratic reciprocity and some of its
applications.
Polynomial Congruences, I

In an earlier chapter, we analyzed the problem of solving linear

congruences of the form ax ≡ b (mod m). We now study the
solutions of congruences of higher degree.
As a first observation, we note that the Chinese Remainder
Theorem reduces the problem of solving any polynomial
congruence q(x) ≡ 0 (mod m) to solving the individual
congruences q(x) ≡ 0 (mod p d ), where the p d are the
prime-power divisors of m.
Polynomial Congruences, II

Example: Solve the equation x 3 + x + 2 ≡ 0 (mod 36).

Polynomial Congruences, II

Example: Solve the equation x 3 + x + 2 ≡ 0 (mod 36).

By the Chinese remainder theorem, it suffices to solve the two
separate equations x 3 + x + 2 ≡ 0 (mod 4) and
x 3 + x + 2 ≡ 0 (mod 9).
We can just test all possible residues to see that the only
solutions are x ≡ 2 (mod 4) and x ≡ 8 (mod 9).
Therefore, by the Chinese remainder theorem, there is a
unique solution; namely, the solution to those simultaneous
congruences, which is x ≡ 26 (mod 36).
Polynomial Congruences, III

Example: Solve the equation x 2 ≡ 0 (mod 12).

Polynomial Congruences, III

Example: Solve the equation x 2 ≡ 0 (mod 12).

By the Chinese remainder theorem, it suffices to solve the two
separate equations x 2 ≡ 0 (mod 4) and x 2 ≡ 0 (mod 3), and
then put the results back together.
The first equation visibly has the solutions x ≡ 0, 2 (mod 4)
while the second equation has the solution x ≡ 0 (mod 3).
Then applying the Chinese remainder theorem to the 2
possible pairs of congruences x ≡ 0 (mod 4), x ≡ 0 (mod 3),
and x ≡ 0 (mod 4), x ≡ 2 (mod 3), yields the solutions
x ≡ 0, 6 (mod 12) to the original equation.
Polynomial Congruences, IV

Example: Solve the equation x 2 ≡ 1 (mod 30).

Polynomial Congruences, IV

Example: Solve the equation x 2 ≡ 1 (mod 30).

By the Chinese remainder theorem, it suffices to solve the
three separate equations x 2 ≡ 1 (mod 2), x 2 ≡ 1 (mod 3),
x 2 ≡ 1 (mod 5).
We can just test all possible residues to see that the solutions
are x ≡ 1 (mod 2), x ≡ 1, 2 (mod 3), and x ≡ 1, 4 (mod 5).
Therefore, by applying the Chinese remainder theorem to all
1 · 2 · 2 = 4 ways to pick a solution from each congruence, we
see that there are 4 solutions modulo 30, and they are
x ≡ 1, 11, 19, 29 (mod 30).
Polynomial Congruences, V
We are therefore reduced to solving a polynomial congruence of
the form q(x) ≡ 0 (mod p d ).
Observe that any solution modulo p d “descends” to a solution
modulo p, simply by considering it modulo p.
For example, any solution to x 3 + x + 3 ≡ 0 (mod 25), such
as x = 6, is also a solution to x 3 + x + 3 ≡ 0 (mod 5).
Our basic idea is that this procedure can also be run in
reverse, by first finding all the solutions modulo p and then
using them to compute the solutions modulo p d .
More explicitly, if we first solve the equation modulo p, we
can then try to “lift” each of these solutions to get all of the
solutions modulo p 2 , then “lift” these to obtain all solutions
modulo p 3 , and so forth, until we have obtained a full list of
solutions modulo p d .
Polynomial Congruences, VI
Example: Solve the congruence x 3 + x + 3 ≡ 0 (mod 25).
Polynomial Congruences, VI
Example: Solve the congruence x 3 + x + 3 ≡ 0 (mod 25).
Since 25 = 52 , we first solve the congruence modulo 5.
If q(x) = x 3 + x + 3, we can just try all residues to see the
only solution is x ≡ 1 (mod 5).
Now we “lift” to find the solutions to the original congruence,
as follows: if x 3 + x + 3 ≡ 0 (mod 25) then we must have
x ≡ 1 (mod 5).
Now write x = 1 + 5a: plugging in yields
(1 + 5a)3 + (1 + 5a) + 3 ≡ 0 (mod 25), which, upon
expanding and reducing, simplifies to 5 + 20a ≡ 0 (mod 25).
Cancelling the factor of 5 yields 4a ≡ 4 (mod 5), which has
the single solution a ≡ 1 (mod 5).
This yields the single solution x ≡ 6 (mod 25) to our original
congruence.
Polynomial Congruences, VII

Example: Solve the congruence x 3 + 4x ≡ 4 (mod 343).

Polynomial Congruences, VII

Example: Solve the congruence x 3 + 4x ≡ 4 (mod 343).

Since 343 = 73 , we first solve the congruence modulo 7, then
modulo 72 , and then finally modulo 73 .
By trying all the residue classes, we see that x 3 + 4x ≡ 4
(mod 7) has the single solution x ≡ 3 (mod 7).
Next we lift to find the solutions modulo 72 : any solution
must be of the form x = 3 + 7a for some a.
Plugging in yields (3 + 7a)3 + 4(3 + 7a) ≡ 4 (mod 72 ), which
eventually simplifies to 21a ≡ 14 (mod 72 ).
Cancelling the factor of 7 yields 3a ≡ 2 (mod 7), which has
the single solution a ≡ 3 (mod 7).
This tells us that x ≡ 24 (mod 49).
Polynomial Congruences, VIII

Example (continued):
Now that we know that we must have x ≡ 24 (mod 49), we
can lift to find the solutions modulo 73 in the same way.
Explicitly, any solution must be of the form x = 24 + 49b for
some b.
Plugging in yields (24 + 72 b)3 + 4(24 + 72 b) ≡ 4 (mod 73 ),
which eventually simplifies to 147b ≡ 147 (mod 73 ).
Cancelling the factor of 72 yields 3b ≡ 3 (mod 7), which has
the single solution b ≡ 1 (mod 7).
Hence we obtain the unique solution
x ≡ 24 + 49b ≡ 73 (mod 73 ).
Polynomial Congruences, IX

Example: Solve the congruence x 3 + 4x ≡ 12 (mod 73 ).

Polynomial Congruences, IX

Example: Solve the congruence x 3 + 4x ≡ 12 (mod 73 ).

We first solve the congruence modulo 7. By trying all the
residue classes, we see that x 3 + 4x ≡ 5 (mod 7) has two
solutions, x ≡ 1 (mod 7) and x ≡ 5 (mod 7).
Next we lift to find the solutions modulo 72 : any solution
must be of the form x = 1 + 7k or x = 5 + 7k for some k.
If x = 1 + 7k, then we get (1 + 7k)3 + 4(1 + 7k) ≡ 12 (mod
72 ), which simplifies to 0 ≡ 7 (mod 72 ). This is contradictory
so there are no solutions in this case.
If x = 5 + 7k, then we get (5 + 7k)3 + 4(5 + 7k) ≡ 12 (mod
72 ), which simplifies to 14k ≡ 14 (mod 72 ). Solving this
linear congruence produces k ≡ 1 (mod 7), so we obtain
x ≡ 12 (mod 49).
Polynomial Congruences, X

Example (continued):
Now we lift to find the solutions modulo 73 : from the previous
slide, any solution must be of the form x = 12 + 49k.
In the same way as before, plugging in yields
(12 + 72 k)3 + 4(12 + 72 k) ≡ 4 (mod 73 ), which after
expanding and reducing, simplifies to 98k ≡ 294 (mod 73 ).
Solving in the same way as before yields k ≡ 5 (mod 7),
whence x ≡ 12 + 49k ≡ 257 (mod 73 ).
Hence, there is a unique solution: x ≡ 257 (mod 73 ).
Polynomial Congruences, XI
Example: Solve the congruence x 2 ≡ 9 (mod 16).
Polynomial Congruences, XI
Example: Solve the congruence x 2 ≡ 9 (mod 16).
Since 16 = 24 , we find the solutions mod 2, then work upward.
It is easy to see that there is a unique solution to x 2 ≡ 9
(mod 2), namely, x ≡ 1 (mod 2).
Next we lift to find the solutions modulo 22 : any solution
must be of the form x = 1 + 2k, so we get (1 + 2k)2 ≡ 9
(mod 22 ), which simplifies to 1 ≡ 9 (mod 22 ). This is always
true, so we get two possible solutions, x ≡ 1, 3 (mod 4).
If x = 1 + 4k, then we get (1 + 4k)2 ≡ 9 (mod 23 ), which
simplifies to 1 ≡ 9 (mod 23 ), which is again always true.
If x = 3 + 4k, then we get (3 + 4k)2 ≡ 9 (mod 23 ), which
simplifies to 9 ≡ 9 (mod 23 ), which is also always true.
Thus we get the four solutions x ≡ 1, 3, 5, 7 (mod 23 ).
Polynomial Congruences, XII
Example (continued):
Finally, we must lift each solution x ≡ 1, 3, 5, 7 (mod 23 ) to
the modulus 24 .
If x = 1 + 8k then we get (1 + 8k)2 ≡ 9 (mod 24 ), which
simplifies to 1 ≡ 9 (mod 24 ), which is contradictory.
If x = 3 + 8k then we get (3 + 8k)2 ≡ 9 (mod 24 ), which
simplifies to 9 ≡ 9 (mod 24 ), which is always true, so we get
two solutions x ≡ 3, 11 (mod 24 ).
If x = 5 + 8k then we get (5 + 8k)2 ≡ 9 (mod 24 ), which
simplifies to 25 ≡ 9 (mod 24 ), which is always true, so we get
two solutions x ≡ 5, 13 (mod 24 ).
If x = 7 + 8k then we get (7 + 8k)2 ≡ 9 (mod 24 ), which
simplifies to 49 ≡ 9 (mod 24 ), which is contradictory.
Thus, we get four solutions in total: x ≡ 3, 5, 11, 13 (mod 24 ).
Polynomial Congruences, XIII: Lucky!

The general procedure will work the same way for any prime power
modulus p n :
We first solve the congruence modulo p. For each solution we
obtain, we then try to lift it to a solution mod p 2 , then lift
each of those to a solution mod p 3 , and so forth, until we get
the full list of solutions mod p n .
In the last few examples we just worked through, we saw a
variety of different behaviors.
Sometimes, when we lift a solution, we obtain exactly one
lifted solution. Other times, the lifting might fail, or it might
yield more than one possible lifted solution.
We would like to understand what determines when each of
these behaviors will occur.
Hensel’s Lemma, I

Rather than building the motivation, we will simply state the result:
Theorem (Hensel’s Lemma)
Suppose q(x) is a polynomial with integer coefficients. If q(a) ≡ 0
(mod p d ) and q 0 (a) 6≡ 0 (mod p), then there is a unique k
(modulo p) such that q(a + kp d ) ≡ 0 (mod q d+1 ). Explicitly, if u
q(a)
is the inverse of q 0 (a) modulo p, then k = −u · d .
p

This result (and a number of variations) is traditionally called

Hensel’s lemma, although for us it is really more of a theorem since
the proof is fairly technical. (The full proof is in the notes, but it is
just a formalized version of the procedure we were using earlier.)
Hensel’s Lemma, II

Example: Show that there is a unique solution to the congruence

x 3 − 2x + 7 ≡ 0 (mod 32020 ).
Hensel’s Lemma, II

Example: Show that there is a unique solution to the congruence

x 3 − 2x + 7 ≡ 0 (mod 32020 ).
The idea is to use Hensel’s lemma to show that the lifting will
always yield a unique solution starting from the bottom level.
First, we solve the congruence modulo 3: testing all 3 possible
residues shows that the only solution is x ≡ 1 (mod 3).
Now we just compute the derivative: if q(x) = x 3 − 2x + 7,
then q 0 (x) = 3x 2 − 2 ≡ 1 (mod 3), no matter what x is.
Therefore, Hensel’s lemma guarantees that we will always
have a unique solution to this congruence modulo 3d for any
d ≥ 1. In particular, the solution is unique modulo 32020 .
Hensel’s Lemma, III
Example (continued): Solutions of x 3 − 2x + 7 ≡ 0 (mod 3d ).
Hensel’s Lemma, III
Example (continued): Solutions of x 3 − 2x + 7 ≡ 0 (mod 3d ).
We can even calculate the various lifts using the formula given
in Hensel’s lemma. (Our direct technique will yield the same
result, since ultimately it is how Hensel’s lemma is proven.)
For example, mod 32 , since q 0 (a) ≡ 1 (mod 3) has inverse
u ≡ 1 (mod 3), we will obtain the solution x = 1 + 3k where
q(a) 6
k = −u · d = −1 · = −2: thus, x ≡ −5 ≡ 4 (mod 9),
p 3
which indeed works.
Lifting again yields x = 4 + 9k where
q(a) 63
k = −u · d = −1 · = −7, yielding x ≡ 4 + 9k ≡ 22
p 9
(mod 27).
We can continue in this way and compute the lifts as high as
we desire.
Summary

We discussed how to solve polynomial congruences modulo m and

modulo prime powers. We discussed how to use Hensel’s lemma to
calculate solutions to congruences modulo p d explicitly in many
cases.

Next lecture: Quadratic Residues and Legendre Symbols