Some notes on Rings topics diverging from book
1. Chinese remainder theorem
Theorem. If m, n are relatively prime numbers then for any two remainders a ∈ Zm , b ∈ Zn ,
there exists a unique remainder r ∈ Zmn with r ≡ a mod m, r ≡ n mod b.
This theorem is named after the mathematician Sunzi, who lived in 3d-5th century China
and left an early treatise on number theory, where this theorem was first proved. A more
ring-theoretic statement of the result is as follows.
Theorem. If m, n are relatively prime then the function
ϕ : Zmn → Zm × Zn
taking a to (a mod m, a mod n) is an isomorphism of rings.
Proof. It is straightforward to see that ϕ is a homomorphism; the slightly more difficult
part is to show ϕ is a bijection. Since ϕ is a function between two finite sets of equal size,
it is a bijection if and only if it an injection. Since ϕ is a homomorphism, proving that it is
injective is equivalent to proving that the kernel Ker(ϕ) = {0} ⊂ Zmn . Why is nothing but
0 in the kernel? Well, if a ∈ Ker(ϕ) then a mod n = 0 and a mod m = 0, but this means
that mn | a, so a = 0 ∈ Zmn .
2. Arithmetic modulo an element
Suppose R is a commutative ring with unit, and a ∈ R is an element. Then we define
(a) ⊂ R, the principal ideal generated by a, also known as the ring of multiples of a, to be
the set (a) := a · R, where a · R is defined to be {ar | r ∈ R}.
Now we define R/(a) to be the set of additive cosets of (a) in R, equivalently the quotient
of R by the equivalence relation ≡a , where x ≡a y if y − x = ar for some r ∈ R. We know
from group theory that this is an equivalence relation, whose classes are additive (a)-cosets.
It is not hard to see that both addition and multiplication modulo a are well-defined.
It is convenient to think of modular arithmetic in terms of a fixed collection of residues,
called a “set of representatives”. A subset S ⊂ R is a set of representatives for the equivalence
relation ≡a if every r ∈ R is equivalent to a unique s ∈ S; equivalently if every class
C ∈ R/(a) (viewed as a coset C ⊂ R) contains exactly one element of S. The advantage
of sets of representatives is that they take out the ambiguity of the notation [r] (which is
the same as [r0 ] so long as r0 ≡a r). Instead, each class can be “labeled” [s] for s ∈ S, now
without ambiguity because there is only one choice of s in every equivalence class.
An example of a set of representatives is the set of residues S = {0, 1, . . . , n − 1} modulo
a number n. To see this is a set of representatives, just check that every number a ∈ Z
2
is equivalent modulo n to exactly one element of S. Importantly, a set of representatives
S ⊂ R will not necessarily be closed under addition or multiplication: however we can always
convert the result of an addition or multiplication of cosets [s] + [t] (for s, t ∈ S) to some
element [s] + [t] = [u] for u ∈ S the unique element of S that differs from s + t by a multiple
of a.
Another example we considered involves Gaussian numbers. Recall the set G of Gaussian
integers {a + bi | a, b ∈ Z} forms a (unital, commutative) ring under +, ·. We can view a
positive integer n ∈ Z as a Gaussian integer (just take nG = n · 1G = n + 0i). What is (n)?
Well, it nG = {ns + nti | s, t ∈ Z}.
Lemma. The set of residues Remn = {a + bi | 0 ≤ a, b ≤ n − 1} ⊂ G is a collection of
representatives for G/ ≡n .
In other words, every Gaussian number is equivalent (up to a multiple of n) to exactly
one sum of two remainders modulo n. This should seem reasonable to you. To prove it
rigorously, notice that if x, y ∈ Z are two integers then we can write x = q · n + a, y = r · n + b
for q, r ∈ Z integers and a, b ∈ {0, . . . , n − 1} remainders. In that case, x + yi ≡n a + bi,
since x + yi − (a + bi) = n · (q + ri). On the other hand if z = a + bi ∈ Remn and
z 0 = a0 + b0 i ∈ Remn are two Gaussian remainders then a, b, a0 , b0 ∈ {0, . . . , n − 1}, which
implies |a − a0 | ≤ n − 1. so the only way one could have n | a − a0 is if a = a0 . Similarly,
the only way n could divide b − b0 is if b = b0 . Thus every coset has an element of Remn and
no two elements of Remn are in the same coset, and so Remn is a set of representatives. Of
course it is not closed under addition and multiplication: to add and multiply while working
with representatives, one might need to re-convert after applying an operation. For example,
taking n = 7 we have (2 + 3i), (5 + 3i) ∈ Rem7 . But (2 + 3i) · (5 + 3i) = 1 + 21i is not in
Rem7 . To perform a calculation in G/(7) in terms of these representatives we should write
[2 + 3i] · [5 + 3i] = [1 + 21i] = [1], as 21 is divisible by 7. Note that what we have just proved
is that in the ring G/(7), the inverse of [2 + 3i] is [5 + 3i].
3. Arithmetic modulo a polynomial
Theorem 1. Let F be any field, and R = F [x] the ring of polynomials (in one indeterminate)
over F . Let f (x) = an xn + · · · + a1 x + a0 ∈ F [x] be a (nonzero, non-constant) polynomial
of degree n. Let P≤n−1 be the set of polynomials of degree ≤ n − 1 (this includes 0). Then
P≤n−1 is a set of representatives for the relation ≡f of equivalence modulo f .
We will prove this theorem in class. Note that for every polynomial of degree n, the same
set P≤n−1 works as a set of representatives! In addition, this set is particularly nice as it
is closed under addition (no pun intended). This means that if α(x), β(x) ∈ P≤n−1 are two
“residue” polynomials representing the classes [α], [β] modulo f , then α + β is once again a
“residue” polynomial representing the class [α] + [β] ∈ R/(a).
3
However, it is not closed under multiplication, and the product of two polynomials in
P≤n−1 might have to be re-converted (by subtracting a multiple of f (x)) to send it back to
P≤n−1 . Here is an example: let f1 (x) = x2 and f2 (x) = 12 x2 − 1, both in Q[x]. Then x ∈ P≤1 .
If we square it, x2 no longer has degree ≤ 1: in R/(f1 ), we have [x]2 = [x2 − f1 (x)] = [0] (zero
divisor alert! Note that 0 has degree that is “less than anything else”). Similarly, in R/(f2 ),
we have [x]2 = [x2 ] but x2 is not one of our representatives. So subtracting a multiple of f ,
we get [x]2 = [x2 − 2 · f2 (x)] = [2] (it is a scalar, hence has degree 0, hence indeed is in our
set of representatives P≤1 .)
The upshot of this discussion is that arithmetic modulo a polynomial f (x) ∈ F [x] works
exactly like arithmetic modulo a number, only better. Namely, if f (x) has degree n ≥ 1,
then one can define a ring of residues modulo f (x) by Rf := (P≤n−1 , +, ∗mod f ), where
r1 (x) ∗mod f r2 (x) = r1 (x) · r2 (x) mod f. Then Rf is isomorphic to F [x]/(f (x)) via r 7→ [r].
When faced with a problem about F [x]/(f (x)), you can always convert it to a problem about
the isomorphic ring of residues Rf via this relabeling (though when possible, try to be clear
whether you are working with residues or equivalence classes and consistent about which
ring you are using). This is very similar to how any problem about Z/ ∼n can be converted
to a problem about the ring of residues Zn = ({0, . . . , n − 1}, +mod n , ·mod n ).
