Interactive · 3 questions
Note: This explanation aims not only to prove common irreducibility criteria but to also demonstrate common methods used in the proofs which appear elsewhere.
There are several common irreducibility criteria for integer polynomials. But before, we can get into those, we need some background knowledge and setup. Define as the set of polynomials with integer coefficients.
What does it mean for a non-constant polynomial to be irreducible in ?
So is irreducible because all we can factor out is a , which is constant. But is reducible because it factors as .
Now, I'll state some well-known and easy results that we'll need before proceeding onwards.
Gauss: A polynomial with integer coefficients which is irreducible in if and only if it is irreducible in . (This means if we can factor it into polynomials with rational coefficients, we can factor it into polynomials with integer coefficients).
Obvious: An irreducible polynomial has no integer roots.
Now we may prove
Eisenstein's Criterion: If , and some prime divides but not , and does not divide , then is irreducible.
How can we even start this? (Almost all irreducibility questions start with this step)
Suppose that (this is the first step of almost any irreducibility proof) for with . I will write
.
Then taking this equation mod , we know that mod . Then some number theory tells us that mod for some with .
This technique is called reduction mod p . Basically, it takes advantage of the fact that polynomials can be uniquely factored mod , so if we take a polynomial mod a prime which its coefficients are divisible by, they all vanish!
But now suppose the constant term of is and the constant term of is . Since is equivalent to , for , its constant term vanishes, meaning we need . Similarly, .
But now take the constant term of : It equals , and since we have . But take a look at the original criterion! We have stated , so we have a contradiction!
This proof is instructive so we will summarize it: We have done two steps.
-
Write as to obtain a contradiction.
-
Reduce mod a prime.
For an exercise, prove that is irreducible for a prime .
Next, we will take a look at a strange criterion:
Cohn's Criterion, simplified: If a prime number may be expressed in base as with , then is irreducibile.
This is a very intuitive result: Basically it says that if you turn a prime into a polynomial, it remains "prime"!
HOw can we start the proof?
As before, write . But this time, we will examine and bound the roots of .
Note that in base , all digits are less than . So we have for all .
Now take an arbitrary root of . Since , we have
.
Since we have . Thus
.
Meanwhile, by the extended triangle inequality, we have
This last quantity equals
.
Thus, we have .
If , then we have , which is a contradiction! (You can prove this via the geometric series formula on the RHS and then cross-multiplying, but I'll leave it out).
Thus we know that all roots of have magnitude less than .
Let the roots be . Note that so one of is since is prime. WLOG let it be , and suppose it equals . Then
for some positive integer and some roots which are also roots of .
But and since , we have . So then , a contradiction. Thus is irreducible as desired.
This emphasizes another technique: Bound the roots of the polynomial
In case you are not familiar with it, we have one final common irreducibility criteria using root-bounding:
Ostrowski: Let . If is prime and , is irreducible.
Proof: I'll just go over it very quickly since this is quite repetitive.
Write , and since is prime, the constant term of one of equals . WLOG suppose the constant term of is 1. Take any root of .
Then using the same technique as earlier, we have that .
If , then we get
, a contradiction! Thus all roots have .
But then the constant term of is the product of many terms of the form , so its magnitude is greater than , a contradiction.
Conclusion:
Today we proved three irreducibility criterion: Eisenstein, Ostrowski, and Cohn, and skimmed over Gauss's Lemma. We also noticed some common techniques for proving irreducibility: Reduction mod , root-bounding, and contradiction.
If you want some more irreducibility criterion, I'll give you some names:
Perron's Criterion: If , and , with , then is irreducible.
STronger Cohn: Same statement as Cohn, except it covers the case.
Cyclotomic Irreducibility: The th cyclotomic polynomial is irreducible.
Irreducibility is a wide topic, and there are many other criterion. Have fun finding them!