特征p>0域上不可分多项式等价性引理的归纳法证明求证

Let's work through the inductive proof of this key lemma step by step, starting with recapping the core definitions and the lemma statement from the undergraduate elliptic curves course:

Relevant Definition #

Let (g(x) \in K[x]) be a polynomial. We call (g(x)) separable if its derivative (g'(x)) is not identically zero; otherwise, (g(x)) is inseparable.

Lemma 2 Statement #

Let (K) be a field of characteristic (p>0). A polynomial (g(x) \in K[x]) is inseparable if and only if there exists a separable polynomial (t(x) \in K[x]) and a positive integer (k) such that (g(x) = t(x^{pk})).

We'll split the proof into sufficiency (the "if" direction) and necessity (the "only if" direction), using induction for the latter.

Sufficiency ((\Leftarrow)): If (g(x) = t(x^{pk})) with (t(x)) separable, then (g(x)) is inseparable #

Calculate the derivative of (g(x)):
[
g'(x) = t'(x^{pk}) \cdot \frac{d}{dx}\left(x^{pk}\right) = t'(x^{pk}) \cdot p^k x^{pk - 1}
]
Since (K) has characteristic (p>0), (p^k \equiv 0 \mod p), so the entire derivative simplifies to the zero polynomial. By definition, (g(x)) is inseparable. This direction is straightforward.

Necessity ((\Rightarrow)): If (g(x)) is inseparable, then (g(x) = t(x^{pk})) for some separable (t(x)) and positive integer (k) #

We use induction on the degree (n = \deg g(x)):

Base Case: (n = 1)

A degree-1 polynomial (g(x) = ax + b) (with (a \neq 0)) has derivative (g'(x) = a \neq 0), so it's always separable. There are no inseparable degree-1 polynomials, so the statement holds vacuously here.

Inductive Step: Assume the statement holds for all inseparable polynomials of degree less than (n). Now consider an inseparable polynomial (g(x)) of degree (n).

Since (g(x)) is inseparable, (g'(x) = 0). Let's break down what this means for (g(x))'s coefficients:
Write (g(x) = \sum_{i=0}^n a_i x^i). The derivative is (g'(x) = \sum_{i=1}^n i a_i x^{i-1}). For (g'(x)) to be identically zero, every coefficient (i a_i) must equal (0) in (K). Since (K) has characteristic (p), this means either (a_i = 0) or (i \equiv 0 \mod p).

In other words, all non-zero terms in (g(x)) have exponents that are multiples of (p). We can rewrite (g(x)) as:
[
g(x) = \sum_{j=0}^m a_{pj} x^{pj} = h_1(x^p)
]
where (h_1(x) = \sum_{j=0}^m a_{pj} x^j), and (\deg h_1(x) = n/p < n) (since (n) must be a multiple of (p); otherwise, the leading term's coefficient (n a_n) would be non-zero, making (g'(x) \neq 0) and contradicting inseparability).

Now we have two cases for (h_1(x)):

1. Case 1: (h_1(x)) is separable
   We're done immediately! Just take (t(x) = h_1(x)) and (k=1), so (g(x) = t(x^p)).
2. Case 2: (h_1(x)) is inseparable
   By our inductive hypothesis (since (\deg h_1(x) < n)), there exists a separable polynomial (t(x)) and positive integer (k_1) such that (h_1(x) = t(x^{p{k_1}})). Substitute back into (g(x)):
   [
   g(x) = h_1(x^p) = t\left( (x^p){p^{k_1}} \right) = t(x^{p{k_1 + 1}})
   ]
   Let (k = k_1 + 1), and we get (g(x) = t(x^{pk})) where (t(x)) is separable.

This completes the inductive step, so the necessity direction holds.

Combining both directions, the lemma is fully proven.

内容的提问来源于stack exchange，提问作者PleasureTheory420