Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

The Primitive Element Theorem states that if $E/F$ is a finite separable field extension, then there exists an element $a$ such that $E=F(a)$.

There's a similar result I found, that I don't quite fully understand. For instance, let $F$ be a field and $F[a_1,\dots,a_n]$ be a finite separable extension. Suppose also that $F_u=F(u_1,\dots,u_n)$ is a purely transcendental extension of $F$, with $u_1,\dots,u_n$ algebraically independent over $F$. Why is it true that $F_u[a_1,\dots,a_n]=F_u[u_1a_1+\cdots+u_na_n]$?

I get that $u_1a_1+\cdots+u_na_n\in F_u[a_1,\dots,a_n]$, and so $F_u[u_1a_1+\cdots+u_na_n]\subseteq F_u[a_1,\dots,a_n]$. However, why is the converse true? I tried reproducing the argument for the primitive element theorem without success. Thank you.

Since the $u_i$ are algebraically independent over $F$, and $F_u[a_1,\dots,a_n]$ is finitely generated over $F_u$, I was trying to use the corollary Bill Cook pointed me to, to conclude that by setting $y_1=\sum_{j=1}^n u_ja_j$, then $F_u[a_1,\dots,a_n]$ is integral over $F_u[y_1]$. From this can I conclude the equality? I wary of how to proceed, as I do not know where separability comes into use.

share|cite|improve this question
I wonder if you're supposed to assume that $u_1, \ldots, u_n$ are algebraically independent (i.e. the transcendence degree of $F_u / F$ is $n$)? – Zhen Lin Feb 3 '12 at 18:13
Dear @Zhen Lin, that hypothesis slipped my mind. I have now added it, thank for pointing it out. – Waldott Feb 3 '12 at 20:59
In other words, $u_1,\ldots,u_n$ is a transcendence basis for $F_u$ over $F$. – Arturo Magidin Feb 3 '12 at 21:05
You might want to take a look at Serge Lang's Algebra: Chapter VIII Theorem 2.2 and surrounding text -- it's very closely related to your question. In my "revised 3rd edition" it's on page 358. – Bill Cook Feb 3 '12 at 21:24
Thanks for the reference, I've been reading through it. – Waldott Feb 3 '12 at 23:33
up vote 2 down vote accepted

I think this statement is proven exactly as the primitive element theorem is proven. I'll give you the argument in the case $n=2$, courtesy of J. S. Milne's notes on field theory, from which the general case follows.

Let $E=F[a,b]$ with $b$ separable over $F_u$. Let $f$ and $g$ be the minimum polynomials of $a$ and $b$ respectively over $F$. Let $a=a_1, a_2, \ldots ,a_s$ be the roots of $f$ in some field $K$ containing $E$, and let $b=b_1, b_2, \ldots ,b_t$ be the roots of $g$. For $j \neq 1$, we have $b_j \neq b_1$ so $$a_i +Xb_j=a_1 +Xb_1$$ has exactly one solution: $X= \displaystyle \frac{a_j -a_i}{b_1-b_j}$. This means that $a_i + \frac{u_2}{u_1} b_j \neq a + \frac{u_2}{u_1} b$ in $K_u$ unless $i=1=j$.

Now let $\gamma = a+\frac{u_2}{u_1}b$. Then the polynomials $g(X)$ and $f(\gamma-\frac{u_2}{u_1}X)$ have coefficients in $F_u[\gamma]$, and have $b$ as a root: $$g(b)=0, f(\gamma-\frac{u_2}{u_1}b)=f(a)=0.$$

Moreover, $b$ is their only common root, because $\gamma - \frac{u_2}{u_1}b_j \neq a_i$ unless $i=1=j$. Therefore $$\gcd(g(X),f(\gamma-\frac{u_2}{u_1}X))=X-b.$$

Though we've computed the gcd in some field that splits $fg$, it is a simple fact that the gcd of two polynomials has coefficients in the same field as the coefficients of the polynomials. Hence $b \in F_u[\gamma]$, and this implies that $a=\gamma - \frac{u_2}{u_1}b$ lies in $F_u[\gamma]$. Hence, $F_u[a,b]=F_u[\gamma]=F_u[a+\frac{u_2}{u_1}b]=F_u[au_1+bu_2]$.

share|cite|improve this answer
Many thanks Parsa. – Waldott Feb 10 '12 at 4:41

Let $\alpha$ be a primitive element for $F[a_1,...,a_n]/F$ and let $d=[F(\alpha):F]$. By looking at the proof of the PET, we can take $\alpha$ to be an $F$-linear combination of the $a_i$, $\alpha=\sum b_i a_i$.

It suffices to prove that $[F_u[a_1 u_1+\cdots + a_n u_n]:F_u]=d$. So suppose $f(a_1 u_1+\cdots + a_n u_n)=0$ where $f(T)\in F_u[T]$ is nonzero. Let $k=deg_T(g)$. Viewing $f(a_1 u_1+\cdots + a_n u_n)$ as an element $g(u_1,...,u_n)\in F[a_1,...,a_n](u_1,...,u_n)$, the transcendence of the $u_i$'s implies that $g=0$. In particular, $g(b_1,...,b_n)=0$. But $g(b_1,...,b_n)$ is a degree $k$ polynomial in $\alpha=\sum b_i a_i$, hence $k\geq d$.

share|cite|improve this answer
Thanks for answering, Ryan. – Waldott Feb 10 '12 at 4:41

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.