2
$\begingroup$

Let field extensions $K \subset L \subset F$ such that $F/L$ is normal extension, and $L/K$ is purely inseparable extension. Show that $F/K$ is normal extension.

My strategy: let $f(x)\in K[x]$ is irreducible polynomial, and $\alpha \in F$ is solution of $f(x)=0$. So if $\alpha \in L$ then $f(x)=0$ has only solution $\alpha$. Thus $f(x)$ split on $F$,then $F/K$ is normal.

PS: I don't what should I do when $\alpha \notin L$.

$\endgroup$
  • $\begingroup$ You asked this question 2 days ago, I gave you a hint, and now you've deleted the question? If my hint wasn't sufficient, you could've commented showing what you'd tried, and where you got stuck... Again, let $f_i$ be a family of polynomials in $L[x]$ whose roots generate $K/L$. Since $L/F$ is purely inseparable, how can you bring the $f_i$ down to $F[x]$ without affecting their roots? You might find the Frobenius endomorphism ($x\mapsto x^p$ in characteristic $p$) useful. $\endgroup$ – Warren Moore Apr 23 '13 at 8:14
  • 1
    $\begingroup$ @Muniain: It is extremely impolite to delete a question once someone has put effort into helping you - effectively you are just destroying their work. I have undeleted this original question, and merged your duplicate question here. Please don't do this again. $\endgroup$ – Zev Chonoles Apr 23 '13 at 8:30
2
$\begingroup$

Hint. Take a set of polynomials $f_i\in L[x]$ such that $F$ is generated by the roots of these polynomials. Since $L/K$ is purely inseparable, how can you make the $f_i$ into polynomials in $K[x]$ without affecting their roots?

You might find the fact that if $\text{char}(K)=p>0$, then $x\mapsto x^p$ is an endomorphism of $K$ (called the Frobenius endomorphism) useful.

$\endgroup$
0
$\begingroup$

Take a $K$ - homomorphism $T$ from the $F$ to the algebraic closure $M$ of $F$, we want to show that $T(F)=F$,(check this equivalence of normality in morandi's book) take an element "a" in $L$, so take then an irreducible polynomial in $K$ that solves "a" we say $f(x)$, then $f(T("a))=O.$ and since $L/K$ is Purele Ins. T("a")="a", in conclusion T(b)=b for every b in L, so T is really an L - homomorphism, since F/L is normal T(F)=F.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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