1
$\begingroup$

Given that $f(x)=\sum\limits_{k=0}^n a_kx^k$ is an irreducible polynomial, show that $g(x)=\sum\limits_{k=0}^n a_kx^{n-k}$ is irreducible too. I don't see how this specific coefficient exchange would preserve the irreducibility. Is there something else I should be looking at?

  • 2
    Hint: $g(x)={x^n}f(\frac 1x)$ Now suppose $g(x)$ can be written as $p(x)q(x)$. – 2017-01-08
  • 1
    Of course you want to assume $a_n \ne 0$, otherwise $f(x) = 1 + x + 0 x^2$ is a counterexample. – 2017-01-08
  • 0
    In addition to the objection raised by @RobertIsrael we have the edge case of the irreducible polynomial $f(x)=x$ when the reciprocal $g(x)$ is a unit of the polynomial ring rather than an irreducible element. – 2017-01-09

3 Answers 3

2

Since $f$ is irreducible, $f\neq a_n X^n$, so $\deg g\geq 1$.

Note that $g(X)=X^nf(\frac 1X)$. Supposing $g(X)=r(X)s(X)$ with $\deg r\geq1$, you get $$X^nr(\frac 1X)s(\frac 1X)=f(X)$$ that is $$ X^{n-\deg g} \left(X^{\deg r}r(\frac 1X)\right) \left(X^{\deg s}s(\frac 1X)\right)=f(X)$$

If $\deg (X^{\deg r}r(\frac 1X))\geq 1$ or $\deg (X^{\deg s}s(\frac 1X))\geq 1$ we get a contradiction with the irreducibility of $f$.

Otherwise, $X^{\deg r}r(\frac 1X)$ and $X^{\deg s}s(\frac 1X)$ are both constant, which implies $n-\deg g = \deg f = n$ hence $\deg g = 0$, which is in contradiction with the first line of my answer.

  • 0
    How is deg(r)+deg(s)=n? Could you explain this further? Thanks. – 2017-01-08
  • 0
    @Desperado I've fixed my argument, I hope it's clear now. – 2017-01-08
1

Hint:

If $\deg f=n$, $g(x)=x^nf(1/x)$ and $f(x)=x^ng(1/x)$.

1

Hint The reversal map $\, f\mapsto x^d f(x^{-1}),\ d=\deg f\,$ is multiplicative, being the product of two multiplicative maps, namely $\,f(x)\mapsto f(x^{-1}),\,$ and $\,f(x)\mapsto x^{\,\deg f}\ $ (an "exponental" of the additive degree map). Being multiplicative, it preserves (ir)reducibility.