2
$\begingroup$

I want to prove that the polynomial

$$ f_p(x) = x^{2p+2} - cx^{2p} - dx^p - 1 $$

,where $c>0$ and $d>0$ are real numbers, has distinct roots. Also $p>0$ is an even integer. How can I prove that the polynomial $f_p(x)$ has distinct roots for any $c$,$d$ and $p$.

PS: There is a similar topic that How to prove that my polynomial has distinct roots?

$\endgroup$
  • 1
    $\begingroup$ Hint: A polynomial has distinct roots if and only if it (the polynomial) and its first derivative do not have any roots in common. See if you can use it here. $\endgroup$ – mathguy Apr 16 '16 at 15:53
  • $\begingroup$ Thank you for response. Of course I know this. But for this polynomial I didn't achieve. I would be grateful if you could give me the solution. $\endgroup$ – drxy Apr 16 '16 at 16:02
  • 6
    $\begingroup$ It seems that for $p=2$, $c=1$, $d\approx 2.6107186132760393498186490083840586275$ a root of $36d^4 + 125d^3 - 133d^2 - 801d - 899$, the polynomial does have multiple roots at $\pm 0.81073923436477156272173034309719150582 i$ (i.e., at the roots of $(6d + 2)x^2 + (d + 9)$, which is the $\gcd(f,f')$ for such $p,c,d$ $\endgroup$ – Hagen von Eitzen Apr 16 '16 at 16:24
  • $\begingroup$ @HagenvonEitzen how did you find the $d$ value ? Have you any program or any other tool for calculation this? Further, how can I prove it theoretically? $\endgroup$ – drxy Apr 20 '16 at 13:24
  • $\begingroup$ @HagenvonEitzen Is there any multiple roots for higher $p$ values? For example $p=4,6,8, \ldots$. $\endgroup$ – drxy Apr 25 '16 at 7:33
4
$\begingroup$

Here is a partial solution that shows there can be no multiple REAL roots. The proof doesn't work for complex numbers though (and I am not sure the result is even true in complex numbers).

$f' = (2p+2) x^{2p+1} - 2pcx^{2p-1} - pdx^{p-1}$. If $x$ is a multiple root of $f$, then both $f$ and $f'$ vanish at $x$. But $(2p+2) f - xf' = -2cx^{2p}-(p+2)dx^p-(2p+2) < 0$ for all $x \in \mathbb R$ because $p$ is even and $c$, $d$ and $p$ are all positive.

$\endgroup$
  • $\begingroup$ Thank you to answer. But I must prove it mathematical way. how can I prove it for complex numbers? $\endgroup$ – drxy Apr 20 '16 at 13:25
  • 1
    $\begingroup$ I don't know what you mean by "mathematical way" - what I gave you is a complete, "mathematical way" proof that the polynomial does not have multiple roots that are real numbers. For complex numbers, I said I don't even believe the result is true (so you can't prove it, if it's not true!) and in a comment to your post, Hagen von Eltzen demonstrated that my suspicion was correct: the result is FALSE in complex numbers. Please see Hagen's comment to your question. $\endgroup$ – mathguy Apr 20 '16 at 13:30
  • $\begingroup$ Mathematical way means that "I must prove it completely" both real and complex numbers as in this topic math.stackexchange.com/questions/1740673/…. Your solution includes real numbers. Besides, it's so important for me that the proof must include the complex numbers. Thank you so much. $\endgroup$ – drxy Apr 20 '16 at 13:34
  • 1
    $\begingroup$ Can you read what I said? In complex numbers the result is FALSE, how do you want to "prove" something that is not true? Hagen gave you a counter-example! $\endgroup$ – mathguy Apr 20 '16 at 13:50
  • $\begingroup$ Sorry my bad English. Thank you for all answers. Have you got any idea for $p=4,6,8$ or higher values. How can I be sure that the polynomial has multiple roots for any $p$ value. $\endgroup$ – drxy Apr 20 '16 at 13:55
-1
$\begingroup$

‘All’ you have to do is computing the g.c.d. of $f_p(x)$ and $f'_p(x)$ via the Euclidean algorithm. A polynomial has only simple roots (in $\mathbf C$) if and only if this polynomial and its derivative are coprime.

$\endgroup$
  • $\begingroup$ This should be the best and most familiar way to go. Moreover, you should expect the two polynomials to be relatively prime. I.e you should expect the gcd to be a constant. $\endgroup$ – user244643 Apr 17 '16 at 1:28
  • $\begingroup$ @Bernard Since I couldn't achieve it, I'm writing here. Thank you very much for valuable answer. $\endgroup$ – drxy Apr 20 '16 at 13:22
  • $\begingroup$ Not very helpful though, eh? HOW do you compute the gcd? Note that the problem is subtle; as shown by Hagen, the result is false in general, but as I have shown, it is true about real roots; so the proof definitely is not based on gcd alone. $\endgroup$ – mathguy Apr 20 '16 at 14:00

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.