I am working on an exercize and I know how to more bruteforcely solve it through pure algebra in its simplest form, but it's such a massive mess to demonstrate so I would like to see if there is another way to solve this

Let $x^3 + ax^2+bx+c$ be a given polynomial (with $a,b,c\in Z$ being acctual numbers but for here I am curiuos in the general case) and we know that it has the roots $x_1,x_2,x_3$ which are given, now determine the polynomial that has the following roots $x_1^2$, $x_2^2$ and $x_3^2$

There are several more but I am mostly curious at how one would go about this in a more elegant mannar than just applying the cubical solution formula and then use $p(x)=(x-x_1^2)(x-x_2^2)(x-x_3^2)$ simplifying and such, while possible it is such a hazzle. Is there a more elegant manner?

For the sake of an example, let's take the polynomial $x^3-7x^2+6x-3$ and it has the roots $x_1$ etc, how would I find the coefficients for the new polynomial with the roots of the stated polynomials being squared?

  • $\begingroup$ An idea would be to factorize $p(x^2)$ as a product of $\;x^3 + ax^2+bx+c\;$ and the 'conjugate' $\;x^3-ax^2+bx-c\;$ (when the sign of the $x_i$ are changed). Replacing $x$ by $x^2$ in the result should return the wished polynomial if I didn't make any error at least... $\endgroup$ – Raymond Manzoni Feb 26 '15 at 14:58

Consider $z:=x^2$ and factorize $\;p(z)=p(x^2)=(x^2-x_1^2)(x^2-x_2^2)(x^2-x_3^2)\;$ as a product of $\;(x^3 + ax^2+bx+c)\;$ and the 'conjugate' $\;(x^3-ax^2+bx-c)=(x+x_1)(x+x_2)(x+x_3)\;$ (obtained when the signs of the three $x_i$ are reversed as you may notice by replacing $x$ with $-x$).

The substitution of $x^2$ by $z$ in the result will return the wished polynomial (since the roots of $\;p(z)=0\,$ are indeed $z=x_i^2\,$).

I obtained thus $\;p(z)=z^3+(2b-a^2)\,z^2+(b^2-2ac)\,z-c^2$.

I considered the simple example $\;(x-1)(x-3)(x-5)=x^3-9\,x^2+23\,x-15\;$ and used the previous formula to get following result (as confirmed by Wolfram Alpha) : $$p(x)=x^3-35\,x^2+259\,x-225$$

  • 1
    $\begingroup$ Agrees with my answer $\ x(x+b)^2-(ax+c)^2\ \ $ $\endgroup$ – Bill Dubuque Feb 26 '15 at 15:32
  • $\begingroup$ @Bill Dubuque: Yes :-) Thanks for the confirmation (both points of view are interesting!). $\endgroup$ – Raymond Manzoni Feb 26 '15 at 15:33
  • $\begingroup$ This is so brilliant! $\endgroup$ – Zelos Malum Feb 26 '15 at 17:36
  • $\begingroup$ Would this method be possible for permutations of the roots and other powers? Like $x_1^3$, $x_1^4$ etc? or $x_1 x_2^2$? Or does an entirely idfferent approch need to be used? $\endgroup$ – Zelos Malum Feb 26 '15 at 19:54
  • $\begingroup$ Thanks @ZelosMalum, see too the method proposed by Bill Dubuque. Well, I don't know concerning the general case but for $(x_i)^n$ for example I would try to replace the $(x,-x)$ that I considered by $(x\,e^{2\pi i k/n})_{k=0,1,\cdots,n-1}$ that is use a complete complex factorization (the method proposed could thus be powerful!). Wishing you fine generalizations, $\endgroup$ – Raymond Manzoni Feb 26 '15 at 22:12

$\!\begin{align} {\rm If}\ \ \color{#c00}{z = x^2}\ \ {\rm then}\ \ 0 = &\ \ \color{#c00}{x^2}x\!+a\color{#c00}{x^2}\!+bx+c\\ = &\ \ \ \color{#c00}z\ x+a\ \color{#c00}z+bx+c\\ \Rightarrow\!&\ \ (z\!+\!b)\,x = -(az\!+\!c)\\ \Rightarrow\!&\ \ (z\!+\!b)^2 z =\ \ (az\!+\!c)^2\ \ \text{by squaring prior}\\ \end{align}$

The prior equation yields a cubic in $z$ having roots $\, z = x^2$


There is a very simple way to do so.
In the cubic polynomial, just replace is $x$ by $\sqrt{x}$
The final answer will be $$\sqrt{x}^3+ax+b\sqrt x+c$$
The reason why that works is this -
In the original polynomial, the roots are $x=x_1,x_2,x_3$
If we want the roots to be $x'=x_1^2,x_2^2,x_3^2$, we define $x$ in terms of the variable $x'$ like this (by taking the square root) $$\sqrt{x'}=x_1,x_2,x_3=x$$ Now we plug $\sqrt{x'}$ in the original polynomial, replace $x'$ by $x$ and we are done!

  • $\begingroup$ Enlighten me further because in your supposed solution I cannot seem to be getting what hte coefficients should be $\endgroup$ – Zelos Malum Feb 26 '15 at 15:11
  • $\begingroup$ The "polynomial" $\sqrt{x}^3+ax+b\sqrt x+c$ isn't a polynomial. $\endgroup$ – Martín-Blas Pérez Pinilla Feb 26 '15 at 15:15
  • $\begingroup$ @ZelosMalum The roots of the polynomial will be $x=x_1,x_2,x_3$. If we define a function in terms of a new variable $x'$, which has roots $x'=x_1^2,x_2^2,x_3^2$, we can get the $x$ in terms of $x'$. When we get $x$ in terms of $x'$, we can plug it back in the original cubic and get our answer. $\endgroup$ – AvZ Feb 26 '15 at 15:17
  • $\begingroup$ @Martín-BlasPérezPinilla Thanks for pointing that out. Fixed. $\endgroup$ – AvZ Feb 26 '15 at 15:18

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.