Factorize $x^4+16x-12$ over reals.

The factor is $x^4+16x-12=(x^2-2x+6)(x^2+2x-2)$

It can be factorized again but I am stuck in this step.If we want to add and then subtract we have a lot of thing to add and subtract.Another idea that I saw in books is writing as this:


and then find $a,b,b',a'$ but there are two problems I can't find these here and we can say maybe it can factorized into one degree $3$ and one degree $1$ polynomial.

Isn't there a nice way to factor this?

  • $\begingroup$ No, probably not, you just got a lucky quartic. $\endgroup$ – Simply Beautiful Art Dec 25 '16 at 20:02
  • $\begingroup$ @SimpleArt But it is the problem of our national math olympiad it should have a way to factor easily. $\endgroup$ – Taha Akbari Dec 25 '16 at 20:03
  • $\begingroup$ The "other idea" you mention is indeed an easy way to factor it. Why are you not satisfied with it? $\endgroup$ – Dietrich Burde Dec 25 '16 at 20:06
  • $\begingroup$ @DietrichBurde I can't find $a,b,a',b'$ $\endgroup$ – Taha Akbari Dec 25 '16 at 20:07
  • 1
    $\begingroup$ Oh, that's easy. $a=2,b=-2$ and $a'=-2,b'=6$ follow directly from the equations by comparison at $1,x,x^2$. $\endgroup$ – Dietrich Burde Dec 25 '16 at 20:08


  • 8
    $\begingroup$ Well, I mean, this is easy to spot if you already know the answer. $\endgroup$ – Simply Beautiful Art Dec 25 '16 at 20:06
  • 2
    $\begingroup$ @SimpleArt, agree. I was also thinking the same thing to post as answer. But I don't think someone can strike it finely if he/she does not know the answer. $\endgroup$ – I am Back Dec 25 '16 at 20:08
  • 4
    $\begingroup$ But it is hard to discover what should we add and subtract. $\endgroup$ – Taha Akbari Dec 25 '16 at 20:08
  • $\begingroup$ you got it @TahaAkbari $\endgroup$ – I am Back Dec 25 '16 at 20:09
  • 4
    $\begingroup$ @SimpleArt However, it is also a standard procedure with the factorisation of a quartic to look for an expression as the difference of two squares, one quadratic and one linear. That is one way in which one can arrive at a cubic, and obtain the general solution of the quartic. $\endgroup$ – Mark Bennet Dec 25 '16 at 20:34

You can check that it's not linear-times-cubic by looking for rational roots, which must have the form $p/q$ where $p$ divides 12 and $q$ divides 1. When that doesn't work, the product-of-quadratics approach is your next best bet.


@JohnHughes explains how to find that this is not a cubic times linear factor. Once you figure that out, we know it's a quadratic times quadratic. To solve this, I'm going to go off the equation you had: $$x^4+16x-12=(x^2+ax+b)(x^2+a'x+b')$$ Expand the right side: $$x^4+16x-12=x^4+(a+a')x^3+(b+b'+aa')x^2+(ab'+ba')x+bb'$$ Now, we have $0x^3$ on the left side and $(a+a')x^3$ on the right side, so we have $a+a'=0 \implies a'=-a$. Furthermore, we have $-12$ as our constant on the left side and $bb'$ as our constant on the right side, so we have $bb'=-12 \implies b'=-\frac{12}b$. Susbtitute: $$x^4+16x-12=x^4+\left(b-\frac{12}{b}-a^2\right)x^2+\left(-\frac{12a}{b}-ab\right)x-12$$ We have $0x^2$ and $+16x$ on the left side, so by comparing such with the $x^2$ and $x$ coefficients on the right side, we get the following equations: $$b-\frac{12}{b}-a^2=0$$ $$-\frac{12a}{b}-ab=16$$ Multiply both sides by both equations $b$: $$b^2-12-ba^2=0$$ $$-12a-ab^2=16b$$

However, instead of guess and check, we can solve for $a$ in terms of $b$ in the second equation (I choose the second equation because it is easier): $$a=\frac{-16b}{12+b^2}$$ Now, remember that $b$ is a factor of $12$ since $bb'=-12$, so $b \in \{-12,-6,-4,-3,-2,-1,1,2,3,4,6,12\}$. Guess and check values of $b$ and solve for $a$. When you get an integer value of $a$, you know you've solved the problem, so substitute back into $a+a'=0$ to find $a'$ and $bb'=-12$ to find $b'$.

  • 1
    $\begingroup$ Though long but nice +1 $\endgroup$ – I am Back Dec 25 '16 at 20:18
  • $\begingroup$ @amWhy Thanks for the catch! I have edited my answer above. $\endgroup$ – Noble Mushtak Dec 25 '16 at 21:10

What you want is to show that the resolvent cubic associated with the quartic has a rational root. Look at this method based on the usual "general solution" for quartics:

Assume a factorization if the form:



Then expand the right side and match terms of like degree:

Degree 4: $1=1$ (forced)

Degree 3: $0=0$ (forced)

Degree 2: $p=-4s+(t_1+t_2)$

Degree 1: $q=(2\sqrt{s})(t_1-t_2)$

Degree 0: $r=t_1t_2$

We may treat the degree 2 and degree 1 equations as a linear system for $t_1, t_2$ in terms of $s$, then shbstitute the solution of this system into the degree 0 equation. Now that equation contains only $s$ as unknown, and after the usual simplifications we get this resolvent cubic:


This cubic equation has three roots corresponding to three ways the quartic roots could be paired to make quadratic factors. If you can find a rational root of this resolvent cubic you get a simplified quadratic factorization.

Here, the resolvent cubic is


with the rational root $s=1$. Then we have a factorization


where $t_1,t_2$ are determined as above with the known resolvent cubic root $s=1$ put into the degree 2 and degree 1 equations. Thus the linear system for the constant coefficients gives $t_1=6$, $t_2=-2$.


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.