2

I have been trying to find all the different methods for factoring cubics and so far in my search I have come across:

1)Using the sum/difference of cubes

2)The grouping method

3)Using the rational root test (and assuming you find a root) followed by synthetic division.

4)The discriminant approach ( which can be a little messy )

But I was looking over an old assignment and there was this question I got wrong at the time:

Determine the splitting field of

$f(x)=x^3-3x+1$ over $\Bbb Q$

Hint: If $\alpha$ is a root compute $f(1-\tfrac{1}{\alpha})$.

But none of the method I mentioned above give roots which are in agreement with the online calculator I'm using.

My questions are :

1) What method for factoring cubics can I use here ?

2) What are some other useful methods of factoring cubics I havent't mentioned here?( I hope to find an exhaustive list so I can always factor any cubic)

3) Is there any method which one can use on ${ANY}$ cubic, to find factors/roots ?

excalibirr
  • 2,855
  • Wikipedia gives the general formula for solving a cubic equation. But it's rather messy, I'm afraid. – TonyK Jun 14 '19 at 11:26
  • @TonyK I tried that approach ( the discrimant) but I was getting complex roots when the onle calculator said that they were real, maybe I did it wrong though. Is the approach you linked supposed to be fullproof ? – excalibirr Jun 14 '19 at 11:31
  • 4
    Well, yes. But not always very useful, as you found out. But you don't have to find the roots (they are all real, by the way); you just have to find the splitting field. The point of the hint is that if $\alpha$ is a root, then $f(1-\frac{1}{\alpha})=0$, so $1-\frac{1}{\alpha}$ is also a root. This, coupled with the fact that $\alpha=1-\frac{1}{\alpha}$ has no real solution, means the roots are of the form $\alpha,\beta,\gamma$, where $\alpha=1-\frac{1}{\beta},\beta=1-\frac{1}{\gamma},$ and $\gamma=1-\frac{1}{\alpha}$. That's as far as I can go with my limited knowledge of Galois theory. – TonyK Jun 14 '19 at 11:44
  • You can find the same question in Dummit and Foote, Abstract Algebra, Third Edition, Page no. 618. They ask to find the splitting field in terms of $\alpha$ only. From Tony's work, it can be easily seen that the splitting field is $\mathbb Q(\alpha)$. – cqfd Jun 14 '19 at 11:58
  • 3
    Here is an answer sheet for your problem. It gives one root as $e^{i2\pi/9}+e^{-i2\pi/9}$; the other two roots can be constructed from my comment. – TonyK Jun 14 '19 at 12:02

2 Answers2

1

This polynomial factors as $$(x - \alpha)(x - (1 - 1/\alpha))(x - (-\alpha^2 - \alpha + 2))$$ where $\alpha$ is any root of the cubic.

You can find this by performing the long division $(x^3 - 3x + 1)/(x-\alpha)$ to get $x^2 + \alpha x + (\alpha^2 - 3)$ and then factoring that quadratic.

But we know that $1 - 1/\alpha$ is also a root so we can just long divide that out to get the final factor.

M. Winter
  • 30,828
1

If $\alpha$ is a root of the cubic, then so is $1-\dfrac{1}{\alpha}$.

Therefore, the third root is $1-\dfrac{1}{1-\dfrac{1}{\alpha}}=\dfrac{1}{1-\alpha}$.

Alternatively, since the product of the three roots is $-1$, third root is $\dfrac{-1}{\alpha\left(1-\dfrac{1}{\alpha}\right)}=\dfrac{1}{1-\alpha}$.

Finally, since $1 = (x^3-3x+1)+(-x^2 - x + 2)(1-x)$, we have $\dfrac{1}{1-\alpha}=-\alpha^2 - \alpha + 2$.

Bottom line, the splitting field of $f(x)=x^3-3x+1$ is $\mathbb Q(\alpha)$, where $\alpha$ is any root of $f$.

lhf
  • 221,500
  • +1 for applying the transformation twice, what an nice solution! –  Aug 01 '20 at 15:06
  • @rain1, see also https://math.stackexchange.com/questions/1767252/expressing-the-roots-of-a-cubic-as-polynomials-in-one-root – lhf Aug 01 '20 at 15:10