4
$\begingroup$

can anyone help me for this cubic equation ? can be solved without delta method?

$X^3-X^2-X-1=0$

(answer is $\sim 1.8393$)

$\endgroup$
8
  • $\begingroup$ Welcome to MSE! Please note, for further reference, the editing I did to your post. $\LaTeX$ is the language of choice for writing math here. Enjoy! $\endgroup$ – Andreas Caranti Jun 3 '14 at 16:48
  • 1
    $\begingroup$ You can rewrite the equation as $x^3 = x^2 + x + 1$, so $x = \left(x^2 + x + 1\right)^{1/3}$. Write this formula in your calculator, start with $x = 1$, and repeatedly substitute the answer as $x$, and you'll get closer and closer to that answer. See en.wikipedia.org/wiki/Fixed-point_iteration $\endgroup$ – M. Vinay Jun 3 '14 at 16:55
  • $\begingroup$ @ReneSchipperus The other two roots are complex. $\endgroup$ – M. Vinay Jun 3 '14 at 16:56
  • $\begingroup$ Some better methods are bisection method, false position (regula-falsi) method, and Newton-Raphson method. But fixed-point iteration is easy and works in this case. $\endgroup$ – M. Vinay Jun 3 '14 at 16:59
  • 1
    $\begingroup$ see mathworld.wolfram.com/TribonacciConstant.html by Cardano's method $\endgroup$ – Will Jagy Jun 3 '14 at 17:37
12
$\begingroup$

Here's a solution that finds all three roots of the function

$$x^3-x^2-x-1=0$$ Substitute $y=x-\frac{1}{3}$: $$-\frac{4}{3}-y-(y+\frac{1}{3})^2+(y+\frac{1}{3})^3=0$$ Expanding this gives: $$y^3-\frac{4}{3}y-\frac{38}{27}=0$$ If $y=\frac{\lambda}{z}+z$ then $z=\frac{1}{2}\left(y+\sqrt{y^2-4\lambda}\right)$ $$-\frac{38}{27}-\frac{4}{3}\left(z+\frac{\lambda}{z}\right)+\left(z+\frac{\lambda}{z}\right)^3=0$$ Multiply both sides by $z^3$ $$z^6+z^4\left(3\lambda-\frac{4}{3}\right)-\frac{38z^3}{27}+z^2\left(3\lambda^2-\frac{4\lambda}{3}\right)+\lambda^3=0$$ Substitute $\lambda=\frac{4}{9}$ and $u=z^3$ $$u^2-\frac{38}{27}u+\frac{64}{729}=0$$ Choose one of the solutions $$u=\frac{1}{27}(19+3\sqrt{33})$$ Substitute back for $u=z^3$ $$z^3=\frac{1}{27}(19+3\sqrt{33})$$ Solving for $z$ $$z=\frac{1}{3}\sqrt[3]{19+3\sqrt{33}}$$ $$z=-\frac{1}{3}\sqrt[3]{-19+3\sqrt{33}}$$ $$z=\frac{1}{3}(-1)^{2/3}\sqrt[3]{19+3\sqrt{33}}$$ Starting with the first solution, we substitute back for $z=\frac{y}{2}+\frac{1}{2}\sqrt{y^2-\frac{16}{9}}$ Solving this for $y$ gives $$y=\frac{4}{3\sqrt[3]{19+3\sqrt{33}}}+\frac{1}{3}\sqrt[3]{19+3\sqrt{33}}$$ Substitute back for $y=x-\frac{1}{3}$ $$x=\frac{1}{3}+\frac{4}{3\sqrt[3]{19+3\sqrt{33}}}+\frac{1}{3}\sqrt[3]{19+3\sqrt{33}}$$ This is the first solution for $x$, going on to the second solution $$z=-\frac{1}{3}\sqrt[3]{-19-3\sqrt{33}}$$ Substitute back for $z=\frac{y}{2}+\frac{1}{2}\sqrt{y^2-\frac{16}{9}}$ $$\frac{y}{2}+\frac{1}{2}\sqrt{y^2-\frac{16}{9}}=-\frac{1}{3}\sqrt[3]{-19-3\sqrt{33}}$$ Solve for $y$ $$y=\frac{4(-1)^{2/3}}{3\sqrt[3]{19+3\sqrt{33}}}-\frac{1}{3}\sqrt[3]{-19-3\sqrt{33}}$$ Substitute back for $y=x-\frac{1}{3}$ $$x=\frac{1}{3}+\frac{4(-1)^{2/3}}{3\sqrt[3]{19+3\sqrt{33}}}-\frac{1}{3}\sqrt[3]{-19-3\sqrt{33}}$$ Here's the second solution $$z=\frac{1}{3}(-1)^{2/3}\sqrt[3]{19+3\sqrt{33}}$$ For the last solution we substitute back for $z=\frac{y}{2}+\frac{1}{2}\sqrt{y^2-\frac{16}{9}}$ $$\frac{y}{2}+\frac{1}{2}\sqrt{y^2-\frac{16}{9}}=\frac{1}{3}(-1)^{2/3}\sqrt[3]{19+3\sqrt{33}}$$ Solve the equation for $y$ $$y=\frac{(-1)^{2/3}}{3}\sqrt[3]{19+3\sqrt{33}}-\frac{4}{3}\sqrt[3]{\frac{-1}{19+3\sqrt{33}}}$$ Substitute back for $x=y-\frac{1}{3}$ $$x=\frac{1}{3}+\frac{(-1)^{2/3}}{3}\sqrt[3]{19+3\sqrt{33}}-\frac{4}{3}\sqrt[3]{\frac{-1}{19+3\sqrt{33}}}$$ I have collected the 3 solutions that I found above here. $$x=\frac{1}{3}+\frac{4}{3\sqrt[3]{19+3\sqrt{33}}}+\frac{1}{3}\sqrt[3]{19+3\sqrt{33}}\approx1.83928675521416113255$$ $$x=\frac{1}{3}+\frac{4(-1)^{2/3}}{3\sqrt[3]{19+3\sqrt{33}}}-\frac{1}{3}\sqrt[3]{-19-3\sqrt{33}}\approx-0.41964-0.60629i$$ $$x=\frac{1}{3}+\frac{(-1)^{2/3}}{3}\sqrt[3]{19+3\sqrt{33}}-\frac{4}{3}\sqrt[3]{\frac{-1}{19+3\sqrt{33}}}\approx-0.41964+0.60629i$$

$\endgroup$
5
$\begingroup$

@Darksonn 's answer certainly works, but if you only want the positive real solution, there is a "cubic formula" that is reasonably useful in this case.

Theorem: If the cubic equation $$X^3 + pX + q$$ ($p, q$ real) satisfies $$\frac{p^3}{27} + \frac{q^2}{4} \geq 0,$$ then a solution to the cubic equation is $$X = \sqrt[3]{- \frac{q}{2} + \sqrt{\frac{p^3}{27} + \frac{q^2}{4}}} + \sqrt[3]{- \frac{q}{2} - \sqrt{\frac{p^3}{27} + \frac{q^2}{4}}}.$$

This is the Cardano-Tartaglia Formula.

In our case, we take @Darksonn 's $y$-substitution $\displaystyle y = x - \frac{1}{3}$ and get

$$y^3 - \frac{4}{3}y - \frac{38}{27} =0.$$

So $p = \displaystyle - \frac{4}{3}$, $q = \displaystyle - \frac{38}{27}$; one may check that in this case, $$ \frac{p^3}{27} + \frac{q^2}{4} = \frac{- 64 \cdot 4 + 38^2}{27^2 \cdot 4} = \frac{297}{27^2}, \quad -\frac{q}{2} = \frac{19}{27},$$ so that the answer becomes \begin{align} y &= \sqrt[3]{\frac{19}{27} + \sqrt{\frac{297}{27^2}}} + \sqrt[3]{\frac{19}{27} - \sqrt{\frac{297}{27^2}}}\\ & = \sqrt[3]{\frac{19 + \sqrt{297}}{27}} + \sqrt[3]{\frac{19 - \sqrt{297}}{27}} \\ &= \frac{1}{3} \left(\sqrt[3]{19 + 3\sqrt{33}} + \sqrt[3]{19 - 3\sqrt{33}} \right). \end{align}

Since $\displaystyle y = x - \frac{1}{3}$, $\displaystyle x = y + \frac{1}{3}$, and we get $$x = \frac{1}{3} + \frac{1}{3} \left(\sqrt[3]{19 + 3\sqrt{33}} + \sqrt[3]{19 - 3\sqrt{33}} \right).$$


To see that our answer matches @Darksonn 's, note that the terms not already matching are $\sqrt[3]{19 - 3\sqrt{33}}$ on the one side and $\displaystyle \frac{4}{\sqrt[3]{19 + 3 \sqrt{33}}}$ on the other side. To see this equality, note that for all positive $a, b$, $$a - b = \frac{a^2 - b^2}{a + b}$$ and letting $a = 19$, $b = 3 \sqrt{33}$, we get $$19 - 3 \sqrt{33} = \frac{64}{19 + 3 \sqrt{33}}$$ Taking the cube root of both sides, we're done.


If you wish to handle the case where the discriminant $$\frac{p^3}{27} + \frac{q^2}{4}$$ is negative, or to understand the shortcomings of the above formula, I strongly recommend looking at Lecture 4, "Equations of Degree Three and Four," in Fuchs and Tabachnikov, Mathematical Omnibus [Amer. Math. Soc., Providence, 2007]. It is a good "elementary" explanation of what is going on.

Yet other good methods are mentioned in the answers to the similar problem https://math.stackexchange.com/questions/612765/find-roots-of-the-cubic-equation-x3-x2-3-0-without-using-calculator?rq=1 .

$\endgroup$

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.