4

How can we solve the following quintic in closed form ? $$16x^5-200x^3-200x^2+25x+30=0$$

What is special about this equation? What can we say about solvability? I'm not particularly into math. I saw this equation in a discussion group. Here are some ideas about the equation from non-mathematicians .

  1. There is no closed-form formula for $5$-degree equations. Therefore there is no solution. In my opinion this is wrong. For example, $x^5=1$ can be easily solved. The true version if that idea would be there is no general solution formula . (by radicals)

  2. The equation is not factored. Therefore, there is no closed-form solution. This idea is also wrong in my opinion. Because the factors do not have to be rational numbers.

  3. Wolfram Alpha does not give closed form for the equation. Therefore, it cannot be solved. I cannot comment on this idea.

I found that the quintic is irreducible over $\Bbb Q$.

I've heard a little bit about Galois theory. But I don't have the mathematical knowledge. If possible, could you help to find out if this equation is solvable? The problem author argues that the equation has a solution.

I tried WA software several times. But, it does not succeed.

hardmath
  • 644

3 Answers3

15

This polynomial $f$ has Galois group $F_5$, the Frobenius group of order 20. Since this group is solvable, then $f$ is solvable by radicals. Here are Magma commands showing this.

R<x> := PolynomialRing(Rationals());
f := 16*x^5-200*x^3-200*x^2+25*x+30;
G, data := GaloisGroup(f);
TransitiveGroupDescription(G);
IsSolvable(G);

which give as output

F(5) = 5:4
true

The docstring for the command SolveByRadicals describes how it gives solutions to a solvable polynomial:

Compute a splitting field as a radical tower and the roots of $f$ as elements in this tower. Return also the non-trivial roots of unity used.

(See here for more details.) We enter

K, roots := SolveByRadicals(f : Name := "K.");
K:Maximal;

which yields output

   K<K.1>
     |
     |
  $1<K.2>
     |
     |
  $2<K.3>
     |
     |
  $3<K.4>
     |
     |
     Q

K : K.1^5 + 1/512(-1875K.4 - 3125)K.3K.2 - 2560000000 $1 : K.2^2 + 15118284881920000K.4 - 34359738368000000 $2 : K.3^2 + 32K.4 + 160 $3 : K.4^2 - 5

Denote by $\alpha_i$ the elements K.i for $i = 1, \ldots, 4$. Then this output is saying that we have a chain of simple radical extensions $$ K = K_1 \supseteq K_2 \supseteq K_3 \supseteq K_4 \supseteq \mathbb{Q} $$ where $K_i = K_{i+1}(\alpha_i)$ and the $\alpha_i$ satisfy the equations \begin{align*} 0&={\alpha_1}^5+1/512(-1875\alpha_4-3125)\alpha_3\alpha_2-2560000000\\ 0&={\alpha_2}^2+15118284881920000\alpha_4-34359738368000000\\ 0&={\alpha_3}^2+32\alpha_4+160\\ 0&={\alpha_4}^2-5\,.\end{align*} Or in other words, \begin{align*} \alpha_4 &= \sqrt{5}\\ \alpha_3 &= \sqrt{-32 \alpha_4 - 160}\\ \alpha_2 &= \sqrt{-15118284881920000 \alpha_4 + 34359738368000000}\\ \alpha_1 &= \sqrt[5]{1/512 (1875 \alpha_4 + 3125) \alpha_3 \alpha_2 + 2560000000} \end{align*} up to choosing the correct roots of unity when taking these roots.

To express the roots of $f$ in terms of these generators, we enter [K!r : r in roots];, which yields

[
(1/268435456000000000*(-3*K.4 - 5)*K.3*K.2 + 1/128000000)*K.1^4 + 1/640000*K.1^3 + (1/33554432000000*(-3*K.4 - 5)*K.3*K.2 + 1/16000)*K.1^2 + 1/80*K.1,
((1/268435456000000000*(2*K.4 + 5)*K.3 + 1/13421772800000000*(K.4 + 3))*K.2 + (1/4096000000*(K.4 - 1)*K.3 + 1/512000000*(-K.4 - 1)))*K.1^4 + (-1/10240000*K.3 + 1/2560000*(K.4 - 1))*K.1^3 + ((1/67108864000000*(-K.4 - 5)*K.3 + 1/838860800000*(K.4 + 2))*K.2 + (1/256000*K.3 + 1/64000*(K.4 - 1)))*K.1^2 + (1/2560*(-K.4 + 1)*K.3 + 1/320*(-K.4 - 1))*K.1,
((1/536870912000000000*(-K.4 - 5)*K.3 + 1/6710886400000000*(K.4 + 2))*K.2 + (1/2048000000*K.3 + 1/512000000*(K.4 - 1)))*K.1^4 + (1/20480000*(K.4 - 1)*K.3 + 1/2560000*(-K.4 - 1))*K.1^3 + ((1/33554432000000*(2*K.4 + 5)*K.3 + 1/1677721600000*(-K.4 - 3))*K.2 + (1/512000*(-K.4 + 1)*K.3 + 1/64000*(-K.4 - 1)))*K.1^2 + (-1/1280*K.3 + 1/320*(K.4 - 1))*K.1,
((1/268435456000000000*(2*K.4 + 5)*K.3 + 1/13421772800000000*(-K.4 - 3))*K.2 + (1/4096000000*(-K.4 + 1)*K.3 + 1/512000000*(-K.4 - 1)))*K.1^4 + (1/10240000*K.3 + 1/2560000*(K.4 - 1))*K.1^3 + ((1/67108864000000*(-K.4 - 5)*K.3 + 1/838860800000*(-K.4 - 2))*K.2 + (-1/256000*K.3 + 1/64000*(K.4 - 1)))*K.1^2 + (1/2560*(K.4 - 1)*K.3 + 1/320*(-K.4 - 1))*K.1,
((1/536870912000000000*(-K.4 - 5)*K.3 + 1/6710886400000000*(-K.4 - 2))*K.2 + (-1/2048000000*K.3 + 1/512000000*(K.4 - 1)))*K.1^4 + (1/20480000*(-K.4 + 1)*K.3 + 1/2560000*(-K.4 - 1))*K.1^3 + ((1/33554432000000*(2*K.4 + 5)*K.3 + 1/1677721600000*(K.4 + 3))*K.2 + (1/512000*(K.4 - 1)*K.3 + 1/64000*(-K.4 - 1)))*K.1^2 + (1/1280*K.3 + 1/320*(K.4 - 1))*K.1
]

You can try this code for yourself using the Magma online calculator.

Angelo
  • 12,328
Viktor Vaughn
  • 19,278
  • 1
    Thanks, the solution is not readable, can you share us exact solution ? – hardmath Jun 16 '23 at 21:05
  • 1
    Viktor, what are $K.4, K.3, K.2, K.1$ ? Could you write explicitly their values? – Angelo Jun 16 '23 at 22:41
  • One minor question, was $F_{20}$ mistyped as $F_5$? – Oscar Lanzi Jun 16 '23 at 22:46
  • @OscarLanzi Is this a step by step solution? – hardmath Jun 16 '23 at 22:58
  • @user1174831 It looks like one to me. Ask the original answerer. – Oscar Lanzi Jun 16 '23 at 23:15
  • @OscarLanzi Original answerer doesn't answer comments. – hardmath Jun 16 '23 at 23:18
  • 1
    @OscarLanzi I think there are differing conventions on how this group is denoted, similar to $D_n$ vs. $D_{2n}$ for dihedral groups. $F_5$ is the notation used both in the LMFDB link I gave and on its GroupNames page. $F_5$ is probably used because it's a subgroup of $S_5$, or maybe because it's isomorphic to the group of affine transformations on $\mathbb{F}_5$. – Viktor Vaughn Jun 17 '23 at 00:54
  • @Angelo The elements K.i are defined by the equations given below the field diagram. I'll add some further clarification. – Viktor Vaughn Jun 17 '23 at 00:55
  • 1
    How did you find the Galois group? – David Lui Jun 17 '23 at 04:41
  • @DavidLui I computed it using Magma, as in the post. – Viktor Vaughn Jun 17 '23 at 04:44
  • Why do you ignore my comment, can I ask ? – hardmath Jun 17 '23 at 04:58
  • 1
    This link contains Magma program to get the solutions of the OP equation. – Angelo Jun 17 '23 at 05:04
  • @user1174831 What is your question? I thought my edit would have addressed confusion over the meaning of the expressions for the roots. – Viktor Vaughn Jun 17 '23 at 05:07
  • @ViktorVaughn, could let me fix a part of your answer in fact it does appear as follows ? Moreover, could I change the final link in order to contain the program which solves the OP equation ? begin{align} 0 &= {\alpha_1}^5 + 1/512 (-1875 \alpha_4 - 3125) \alpha_3 \alpha_2 - 2560000000\ 0 &= {\alpha_2}^2 + 15118284881920000 \alpha_4 - 34359738368000000\ 0 &= {\alpha_3}^2 + 32 \alpha_4 + 160\ 0 &= {\alpha_4}^2 - 5, . \end{align} – Angelo Jun 17 '23 at 05:10
8

I have no idea how to solve the equation by radicals. However, I wanted to write it assuming that this answer will somehow work for us as well .


$$ \begin{align}16x^5-200x^3-200x^2&+25x+30=0\end{align} $$

Using the linear substitution $\thinspace x:=u-\dfrac 12\thinspace $, we have :

$$ \begin{align}2u^5-5u^4-20u^3+10u^2&+10u-1=0\end{align} $$

Then, recall the following trigonometric angle formula :

$$ \begin{align}\tan (5\alpha)&=\frac{\tan^5\alpha-10\tan^3\alpha+5\tan \alpha}{5\tan^4\alpha-10\tan^2\alpha+1}\end{align} $$

Let $\thinspace\tan\alpha=u\thinspace$ and $\thinspace\tan (5\alpha)=a\thinspace$, then :

$$ \begin{align}u^5-5ay^4-10u^3+10au^2&+5u-a=0\end{align} $$

which is equivalent to :

$$ \begin{align}2u^5-10au^4-20u^3+20au^2&+10u-2a=0\end{align} $$

Therefore, taking $\thinspace a=\dfrac 12\thinspace $ then reduces the original equation to :

$$ \begin{align}&\tan (5\alpha)=\frac 12\\ \implies &\alpha=\frac 15\left(\arctan \left(\frac 12\right)+\pi k\right)\end{align} $$


Thus, the final answer becomes :

$$ \begin{align}\bbox[5px,border:2px solid #C0A000]{x=-\frac 12+\tan\left(\frac 15\left(\arctan \left(\frac 12\right)+\pi k\right)\right)}\end{align} $$

where $\thinspace k\in\{0,1,2,3,4\}\thinspace .$

lone student
  • 14,709
  • I would appreciate insight into how you decided which trig function to recall in this situation! It seems a bit miraculous without further context. – Benjamin Dickman Jun 17 '23 at 18:18
  • 2
    @BenjaminDickman Spend enough time with equations, one becomes familiar with some forms. For ex, the DeMoivre quintic is easy to recognize, $$x^5+5ax^3+5a^2x+b=0$$ But the tangent quintic is harder to spot, $$x^5 + 5a x^4 + 10 b x^3 + 10a b x^2 + 5b^2 x + a b^2=0$$ By the OP's inspired change of variables to get, $$2u^5-5u^4-20u^3+10u^2+10u-1=0$$ which we divide by 2, $$u^5-\tfrac52 u^4-10u^3+5u^2+5u-\tfrac12=0$$ one can now recognize the tangent quintic with $a=-\frac12$ and $b=-1$. Of course, what changes to do in the first place, there's where inspiration and ingenuity comes in. – Tito Piezas III Jun 17 '23 at 18:53
  • 2
    In effect the trigonometric solution is a radical solution; the forms are interconverted using the relationships between trig functions and complex exponentials. The given solution for $x$ may thus be rendered as $(-1/2)+[(1/i)(a-1)/(a+1)]$ where $a$ runs through all fifth roots of $(2+i)/(2-i)=(3+4i)/5$. Since all roots are real the radical form must involve complex radicands. – Oscar Lanzi Jun 17 '23 at 23:06
4

If we are content to evaluate the solvability, we can use the Cayley resolvent described by Wikipedia. This gives an unexpected result.

First we note that the given equation has a leading coefficient of $2^4$. If we multiply by $2$ and define $y=2x$, we get an equation with smaller integer coefficients:

$ y^5-50y^3-100y^2+25y+60=0$

The Cayley resolvent method is applied to this equation. The quintic is solvable if the combination

$[P_3(z)]^2-1024z\Delta$

has a rational zero, where $\Delta$ and the coefficients of the cubic polynomial are evaluated from the coefficients of the reduced quintic. Formulas are given in the referenced Wikipedia article.

For the case at hand, the relevant outcome is that $P_3$ happens to have a zero constant term:

$P_3(z)=z^3-3000z^2+12800000z+\color{blue}{0}.$

This forces the complete sextic to also have a zero constant term, and so the sought rational root of the Cayley resolvent is simply zero!

Thus this particular quintic is solvable. But actually obtaining the solution requires a complex formula which, as reported by Wikipedia, runs to three pages (with the Cayley resolvent root already in hand). We might be content just to use the Magma online calculator from here on out, as Viktor Vaughn does, or use the trigonometric trick by lone student.

Oscar Lanzi
  • 39,403
  • 2
    Given the quintic, $$x^5 + 10c x^3 + 10d x^2 + 5 e x + f = 0$$ then Watson's resolvent sextic has a zero constant term if the quintic's coefficients form a "square", $$\big(c^3 + d^2 - c e\big) \big(25 c^4 - 10 c^2 e + e^2 + 16 c d^2\big) = \big(c^2 d + d e - c f\big)^2$$ I tested it on your quintic in $y$ and it obeys the relation as well, so Cayley's and Watson's resolvents must be related. (Note: I just got a ping on an old post asking about resolvents with a zero root. Must be due to this new question.) – Tito Piezas III Jun 17 '23 at 14:37
  • 1
    @TitoPiezasIII I found that, the trigo formula worked . – lone student Jun 17 '23 at 16:33
  • 2
    @lonestudent Yes, I upvoted your answer. Very elegant and concise solution. – Tito Piezas III Jun 17 '23 at 17:07
  • 1
    @TitoPiezasIII Ahh, many thank you, I just wanted to ping you . Thank you very much . – lone student Jun 17 '23 at 17:32
  • @TitoPiezasIII "ping ... must be due to this new question" $;-;$ Right, and thanks again for the followup there. The example in my comment was this quintic, just with $x \mapsto x/2$ to make it a simpler monic. The two quadratics you mention happen to have integer coefficients in this case, if I am not mistaken. – dxiv Jun 17 '23 at 21:17
  • @TitoPiezasIII the two resolvents may be related. If we apply root squaring to the Watson form, as described in my answer to the pinged question, it looks like the Cayley form. – Oscar Lanzi Jun 17 '23 at 23:35