24
$\begingroup$

The quintic can be transformed to the one-parameter Brioschi quintic,

$$u^5-10\alpha u^3+45\alpha^2u-\alpha^2 = 0\tag1$$

This form is well-known for its connection to the symmetries of the icosahedron. I found that using a similar transformation, the general quintic can be reduced, also in radicals, to,

$$v^5-5\beta v^3+10\beta^2v-\beta^2 = 0\tag2$$

Question: Does $(2)$ appear anywhere when studying icosahedral symmetry or similar objects? If not, then what is the reason why the quintic can be reduced, in radicals, to this one-parameter form?

P.S. Incidentally, doing the transformation $u = 1/(x^2+20)$ on $(1)$ and $v = 4/(y^2+15)$ on $(2)$ reduces them to the rather nice similar forms,

$$(x^2+20)^2(x-5)+\frac{1}{\alpha}=0\tag3$$

$$(y^2+15)^2(y-5)+\frac{32}{\beta}=0\tag4$$

$\endgroup$

2 Answers 2

11
$\begingroup$

It took almost 20 long years (I first found the quintic back in April 2006), but I finally have the answer. Given the two similar one-parameter forms,

$$u^5-10\alpha u^3+45\alpha^2u-\alpha^2 = 0\tag1$$ $$v^5-5\beta v^3+10\beta^2v-\beta^2 = 0\tag2$$

The first is well-known for its connection to the icosahedron, Rogers-Ramanujan continued fraction $R(q)$, and the McKay-Thompson series of class $1A$ for the Monster.

The second (it turns out) is more for the tetrahedron, Ramanujan's cubic continued fraction $C(q)$, and the McKay-Thompson series of class $6A$ for the Monster.


I. Definitions

Let $q=e^{2\pi i \tau}$ and Dedekind eta function $\eta(\tau)$, then,

$$C(\tau) = \cfrac{q^{1/3}}{1 + \cfrac{q+q^2}{1 + \cfrac{q^2+q^4}{1 + \ddots}}}$$

Define the eta quotients,

$$B(\tau) = \frac{\eta(\tau)\,\eta^2(6\tau)}{\eta(3\tau)\,\eta^2(2\tau)},\quad C(\tau)\, = \frac{\eta(\tau)\,\eta^3(6\tau)}{\eta(2\tau)\,\eta^3(3\tau)}$$

as well as,

$$D_1(\tau) = \frac{\eta(\tau)\,\eta(3\tau)}{\eta(2\tau)\,\eta(6\tau)},\quad D_2(\tau) = \frac{\eta(\tau)\,\eta(2\tau)}{\eta(3\tau)\,\eta(6\tau)} $$

Suppressing $\tau$ for brevity, we have the neat relations,

\begin{align} \frac1{B^4}-\frac1{C^3} &= 1\\[6pt] \frac1{B^4}+9B^4 &= D_1^6+10\\[6pt] \frac1{C^3}-8C^3 &= D_2^4+7 \end{align}

Let $D_3=\frac{\eta(\tau)}{\eta(5\tau)}$ which is the McKay-Thompson series of class 5B for the Monster. These are then analogous to Ramanujan's well-known,

$$\frac1{R^5}-R^5 \,=\, D_3^6+11$$

but are McKay-Thompson series of class 6. The quotient $D_1^6$ is class 6C, $D_2^4$ is class 6D, while $\dfrac1{B^4}$ and $\dfrac1{C^3}$ are variants of class 6E (being A128633 and A105559). The forms $C$ and $D_2$ can solve the Bring quintic in a previous MO post, but $B$ and $D_1$ will be used in this one.


II. Modular equation

The modular equation between $B(\tau)$ and $B(5\tau)$ is more aesthetic,

$$\left(\frac{u^3}{v^3}-\frac{v^3}{u^3}\right)-5\left(\frac{u^2}{v^2}+\frac{v^2}{u^2}\right)+5\left(\frac{u}{v}-\frac{v}{u}\right)+3\left(\frac{1}{3u^2v^2}+3u^2v^2\right)=0$$

If $v = B(\tau)$, then the six roots $u$ are,

$$u = B(5\tau),\;-B\big(\tau/5\big),\;B\big(\tfrac{\tau+2}{5}\big),\;-B\big(\tfrac{\tau+4}{5}\big),\;B\big(\tfrac{\tau+6}{5}\big),\;-B\big(\tfrac{\tau+8}{5}\big)$$

Define,

$$X(\tau) = \left(\frac{B(5\tau)+B(\tau/5)}{\sqrt5\, B^3(\tau)\,D_1^3(\tau)}\right) \left(B\big(\tfrac{\tau+2}{5}\big)+B\big(\tfrac{\tau+8}{5}\big)\right) \left(B\big(\tfrac{\tau+4}{5}\big)+B\big(\tfrac{\tau+6}{5}\big)\right)$$

This function can solve the quintic.


III. Connection to cubic continued fraction

With scaling and change of variable, the equation

$$v^5-5\beta v^3+10\beta^2v-\beta^2 = 0$$

can be transformed to the simpler,

$$x^5+5x^3+10x+d=0$$

On the other hand, using the function $X(\tau)$ above, expand,

$$\prod_{n=0}^4 \big(x-X(\tau+2n)\big) = x^5+5x^3+10x+\sqrt{j_{6A}(\tau)-32\,}=0$$

where $j_{6A}(\tau)=\left(D_1^3(\tau)+\dfrac8{D_1^3(\tau)}\right)^2$ and is class 6A. Given $d$, one has to solve for $\tau$ in,

$$\sqrt{j_{6A}(\tau)-32\,} = d$$

which can be done. The roots $x_n$ for $n=0,1,2,3,4,$ are then,

$$x_n = X(\tau+2n)$$

$$\tau = \frac{_2F_1\big(\tfrac13,\tfrac23,1,\,1-\mu\big)}{_2F_1\big(\tfrac13,\tfrac23,1,\,\mu\big)}\sqrt{-\frac13}$$

$$\mu= \frac12\left(1-\sqrt{1-\frac{108}{j_3}}\right)$$

$$j_3 = \frac{432+45d^2+d^4-d(27+d^2)\sqrt{32+d^2}}2$$

Note: For simplicity, we used Ramanujan's theory of elliptic functions for signature $3$.


IV. Connection to tetrahedron

Given Ramanujan's cubic continued fraction $C=C(\tau)$. Define $z=-\dfrac{4C^2+C^{-1}}{3\sqrt2}$. (One can also define $z$ in terms of $B$.) Then a formula for the j-function is,

$$j(\tau) = \frac{-12^3(z^4-2\sqrt2\,z)^3}{(2\sqrt2\,z^3+1)^3}$$

However, these are polynomial invariants of the tetrahedral equation,

\begin{align} &a :=z^4-2\sqrt2\,z\\ &b :=2\sqrt2\,z^3+1\\ &c :=z^6+5\sqrt2\,z^3-1 \end{align}

such that,

$$a^3+b^3=c^2$$


V. Connection to McKay-Thompson series 6A

Recall that,

$$x^5+5x^3+10x+\sqrt{j_{6A}(\tau)-32\,}=0$$

with $j_{6A}(\tau)$ as defined above. Then for $\tau=\sqrt{-d}$, the quintic has a solvable Galois group. For example, $j_{6A}\big(\sqrt{-5/6}\big)=320$. Then,

$$x^5+5x^3+10x+\sqrt{288}=0$$

is solvable in radicals.


VI. Tschirnhausen transformations

A. Given the Brioschi,

$$u^5-10c u^3+45c^2u-c^2 = 0$$

and the transformation,

$$x = \frac{au+b}{c^{-1}u^2-\color{blue}3}\quad$$

with unknowns $(a,b,c)$ to eliminate $u$ using resultants transforms the Brioschi to the form $x^5+f_1x^2+f_2x+f_3=0$. Equating this to a principal quintic $x^5+5px^2+5qx+r=0$, then one has a system of 3 equations in 3 unknowns where the final equation in $b$ has a deg-$2$ factor.

B. Given the quintic,

$$v^5-5c v^3+10c^2v-c^2 = 0$$

and the transformation,

$$x = \frac{av+b}{c^{-1}v^2-\color{blue}2},\quad c = -\frac{a b+2b^2}{4a^2}$$

transforms said quintic into Euler form $x^5+f_1x^2+f_2=0$. Equating this to $x^5+5px^2+q=0$, one has a system of 2 equations in 2 unknowns $(a,b)$ where the final equation in $b$ has a deg-$3$ factor.

Or if we do the transformation,

$$x = \frac{av+b}{c^{-1}v^2-\color{blue}1},\quad c = -\frac{a b+3b^2}{3a^2}$$

transforms said quintic into Bring form $x^5+f_1x+f_2=0$. Equating this to $x^5+5px+q=0$, one again has a system of 2 equations in 2 unknowns $(a,b)$ where the final equation in $b$ has a deg-$4$ factor.

$\endgroup$
2
$\begingroup$

Geometry behind the Brioschi quintic is explained in Jerry Shurman book "Geometry of the Quintic" (available from his website http://people.reed.edu/~jerry/ ). However, it seems no explanation of (2) is given there.

P.S. Recent interesting paper: The Quintic, the Icosahedron, and Elliptic Curves by Bruce Bartlett. The quintic (2) is again not discussed.

$\endgroup$
1
  • 1
    $\begingroup$ My feeling is that since it is not trivial to reduce the quintic to one-parameter form, then there has to be some reason (geometric or other) why it can transformed into something so similar to the Brioschi form. Note that it is also very reminiscent to the solvable DeMoivre quintic $$x^5-5\alpha\, x^3+5\alpha^2 x+\beta = 0$$ $\endgroup$ Commented Dec 7, 2015 at 6:34

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .