In mathematics, the Dixon elliptic functions sm and cm are two elliptic functions (doubly periodic meromorphic functions on the complex plane) that map from each regular hexagon in a hexagonal tiling to the whole complex plane. Because these functions satisfy the identity




cm

3


⁡
z
+

sm

3


⁡
z
=
1


{\displaystyle \operatorname {cm} ^{3}z+\operatorname {sm} ^{3}z=1}

, as real functions they parametrize the cubic Fermat curve




x

3


+

y

3


=
1


{\displaystyle x^{3}+y^{3}=1}

, just as the trigonometric functions sine and cosine parametrize the unit circle




x

2


+

y

2


=
1


{\displaystyle x^{2}+y^{2}=1}

.
They were named sm and cm by Alfred Dixon in 1890, by analogy to the trigonometric functions sine and cosine and the Jacobi elliptic functions sn and cn; Göran Dillner described them earlier in 1873.

Definition
The functions sm and cm can be defined as the solutions to the initial value problem:

d

d
z



cm
⁡
z
=
−

sm

2


⁡
z
,



d

d
z



sm
⁡
z
=

cm

2


⁡
z
,

cm
⁡
(
0
)
=
1
,

sm
⁡
(
0
)
=
0


{\displaystyle {\frac {d}{dz}}\operatorname {cm} z=-\operatorname {sm} ^{2}z,\ {\frac {d}{dz}}\operatorname {sm} z=\operatorname {cm} ^{2}z,\ \operatorname {cm} (0)=1,\ \operatorname {sm} (0)=0}

Or as the inverse of the Schwarz–Christoffel mapping from the complex unit disk to an equilateral triangle, the Abelian integral:

z
=

∫

0


sm
⁡
z





d
w


(
1
−

w

3



)

2

/

3





=

∫

cm
⁡
z


1





d
w


(
1
−

w

3



)

2

/

3







{\displaystyle z=\int _{0}^{\operatorname {sm} z}{\frac {dw}{(1-w^{3})^{2/3}}}=\int _{\operatorname {cm} z}^{1}{\frac {dw}{(1-w^{3})^{2/3}}}}

which can also be expressed using the hypergeometric function:

sm

−
1


⁡
(
z
)
=
z

2



F

1




(





1
3



,



2
3



;



4
3



;

z

3




)




{\displaystyle \operatorname {sm} ^{-1}(z)=z\;{}_{2}F_{1}{\bigl (}{\tfrac {1}{3}},{\tfrac {2}{3}};{\tfrac {4}{3}};z^{3}{\bigr )}}

Parametrization of the cubic Fermat curve

Both sm and cm have a period along the real axis of




π

3


=

B



(





1
3



,



1
3





)


=




3


2
π





Γ

3




(





1
3





)


≈
5.29991625


{\displaystyle \pi _{3}=\mathrm {B} {\bigl (}{\tfrac {1}{3}},{\tfrac {1}{3}}{\bigr )}={\tfrac {\sqrt {3}}{2\pi }}\Gamma ^{3}{\bigl (}{\tfrac {1}{3}}{\bigr )}\approx 5.29991625}

with




B



{\displaystyle \mathrm {B} }

the beta function and



Γ


{\displaystyle \Gamma }

the gamma function:

1
3




π

3





=

∫

−
∞


0





d
x


(
1
−

x

3



)

2

/

3





=

∫

0


1





d
x


(
1
−

x

3



)

2

/

3





=

∫

1


∞





d
x


(
1
−

x

3



)

2

/

3











≈
1.76663875






{\displaystyle {\begin{aligned}{\tfrac {1}{3}}\pi _{3}&=\int _{-\infty }^{0}{\frac {dx}{(1-x^{3})^{2/3}}}=\int _{0}^{1}{\frac {dx}{(1-x^{3})^{2/3}}}=\int _{1}^{\infty }{\frac {dx}{(1-x^{3})^{2/3}}}\\[8mu]&\approx 1.76663875\end{aligned}}}

They satisfy the identity




cm

3


⁡
z
+

sm

3


⁡
z
=
1


{\displaystyle \operatorname {cm} ^{3}z+\operatorname {sm} ^{3}z=1}

. The parametric function



t
↦
(
cm
⁡
t
,

sm
⁡
t
)
,


{\displaystyle t\mapsto (\operatorname {cm} t,\,\operatorname {sm} t),}





t
∈


[



−



1
3





π

3


,



2
3




π

3




]




{\displaystyle t\in {\bigl [}{-{\tfrac {1}{3}}}\pi _{3},{\tfrac {2}{3}}\pi _{3}{\bigr ]}}

parametrizes the cubic Fermat curve




x

3


+

y

3


=
1
,


{\displaystyle x^{3}+y^{3}=1,}

with






1
2



t


{\displaystyle {\tfrac {1}{2}}t}

representing the signed area lying between the segment from the origin to



(
1
,

0
)


{\displaystyle (1,\,0)}

, the segment from the origin to



(
cm
⁡
t
,

sm
⁡
t
)


{\displaystyle (\operatorname {cm} t,\,\operatorname {sm} t)}

, and the Fermat curve, analogous to the relationship between the argument of the trigonometric functions and the area of a sector of the unit circle. To see why, apply Green's theorem:

A
=



1
2




∫

0


t


(
x

d
y

−
y

d
x

)
=



1
2




∫

0


t


(

cm

3


⁡
t
+

sm

3


⁡
t
)

d
t

=



1
2




∫

0


t


d
t
=



1
2



t
.


{\displaystyle A={\tfrac {1}{2}}\int _{0}^{t}(x\mathop {dy} -y\mathop {dx} )={\tfrac {1}{2}}\int _{0}^{t}(\operatorname {cm} ^{3}t+\operatorname {sm} ^{3}t)\mathop {dt} ={\tfrac {1}{2}}\int _{0}^{t}dt={\tfrac {1}{2}}t.}

Notice that the area between the



x
+
y
=
0


{\displaystyle x+y=0}

and




x

3


+

y

3


=
1


{\displaystyle x^{3}+y^{3}=1}

can be broken into three pieces, each of area






1
6




π

3




{\displaystyle {\tfrac {1}{6}}\pi _{3}}

:

1
2




π

3





=

∫

−
∞


∞




(


(
1
−

x

3



)

1

/

3


+
x


)



d
x








1
6




π

3





=

∫

−
∞


0




(


(
1
−

x

3



)

1

/

3


+
x


)



d
x

=

∫

0


1


(
1
−

x

3



)

1

/

3



d
x

.






{\displaystyle {\begin{aligned}{\tfrac {1}{2}}\pi _{3}&=\int _{-\infty }^{\infty }{\bigl (}(1-x^{3})^{1/3}+x{\bigr )}\mathop {dx} \\[8mu]{\tfrac {1}{6}}\pi _{3}&=\int _{-\infty }^{0}{\bigl (}(1-x^{3})^{1/3}+x{\bigr )}\mathop {dx} =\int _{0}^{1}(1-x^{3})^{1/3}\mathop {dx} .\end{aligned}}}

Symmetries

The function



sm
⁡
z


{\displaystyle \operatorname {sm} z}

has zeros at the complex-valued points



z
=



1

3





π

3


i
(
a
+
b
ω
)


{\displaystyle z={\tfrac {1}{\sqrt {3}}}\pi _{3}i(a+b\omega )}

for any integers



a


{\displaystyle a}

and



b


{\displaystyle b}

, where



ω


{\displaystyle \omega }

is a cube root of unity,



ω
=
exp
⁡



2
3



i
π
=
−



1
2



+




3

2



i


{\displaystyle \omega =\exp {\tfrac {2}{3}}i\pi =-{\tfrac {1}{2}}+{\tfrac {\sqrt {3}}{2}}i}

(that is,



a
+
b
ω


{\displaystyle a+b\omega }

is an Eisenstein integer). The function



cm
⁡
z


{\displaystyle \operatorname {cm} z}

has zeros at the complex-valued points



z
=



1
3




π

3


+



1

3





π

3


i
(
a
+
b
ω
)


{\displaystyle z={\tfrac {1}{3}}\pi _{3}+{\tfrac {1}{\sqrt {3}}}\pi _{3}i(a+b\omega )}

. Both functions have poles at the complex-valued points



z
=
−



1
3




π

3


+



1

3





π

3


i
(
a
+
b
ω
)


{\displaystyle z=-{\tfrac {1}{3}}\pi _{3}+{\tfrac {1}{\sqrt {3}}}\pi _{3}i(a+b\omega )}

.
On the real line,



sm
⁡
x
=
0
↔
x
∈

π

3



Z



{\displaystyle \operatorname {sm} x=0\leftrightarrow x\in \pi _{3}\mathbb {Z} }

, which is analogous to



sin
⁡
x
=
0
↔
x
∈
π

Z



{\displaystyle \sin x=0\leftrightarrow x\in \pi \mathbb {Z} }

.

Fundamental reflections, rotations, and translations
Both cm and sm commute with complex conjugation,

cm
⁡



z
¯






=



cm
⁡
z

¯


,




sm
⁡



z
¯






=



sm
⁡
z

¯


.






{\displaystyle {\begin{aligned}\operatorname {cm} {\bar {z}}&={\overline {\operatorname {cm} z}},\\\operatorname {sm} {\bar {z}}&={\overline {\operatorname {sm} z}}.\end{aligned}}}

Analogous to the parity of trigonometric functions (cosine an even function and sine an odd function), the Dixon function cm is invariant under






1
3





{\textstyle {\tfrac {1}{3}}}

turn rotations of the complex plane, and






1
3





{\textstyle {\tfrac {1}{3}}}

turn rotations of the domain of sm cause






1
3





{\displaystyle {\tfrac {1}{3}}}

turn rotations of the codomain:

cm
⁡
ω
z



=
cm
⁡
z
=
cm
⁡

ω

2


z
,




sm
⁡
ω
z



=
ω
sm
⁡
z
=

ω

2


sm
⁡

ω

2


z
.






{\displaystyle {\begin{aligned}\operatorname {cm} \omega z&=\operatorname {cm} z=\operatorname {cm} \omega ^{2}z,\\\operatorname {sm} \omega z&=\omega \operatorname {sm} z=\omega ^{2}\operatorname {sm} \omega ^{2}z.\end{aligned}}}

Each Dixon elliptic function is invariant under translations by the Eisenstein integers



a
+
b
ω


{\displaystyle a+b\omega }

scaled by




π

3


,


{\displaystyle \pi _{3},}

cm
⁡


(


z
+

π

3


(
a
+
b
ω
)


)


=
cm
⁡
z
,




sm
⁡


(


z
+

π

3


(
a
+
b
ω
)


)


=
sm
⁡
z
.






{\displaystyle {\begin{aligned}\operatorname {cm} {\bigl (}z+\pi _{3}(a+b\omega ){\bigr )}=\operatorname {cm} z,\\\operatorname {sm} {\bigl (}z+\pi _{3}(a+b\omega ){\bigr )}=\operatorname {sm} z.\end{aligned}}}

Negation of each of cm and sm is equivalent to a






1
3




π

3




{\displaystyle {\tfrac {1}{3}}\pi _{3}}

translation of the other,

cm
⁡
(
−
z
)



=


1

cm
⁡
z



=
sm
⁡


(


z
+



1
3




π

3




)


,




sm
⁡
(
−
z
)



=
−



sm
⁡
z


cm
⁡
z



=


1

sm
⁡


(


z
−



1
3




π

3




)





=
cm
⁡


(


z
+



1
3




π

3




)


.






{\displaystyle {\begin{aligned}\operatorname {cm} (-z)&={\frac {1}{\operatorname {cm} z}}=\operatorname {sm} {\bigl (}z+{\tfrac {1}{3}}\pi _{3}{\bigr )},\\\operatorname {sm} (-z)&=-{\frac {\operatorname {sm} z}{\operatorname {cm} z}}={\frac {1}{\operatorname {sm} {\bigl (}z-{\tfrac {1}{3}}\pi _{3}{\bigr )}}}=\operatorname {cm} {\bigl (}z+{\tfrac {1}{3}}\pi _{3}{\bigr )}.\end{aligned}}}

For



n
∈

{

0
,
1
,
2
}
,


{\displaystyle n\in \mathbb {\{} 0,1,2\},}

translations by






1
3




π

3


ω


{\displaystyle {\tfrac {1}{3}}\pi _{3}\omega }

give

cm
⁡


(


z
+



1
3




ω

n



π

3




)





=

ω

2
n





−
sm
⁡
z


cm
⁡
z



,




sm
⁡


(


z
+



1
3




ω

n



π

3




)





=

ω

n




1

cm
⁡
z



.






{\displaystyle {\begin{aligned}\operatorname {cm} {\bigl (}z+{\tfrac {1}{3}}\omega ^{n}\pi _{3}{\bigr )}&=\omega ^{2n}{\frac {-\operatorname {sm} z}{\operatorname {cm} z}},\\\operatorname {sm} {\bigl (}z+{\tfrac {1}{3}}\omega ^{n}\pi _{3}{\bigr )}&=\omega ^{n}{\frac {1}{\operatorname {cm} z}}.\end{aligned}}}

Specific values

More specific values

Sum and difference identities
The Dixon elliptic functions satisfy the argument sum and difference identities:

cm
⁡
(
u
+
v
)



=



sm
⁡
u

cm
⁡
u
−
sm
⁡
v

cm
⁡
v


sm
⁡
u


cm

2


⁡
v
−

cm

2


⁡
u

sm
⁡
v







cm
⁡
(
u
−
v
)



=




cm

2


⁡
u

cm
⁡
v
−
sm
⁡
u


sm

2


⁡
v


cm
⁡
u


cm

2


⁡
v
−

sm

2


⁡
u

sm
⁡
v







sm
⁡
(
u
+
v
)



=




sm

2


⁡
u

cm
⁡
v
−
cm
⁡
u


sm

2


⁡
v


sm
⁡
u


cm

2


⁡
v
−

cm

2


⁡
u

sm
⁡
v







sm
⁡
(
u
−
v
)



=



sm
⁡
u

cm
⁡
u
−
sm
⁡
v

cm
⁡
v


cm
⁡
u


cm

2


⁡
v
−

sm

2


⁡
u

sm
⁡
v









{\displaystyle {\begin{aligned}\operatorname {cm} (u+v)&={\frac {\operatorname {sm} u\,\operatorname {cm} u-\operatorname {sm} v\,\operatorname {cm} v}{\operatorname {sm} u\,\operatorname {cm} ^{2}v-\operatorname {cm} ^{2}u\,\operatorname {sm} v}}\\[8mu]\operatorname {cm} (u-v)&={\frac {\operatorname {cm} ^{2}u\,\operatorname {cm} v-\operatorname {sm} u\,\operatorname {sm} ^{2}v}{\operatorname {cm} u\,\operatorname {cm} ^{2}v-\operatorname {sm} ^{2}u\,\operatorname {sm} v}}\\[8mu]\operatorname {sm} (u+v)&={\frac {\operatorname {sm} ^{2}u\,\operatorname {cm} v-\operatorname {cm} u\,\operatorname {sm} ^{2}v}{\operatorname {sm} u\,\operatorname {cm} ^{2}v-\operatorname {cm} ^{2}u\,\operatorname {sm} v}}\\[8mu]\operatorname {sm} (u-v)&={\frac {\operatorname {sm} u\,\operatorname {cm} u-\operatorname {sm} v\,\operatorname {cm} v}{\operatorname {cm} u\,\operatorname {cm} ^{2}v-\operatorname {sm} ^{2}u\,\operatorname {sm} v}}\end{aligned}}}

These formulas can be used to compute the complex-valued functions in real components:

cm
⁡
(
x
+
ω
y
)



=



sm
⁡
x

cm
⁡
x
−
ω

sm
⁡
y

cm
⁡
y


sm
⁡
x


cm

2


⁡
y
−
ω


cm

2


⁡
x

sm
⁡
y









=



cm
⁡
x
(

sm

2


⁡
x


cm

2


⁡
y
+
cm
⁡
x


sm

2


⁡
y

cm
⁡
y
+
sm
⁡
x


cm

2


⁡
x

sm
⁡
y
)



sm

2


⁡
x


cm

4


⁡
y
+
sm
⁡
x


cm

2


⁡
x

sm
⁡
y


cm

2


⁡
y
+

cm

4


⁡
x


sm

2


⁡
y










+
ω



sm
⁡
x

sm
⁡
y
(

cm

3


⁡
x
−

cm

3


⁡
y
)



sm

2


⁡
x


cm

4


⁡
y
+
sm
⁡
x


cm

2


⁡
x

sm
⁡
y


cm

2


⁡
y
+

cm

4


⁡
x


sm

2


⁡
y







sm
⁡
(
x
+
ω
y
)



=




sm

2


⁡
x

cm
⁡
y
−

ω

2



cm
⁡
x


sm

2


⁡
y


sm
⁡
x


cm

2


⁡
y
−
ω


cm

2


⁡
x

sm
⁡
y









=



sm
⁡
x
(
sm
⁡
x

cm
⁡
x


cm

2


⁡
y
+
sm
⁡
y


cm

3


⁡
x
+
sm
⁡
y


cm

3


⁡
y
)



sm

2


⁡
x


cm

4


⁡
y
+
sm
⁡
x


cm

2


⁡
x

sm
⁡
y


cm

2


⁡
y
+

cm

4


⁡
x


sm

2


⁡
y










+
ω



sm
⁡
y
(
sm
⁡
x


cm

3


⁡
x
+
sm
⁡
x


cm

3


⁡
y
+

cm

2


⁡
x

sm
⁡
y

cm
⁡
y
)



sm

2


⁡
x


cm

4


⁡
y
+
sm
⁡
x


cm

2


⁡
x

sm
⁡
y


cm

2


⁡
y
+

cm

4


⁡
x


sm

2


⁡
y









{\displaystyle {\begin{aligned}\operatorname {cm} (x+\omega y)&={\frac {\operatorname {sm} x\,\operatorname {cm} x-\omega \,\operatorname {sm} y\,\operatorname {cm} y}{\operatorname {sm} x\,\operatorname {cm} ^{2}y-\omega \,\operatorname {cm} ^{2}x\,\operatorname {sm} y}}\\[4mu]&={\frac {\operatorname {cm} x(\operatorname {sm} ^{2}x\,\operatorname {cm} ^{2}y+\operatorname {cm} x\,\operatorname {sm} ^{2}y\,\operatorname {cm} y+\operatorname {sm} x\,\operatorname {cm} ^{2}x\,\operatorname {sm} y)}{\operatorname {sm} ^{2}x\,\operatorname {cm} ^{4}y+\operatorname {sm} x\,\operatorname {cm} ^{2}x\,\operatorname {sm} y\,\operatorname {cm} ^{2}y+\operatorname {cm} ^{4}x\,\operatorname {sm} ^{2}y}}\\[4mu]&\qquad +\omega {\frac {\operatorname {sm} x\,\operatorname {sm} y(\operatorname {cm} ^{3}x-\operatorname {cm} ^{3}y)}{\operatorname {sm} ^{2}x\,\operatorname {cm} ^{4}y+\operatorname {sm} x\,\operatorname {cm} ^{2}x\,\operatorname {sm} y\,\operatorname {cm} ^{2}y+\operatorname {cm} ^{4}x\,\operatorname {sm} ^{2}y}}\\[8mu]\operatorname {sm} (x+\omega y)&={\frac {\operatorname {sm} ^{2}x\,\operatorname {cm} y-\omega ^{2}\,\operatorname {cm} x\,\operatorname {sm} ^{2}y}{\operatorname {sm} x\,\operatorname {cm} ^{2}y-\omega \,\operatorname {cm} ^{2}x\,\operatorname {sm} y}}\\[4mu]&={\frac {\operatorname {sm} x(\operatorname {sm} x\,\operatorname {cm} x\,\operatorname {cm} ^{2}y+\operatorname {sm} y\,\operatorname {cm} ^{3}x+\operatorname {sm} y\,\operatorname {cm} ^{3}y)}{\operatorname {sm} ^{2}x\,\operatorname {cm} ^{4}y+\operatorname {sm} x\,\operatorname {cm} ^{2}x\,\operatorname {sm} y\,\operatorname {cm} ^{2}y+\operatorname {cm} ^{4}x\,\operatorname {sm} ^{2}y}}\\[4mu]&\qquad +\omega {\frac {\operatorname {sm} y(\operatorname {sm} x\,\operatorname {cm} ^{3}x+\operatorname {sm} x\,\operatorname {cm} ^{3}y+\operatorname {cm} ^{2}x\,\operatorname {sm} y\,\operatorname {cm} y)}{\operatorname {sm} ^{2}x\,\operatorname {cm} ^{4}y+\operatorname {sm} x\,\operatorname {cm} ^{2}x\,\operatorname {sm} y\,\operatorname {cm} ^{2}y+\operatorname {cm} ^{4}x\,\operatorname {sm} ^{2}y}}\end{aligned}}}

Multiple-argument identities
Argument duplication and triplication identities can be derived from the sum identity:

cm
⁡
2
u



=




cm

3


⁡
u
−

sm

3


⁡
u


cm
⁡
u
(
1
+

sm

3


⁡
u
)



=



2

cm

3


⁡
u
−
1


2
cm
⁡
u
−

cm

4


⁡
u



,




sm
⁡
2
u



=



sm
⁡
u
(
1
+

cm

3


⁡
u
)


cm
⁡
u
(
1
+

sm

3


⁡
u
)



=



2
sm
⁡
u
−

sm

4


⁡
u


2
cm
⁡
u
−

cm

4


⁡
u



,




cm
⁡
3
u



=




cm

9


⁡
u
−
6

cm

6


⁡
u
+
3

cm

3


⁡
u
+
1



cm

9


⁡
u
+
3

cm

6


⁡
u
−
6

cm

3


⁡
u
+
1



,




sm
⁡
3
u



=



3
sm
⁡
u

cm
⁡
u
(

sm

3


⁡
u


cm

3


⁡
u
−
1
)



cm

9


⁡
u
+
3

cm

6


⁡
u
−
6

cm

3


⁡
u
+
1



.






{\displaystyle {\begin{aligned}\operatorname {cm} 2u&={\frac {\operatorname {cm} ^{3}u-\operatorname {sm} ^{3}u}{\operatorname {cm} u(1+\operatorname {sm} ^{3}u)}}={\frac {2\operatorname {cm} ^{3}u-1}{2\operatorname {cm} u-\operatorname {cm} ^{4}u}},\\[5mu]\operatorname {sm} 2u&={\frac {\operatorname {sm} u(1+\operatorname {cm} ^{3}u)}{\operatorname {cm} u(1+\operatorname {sm} ^{3}u)}}={\frac {2\operatorname {sm} u-\operatorname {sm} ^{4}u}{2\operatorname {cm} u-\operatorname {cm} ^{4}u}},\\[5mu]\operatorname {cm} 3u&={\frac {\operatorname {cm} ^{9}u-6\operatorname {cm} ^{6}u+3\operatorname {cm} ^{3}u+1}{\operatorname {cm} ^{9}u+3\operatorname {cm} ^{6}u-6\operatorname {cm} ^{3}u+1}},\\[5mu]\operatorname {sm} 3u&={\frac {3\operatorname {sm} u\,\operatorname {cm} u(\operatorname {sm} ^{3}u\,\operatorname {cm} ^{3}u-1)}{\operatorname {cm} ^{9}u+3\operatorname {cm} ^{6}u-6\operatorname {cm} ^{3}u+1}}.\end{aligned}}}

Specific value identities
The



cm


{\displaystyle \operatorname {cm} }

function satisfies the identities

cm
⁡



2
9




π

3





=
−
cm
⁡



1
9




π

3



cm
⁡



4
9




π

3


,




cm
⁡



1
4




π

3





=
cl
⁡



1
3



ϖ
,






{\displaystyle {\begin{aligned}\operatorname {cm} {\tfrac {2}{9}}\pi _{3}&=-\operatorname {cm} {\tfrac {1}{9}}\pi _{3}\,\operatorname {cm} {\tfrac {4}{9}}\pi _{3},\\[5mu]\operatorname {cm} {\tfrac {1}{4}}\pi _{3}&=\operatorname {cl} {\tfrac {1}{3}}\varpi ,\end{aligned}}}

where



cl


{\displaystyle \operatorname {cl} }

is lemniscate cosine and



ϖ


{\displaystyle \varpi }

is Lemniscate constant.

Power series
The cm and sm functions can be approximated for




|

z

|

<



1
3




π

3




{\displaystyle |z|<{\tfrac {1}{3}}\pi _{3}}

by the Taylor series

cm
⁡
z



=

c

0


+

c

1



z

3


+

c

2



z

6


+

c

3



z

9


+
⋯
+

c

n



z

3
n


+
⋯




sm
⁡
z



=

s

0


z
+

s

1



z

4


+

s

2



z

7


+

s

3



z

10


+
⋯
+

s

n



z

3
n
+
1


+
⋯






{\displaystyle {\begin{aligned}\operatorname {cm} z&=c_{0}+c_{1}z^{3}+c_{2}z^{6}+c_{3}z^{9}+\cdots +c_{n}z^{3n}+\cdots \\[4mu]\operatorname {sm} z&=s_{0}z+s_{1}z^{4}+s_{2}z^{7}+s_{3}z^{10}+\cdots +s_{n}z^{3n+1}+\cdots \end{aligned}}}

whose coefficients satisfy the recurrence




c

0


=

s

0


=
1
,


{\displaystyle c_{0}=s_{0}=1,}

c

n





=
−


1

3
n




∑

k
=
0


n
−
1



s

k



s

n
−
1
−
k







s

n





=


1

3
n
+
1




∑

k
=
0


n



c

k



c

n
−
k








{\displaystyle {\begin{aligned}c_{n}&=-{\frac {1}{3n}}\sum _{k=0}^{n-1}s_{k}s_{n-1-k}\\[4mu]s_{n}&={\frac {1}{3n+1}}\sum _{k=0}^{n}c_{k}c_{n-k}\end{aligned}}}

These recurrences result in:

cm
⁡
z



=
1
−


1
3



z

3


+


1
18



z

6


−


23
2268



z

9


+


25
13608



z

12


−


619
1857492



z

15


+
⋯




sm
⁡
z



=
z
−


1
6



z

4


+


2
63



z

7


−


13
2268



z

10


+


23
22113



z

13


−


2803
14859936



z

16


+
⋯






{\displaystyle {\begin{aligned}\operatorname {cm} z&=1-{\frac {1}{3}}z^{3}+{\frac {1}{18}}z^{6}-{\frac {23}{2268}}z^{9}+{\frac {25}{13608}}z^{12}-{\frac {619}{1857492}}z^{15}+\cdots \\[8mu]\operatorname {sm} z&=z-{\frac {1}{6}}z^{4}+{\frac {2}{63}}z^{7}-{\frac {13}{2268}}z^{10}+{\frac {23}{22113}}z^{13}-{\frac {2803}{14859936}}z^{16}+\cdots \end{aligned}}}

Relation to other elliptic functions

Weierstrass elliptic function

The equianharmonic Weierstrass elliptic function



℘
(
z
)
=
℘


(


z
;
0
,



1
27





)


,


{\displaystyle \wp (z)=\wp {\bigl (}z;0,{\tfrac {1}{27}}{\bigr )},}

with lattice



Λ
=

π

3



Z

⊕

π

3


ω

Z



{\displaystyle \Lambda =\pi _{3}\mathbb {Z} \oplus \pi _{3}\omega \mathbb {Z} }

a scaling of the Eisenstein integers, can be defined as:

℘
(
z
)
=


1

z

2




+

∑

λ
∈
Λ
∖
{
0
}




(



1

(
z
−
λ

)

2





−


1

λ

2





)



{\displaystyle \wp (z)={\frac {1}{z^{2}}}+\sum _{\lambda \in \Lambda \smallsetminus \{0\}}\!\left({\frac {1}{(z-\lambda )^{2}}}-{\frac {1}{\lambda ^{2}}}\right)}

The function



℘
(
z
)


{\displaystyle \wp (z)}

solves the differential equation:

℘
′

(
z

)

2


=
4
℘
(
z

)

3


−



1
27





{\displaystyle \wp '(z)^{2}=4\wp (z)^{3}-{\tfrac {1}{27}}}

We can also write it as the inverse of the integral:

z
=

∫

∞


℘
(
z
)





d
w


4

w

3


−



1
27








{\displaystyle z=\int _{\infty }^{\wp (z)}{\frac {dw}{\sqrt {4w^{3}-{\tfrac {1}{27}}}}}}

In terms of



℘
(
z
)


{\displaystyle \wp (z)}

, the Dixon elliptic functions can be written:

cm
⁡
z
=



3

℘
′

(
z
)
+
1


3

℘
′

(
z
)
−
1



,

sm
⁡
z
=



−
6
℘
(
z
)


3

℘
′

(
z
)
−
1





{\displaystyle \operatorname {cm} z={\frac {3\wp '(z)+1}{3\wp '(z)-1}},\ \operatorname {sm} z={\frac {-6\wp (z)}{3\wp '(z)-1}}}

Likewise, the Weierstrass elliptic function



℘
(
z
)
=
℘


(


z
;
0
,



1
27





)




{\displaystyle \wp (z)=\wp {\bigl (}z;0,{\tfrac {1}{27}}{\bigr )}}

can be written in terms of Dixon elliptic functions:

℘
′

(
z
)
=



cm
⁡
z
+
1


3
(
cm
⁡
z
−
1
)



,

℘
(
z
)
=



−
sm
⁡
z


3
(
cm
⁡
z
−
1
)





{\displaystyle \wp '(z)={\frac {\operatorname {cm} z+1}{3(\operatorname {cm} z-1)}},\ \wp (z)={\frac {-\operatorname {sm} z}{3(\operatorname {cm} z-1)}}}

Jacobi elliptic functions
The Dixon elliptic functions can also be expressed using Jacobi elliptic functions, which was first observed by Cayley. Let



k
=

e

5
i
π

/

6




{\displaystyle k=e^{5i\pi /6}}

,



θ
=

3


1
4




e

5
i
π

/

12




{\displaystyle \theta =3^{\frac {1}{4}}e^{5i\pi /12}}

,



s
=
sn
⁡
(
u
,
k
)


{\displaystyle s=\operatorname {sn} (u,k)}

,



c
=
cn
⁡
(
u
,
k
)


{\displaystyle c=\operatorname {cn} (u,k)}

, and



d
=
dn
⁡
(
u
,
k
)


{\displaystyle d=\operatorname {dn} (u,k)}

. Then, let

ξ
(
u
)
=



−
1
+
θ
s
c
d


1
+
θ
s
c
d





{\displaystyle \xi (u)={\frac {-1+\theta scd}{1+\theta scd}}}

,



η
(
u
)
=




2

1

/

3



(

1
+

θ

2



s

2



)



1
+
θ
s
c
d



.


{\displaystyle \eta (u)={\frac {2^{1/3}\left(1+\theta ^{2}s^{2}\right)}{1+\theta scd}}.}

Finally, the Dixon elliptic functions are as so:

sm
⁡
(
z
)
=
ξ

(



z
+

π

3



/

6



2

1

/

3


θ



)

,


{\displaystyle \operatorname {sm} (z)=\xi \left({\frac {z+\pi _{3}/6}{2^{1/3}\theta }}\right),}





cm
⁡
(
z
)
=
η

(



z
+

π

3



/

6



2

1

/

3


θ



)

.


{\displaystyle \operatorname {cm} (z)=\eta \left({\frac {z+\pi _{3}/6}{2^{1/3}\theta }}\right).}

Generalized trigonometry
Several definitions of generalized trigonometric functions include the usual trigonometric sine and cosine as an



n
=
2


{\displaystyle n=2}

case, and the functions sm and cm as an



n
=
3


{\displaystyle n=3}

case.
For example, defining




π

n


=

B



(





1
n



,



1
n





)




{\displaystyle \pi _{n}=\mathrm {B} {\bigl (}{\tfrac {1}{n}},{\tfrac {1}{n}}{\bigr )}}

and




sin

n


⁡
z
,


cos

n


⁡
z


{\displaystyle \sin _{n}z,\,\cos _{n}z}

the inverses of an integral:

z
=

∫

0



sin

n


⁡
z





d
w


(
1
−

w

n



)

(
n
−
1
)

/

n





=

∫


cos

n


⁡
z


1





d
w


(
1
−

w

n



)

(
n
−
1
)

/

n







{\displaystyle z=\int _{0}^{\sin _{n}z}{\frac {dw}{(1-w^{n})^{(n-1)/n}}}=\int _{\cos _{n}z}^{1}{\frac {dw}{(1-w^{n})^{(n-1)/n}}}}

The area in the positive quadrant under the curve




x

n


+

y

n


=
1


{\displaystyle x^{n}+y^{n}=1}

is

∫

0


1


(
1
−

x

n



)

1

/

n




d

x
=



π

n



2
n



.


{\displaystyle \int _{0}^{1}(1-x^{n})^{1/n}\,\mathrm {d} x={\frac {\pi _{n}}{2n}}.}

The quartic



n
=
4


{\displaystyle n=4}

case results in a square lattice in the complex plane, related to the lemniscate elliptic functions.

Applications

The Dixon elliptic functions are conformal maps from an equilateral triangle to a disk, and are therefore helpful for constructing polyhedral conformal map projections involving equilateral triangles, for example projecting the sphere onto a triangle, hexagon, tetrahedron, octahedron, or icosahedron.

See also
Eisenstein integer
Elliptic function
Abel elliptic functions
Jacobi elliptic functions
Lemniscate elliptic functions
Weierstrass elliptic function
Lee conformal world in a tetrahedron
Schwarz–Christoffel mapping

Notes

References

External links
Desmos plots:
Real-valued Dixon elliptic functions https://www.desmos.com/calculator/5s4gdcnxh2.
Parametrizing the cubic Fermat curve, https://www.desmos.com/calculator/elqqf4nwas
On-Line Encyclopedia of Integer Sequences pages:
“Coefficient of x^(3n+1)/(3n+1)! in the Maclaurin expansion of the Dixon elliptic function sm(x,0).” https://oeis.org/A104133
“Coefficient of x^(3n)/(3n)! in the Maclaurin expansion of the Dixon elliptic function cm(x,0).” https://oeis.org/A104134
“Pi(3): fundamental real period of the Dixonian elliptic functions sm(z) and cm(z).” https://oeis.org/A197374
Mathematics Stack Exchange discussions:
“On




x

3


+

y

3


=

z

3




{\displaystyle x^{3}+y^{3}=z^{3}}

, the Dixonian elliptic functions, and the Borwein cubic theta functions”, https://math.stackexchange.com/q/2090523/
“doubly periodic functions as tessellations (other than parallelograms)”, https://math.stackexchange.com/q/35671/