The Borwein cubic theta functions $a(q),b(q),c(q)$ and the doubly-periodic Dixonian elliptic functions $\mathrm{sm}(z), \mathrm{cm}(z)$ parameterize the Fermat cubic $x^3+y^3=1$, so

$$\frac{b^3(q)}{a^3(q)}+\frac{c^3(q)}{a^3(q)} =1\tag1$$


This post asks if there is any relation between the Borwein and Dixonian versions. It is known that the former is associated with $_2F_1\big(\tfrac13,\tfrac23;\color{blue}1;u\big)$ while the latter is with $_2F_1\big(\tfrac13,\tfrac23;\color{blue}{\tfrac43};v\big)$.

For argument $q=e^{2\pi i \sqrt{-n}}$, we have the nice relationship,

$$\large \frac{_2F_1\Big(\tfrac13,\tfrac23;1;\tfrac{b^3(q)}{a^3(q)} \Big)}{_2F_1\Big(\tfrac13,\tfrac23;1;\tfrac{c^3(q)}{a^3(q)} \Big)}=\sqrt{3n}$$

Q: By analogy, does the ratio, $$\frac{_2F_1\Big(\tfrac13,\tfrac23;\tfrac43;\mathrm{sm}^3(z) \Big)}{_2F_1\Big(\tfrac13,\tfrac23;\tfrac43;\mathrm{cm}^3(z) \Big)}=\,??\tag3$$ evaluate to anything "meaningful" for appropriate choice of argument $z$?

P.S. To test this conjecture, for the circle $x^2+y^2=1$, we have the Dedekind eta quotients $\alpha(q), \beta(q)$ parameterization (given here) and the singly-periodic trigonometric functions,

$$\alpha^8(q)+\beta^8(q)=1\tag4$$ $$\sin^2 z+\cos^2 z=1\tag5$$

The analogous ratios, $$\frac{_2F_1\Big(\tfrac12,\tfrac12;1; \beta^8(q) \Big)}{ _2F_1\Big(\tfrac12,\tfrac12;1; \alpha^8(q)\Big)}=\sqrt{4n}$$ and choice of $0<z<1$,

$$\frac{_2F_1\Big(\tfrac12,\tfrac12;\tfrac32; \cos^2 z \Big)}{ _2F_1\Big(\tfrac12,\tfrac12;\tfrac32; \sin^2 z \Big)}=\Big(\frac{\pi}{2z}-1\Big) \tan z$$ so $(3)$ might indeed evaluate to some closed-form, perhaps involving the fundamental constant of Dixonian functions $\large \pi_3=\frac{\sqrt3}{2\pi}\Gamma^3\big(\tfrac13\big)$.

  • $\begingroup$ As soon as I saw the title of this question, I thought that you were probably the author. I admire the work you do here. $\endgroup$ – marty cohen Jan 9 '17 at 17:19
  • $\begingroup$ @martycohen: Thanks, Marty. :) $\endgroup$ – Tito Piezas III Jan 9 '17 at 17:35
  • $\begingroup$ After a good night's rest, I figured out the answer. See this MO post for a possible generalization $\endgroup$ – Tito Piezas III Jan 10 '17 at 5:36

Recall that

$$\int_0^z\frac{\mathrm du}{\left(1-u^3\right)^{2/3}}=\operatorname{sm}^{(-1)}(z)=z\;{}_2 F_1\left({{\frac13,\frac23}\atop{\frac43}}\middle|z^3\right)$$


$$\int_z^1\frac{\mathrm du}{\left(1-u^3\right)^{2/3}}=\operatorname{cm}^{(-1)}(z)$$

along with

$$\frac{\pi_3}{3}=\int_0^1\frac{\mathrm du}{\left(1-u^3\right)^{2/3}}=\frac13 B\left(\frac13,\frac13\right)$$

so within a limited domain, your expression ought to simplify to

$$\frac{{}_2 F_1\left({{\frac13,\frac23}\atop{\frac43}}\middle|\operatorname{sm}^3(z)\right)}{{}_2 F_1\left({{\frac13,\frac23}\atop{\frac43}}\middle|\operatorname{cm}^3(z)\right)}=\frac{\frac{z}{\operatorname{sm}(z)}}{\frac{\frac{\pi_3}{3}-z}{\operatorname{cm}(z)}}=\frac{z}{\frac{\pi_3}{3}-z}\frac{\operatorname{cm}(z)}{\operatorname{sm}(z)}$$

Note the phrase "limited domain". This is expected, in the same way that $\arcsin\sin z=z$ and $\arccos\cos z=z$ are also only valid within a limited domain. In particular, the relation only works within the intersection of the domains of validity of $\operatorname{sm}^{(-1)}(\operatorname{sm}(z))$ and $\operatorname{cm}^{(-1)}(\operatorname{cm}(z))$ (both of which are, perhaps not surprisingly, hexagons), which is the rhombus with corners at $0,\frac{\pi_3}{3}\exp\left(\frac{\pi i}{3}\right),\frac{\pi_3}{3},\frac{\pi_3}{3}\exp\left(-\frac{\pi i}{3}\right)$.

  • $\begingroup$ @JM: Yes, I believe it is $0<z<1$. This morning I posted such relations in MO for $\sin, \mathrm{sn}, \mathrm{sm}$ and asked if it could be generalized from $a=1/2$ and $a=2/3$, to $a=1/4$ and $a=1/6$. $\endgroup$ – Tito Piezas III Jan 10 '17 at 16:37
  • $\begingroup$ @JM: Perhaps you can answer that MO post as well for $a=1/4$? :) $\endgroup$ – Tito Piezas III Jan 10 '17 at 16:52
  • $\begingroup$ The cases $a=1/4,1/6$ have been posted here mathoverflow.net/a/259267/82588 $\endgroup$ – Nemo Jan 10 '17 at 22:20
  • $\begingroup$ @Tito, $0<z<1$ is but a small cut of the rhombus I mentioned in my answer. :) $\endgroup$ – J. M. is a poor mathematician Jan 12 '17 at 15:42
  • $\begingroup$ @J.M.: And someone also did work on $x^5+y^5=1$. $\endgroup$ – Tito Piezas III Jan 12 '17 at 15:49

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.