1
$\begingroup$

I have a Jacobian elliptic function as a function of two independent variables $x$ and $y$. The elliptic parameter $m=m(y)$, $0 \leq m \leq 1$, is also a function of the variable $y$, and thus the elliptic nome $q=q(m)=q(m(y))$ also depends on $y$.

For example, I define:

$$h(y) = \ln \left(\frac{5+\sin y}{2- \cos y}\right)$$

and take $q$ such that $q = -\frac{\pi^{2}}{h(y)}$. My elliptic function is

$$f(x,y)=sn\left(\frac{2K(m)x}{h(y)},m\right),$$

where $sn$ is the Jacobian elliptic function and $K(m)=K(m(y))$ is the complete elliptic integral of the first kind, and $x \in \mathbb{R}$.

What I want is to rewrite $f(x,y)$ in terms of $x$ and $m$, instead of $y$, (and suitably normalise it) so that I could then make a contour plot of $f(x,m)$ for some level set of $f$. The problem is that I don't know how to get $Mathematica$ to invert the elliptic nome so that $y$ is expressed in terms of $m$.

My code is:

h[y_] := Log[(5 + Sin[y])/(2 - Cos[y])]

f[x_, y_] := (JacobiSN[2*EllipticK[InverseEllipticNomeQ[Exp[-(Pi)^2/h[y]]]]*x/h[y], InverseEllipticNomeQ[Exp[-(Pi)^2/h[y]]]])

ContourPlot[f[x, y] == 0, {x, 0, 10}, {y, 0, 2*Pi}]

But this is not what I need.

Edit. A bit more info:

When I run the code I wrote above, the contour plot gives me a whole $\textit{set}$ of level curves of $f$, because as it stands, for any fixed $y$, the period of $f$ depends on $y$ (since the period of elliptic functions is expressed through $K(m)$, and $y$ runs through the interval $[0, 2\pi]$.

Fixing $y = y^{*}$ for a few random values $y^{*}$ reveals that the the graph of $f(x, y^{*})$ has just one root in its appropriate period determined by $h(y)$.

What I want is to show that this is true for all set of values $0 \leq y \leq 2\pi$, from looking at the contour plot. As it stands, this is not clear. However, my expectation is that if I normalise $f$ properly, and use $m$ instead of $y$, as the independent variable with $ 0\leq m \leq 1$, the contour plot should give me just one level curve, which proves the above.

$\endgroup$
  • $\begingroup$ If I solve the equation $w=h(y)$ for $y$, I get a complicated expression with a choice of branches for the square root and the arctangent in the result. You might want to think about which branch is most appropriate for what you want to do. $\endgroup$ – J. M. will be back soon Jun 7 '16 at 16:07
  • $\begingroup$ The $h(y)$ above is just a hypothetical example, my real $h$ is even more complicated. What I want is $0 \leq y \leq 2\pi$. So you suggest that I start by resolving the expression for $y$ first (say using the Inverse Function commmand)? @J.M. $\endgroup$ – Alex Jun 7 '16 at 16:14
  • $\begingroup$ "my real $h$ is even more complicated" tells me that you'll have to resort to a numerical method (i.e. FindRoot[]) in the worst case; as you know, most transcendental equations do not lend themselves to simple solutions. But wait, if $h(y)=-\frac{\pi^2}{q(m)}$, why not make that replacement everywhere you can? $\endgroup$ – J. M. will be back soon Jun 7 '16 at 16:16
  • $\begingroup$ Is this of some help Jacobi Elliptic Functions $\endgroup$ – user9660 Jun 7 '16 at 16:28
  • 1
    $\begingroup$ Another thought: I guess this essentially amounts to ensuring that "InverseEllipticNomeQ[Exp[-(Pi)^2/h[y]]]" is always in the interval [0,1], right? @J.M. $\endgroup$ – Alex Jun 7 '16 at 17:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.