The complete elliptic integral of the first kind $K(x)$ has a branch cut on the real axis for $|x|>0$.

One has: \begin{align} &1.) \qquad \lim_{\eta \to 0^-}K(x+i\eta)=K(x) &\text{ with } x,\eta\in \mathbb{R} \text{ and } x>1 \\ &2.) \qquad \lim_{\eta \to 0^+}K(x+i\eta)=K(x)+2i K(1-x) &\text{ with } x,\eta\in \mathbb{R} \text{ and } x>1 \end{align}

Therefore, for an evaluation on the branch cut, I have to chose an $i \eta$ prescription. My numerical tests suggest, that Mathematica has the default prescription $1.)$, but I want to be absolutely certain.

Can someone please point me to a page of Mathematica or something else, which confirms that Mathematica has a default prescription and that it is $1.)$.

On a side note: I am aware of the fact, that I could use functional identities of $K$ to map the evaluation to arguments to $|x|<1$ where everything is nice and cosy, but I would rather avoid them for some specific reasons.

  • 1
    $\begingroup$ The Wolfram Functions website describes the branch cut of the complete elliptic integral as you do. I have always assumed that the conventions are the same between the Wolfram Functions website and the Mathematica software, but I admittedly do not know for sure. $\endgroup$ – Michael Seifert May 4 '17 at 18:25
  • 2
    $\begingroup$ Also, a "brunch cut" (line below the equations) sounds pretty tasty. :-) $\endgroup$ – Michael Seifert May 4 '17 at 18:27
  • $\begingroup$ @MichaelSeifert Thx! $\endgroup$ – Armin May 4 '17 at 18:29
  • $\begingroup$ @MichaelSeifert But how do I know, what value it gives me if I evaluate e.g. EllipticK[2]? Does it evaluate ``K[2]+2 I K[-1]" or "K[2]"? $\endgroup$ – Armin May 4 '17 at 18:32
  • $\begingroup$ @MichaelSeifert I do understand now, what you meant and that you answered my question already... Sry, for me being so slow... $\endgroup$ – Armin May 5 '17 at 7:04

Taking the directional limits will reveal Mathematica's conventions:

Assuming[x > 1, Limit[EllipticK[x + I ε], ε -> 0, Direction -> 1]]


Assuming[x > 1, Limit[EllipticK[x + I ε], ε -> 0, Direction -> -1]]

2 I EllipticK[1 - x] + EllipticK[x]

| improve this answer | |
  • $\begingroup$ many thanks. good idea! $\endgroup$ – Armin May 5 '17 at 6:12

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.