1
$\begingroup$

Sometimes when I try to use Mathematica to evaluate some integrals depending on some parameter $\alpha\in\mathbb{C}$ the result ends up involving $(-1)^{\alpha}$. Now this is ambiguous if $\alpha\notin\mathbb{Z}$. In fact it could be either $e^{\pm i\pi\alpha}$ and indeed the two can be distinct $$e^{i\pi\alpha}=e^{2\pi i\alpha}e^{-i\pi\alpha},$$

and when $\alpha\notin\mathbb{Z}$ the factor $e^{2\pi i\alpha}\neq 1$. All that said, if Mathematica outputs the result as $(-1)^{\alpha}$ I don't really know which of the two possibilities it is supposed to be.

How can I make Mathematica write $(-1)^\alpha$ correctly for general $\alpha$ when evaluating an integral, taking branch cuts into consideration?

$\endgroup$

1 Answer 1

5
$\begingroup$

It's always interpreted by Mathematica as $e^{i \pi \alpha}$:

Reduce[(-1)^x == Exp[I \[Pi] x], x]
(* True *)

Reduce[(-1)^x == Exp[-I \[Pi] x], x]
(* C[1] \[Element] Integers && x == C[1] *)

According to http://reference.wolfram.com/language/ref/Power.html

For complex numbers $x$ and $y$, Power gives the principal value of $e^{y \log (x)}$.

$\endgroup$
2
  • $\begingroup$ So the result is $e^{i\pi\alpha }$ it is going to write $(-1)^\alpha$ and if the results s $e^{-i\pi\alpha}$ it is going to write $(-1)^{-\alpha}$? That's how it makes a difference between the two? $\endgroup$ Commented Aug 28, 2023 at 13:40
  • $\begingroup$ Yes, that's how it makes a difference between the two. Try ReImPlot[(-1)^x, {x,-3,3}] to see it. $\endgroup$ Commented Aug 28, 2023 at 15:16

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.