1

All books that i am reading are telling that the identiy $ a^x = a^y \Longrightarrow x=y ; \,\,\,\,\, a \in \mathbb{R} - \{0,1\} $ $ \,\,\,\,$ also do not work for $a<0$, but, for example, if $ (-2)^2 = (-2)^x $, than for sure $x=2$.

Than, the identity works for $a<0$, correct?

Voyager
  • 1,042

2 Answers2

3

It does not work for $a\lt 0$.

If $$(-1)^x=(-1)^2,$$ then $x=4$ is another solution.

In general, $$(-1)^x=(-1)^2\Rightarrow x=2m (m\in\mathbb Z)$$

mathlove
  • 139,939
3

Note that $a^b := \exp(b \ln a)$ and even the definition of $\ln a$ is problematic for arbitrary $a\in\mathbb C$ and the complex exponential isn't bijective (unlike the real). The principal branch of the natural logarithm is not defined for $a\in\mathbb R_-\cup\{0\}$.

AlexR
  • 24,905
  • 1
    The principal branch of $\ln$ is defined for $a < 0$, it is $\ln|a| + i\pi$, but it is not continous across the branch cut. – gammatester Jan 20 '14 at 11:08
  • @gammatester this is a matter of convention. If you leave it undefined, it is analytic on its domain; if you do define it, it's no longer analytic, making things a lot harder. – AlexR Jan 20 '14 at 11:09