9

Why does the following law

$$(z_1z_2)^w=z_1^wz_2^w$$

sometimes fail, such as when $z_1=z_2=-1$ and $w=-i$?

And what are the conditions I need for $z_1$ and $z_2$ so that the above identity hold for all complex values of w?

MR_BD
  • 5,942
  • 2
    Note that the exponential function is not injective. So, the above law always holds. However, if you apply it backwards, you may get multiple solutions, and the solution you may choose might not be the one you wanted. That's where proofs like 0 = 1 come from. – Kaynex Jan 01 '17 at 19:51

2 Answers2

6

The if we write $z_i = r_ie^{i\theta_i}$ and $w = x + iy$ then we have $$(z_1z_2)^w = e^{\ln(r_1r_2)x}e^{i\ln(r_1r_2)y}e^{i(\theta_1+\theta_2) x}e^{-y(\theta_1 + \theta_2)}$$

The same formal identity holds for $z_1^wz_2^w.$ The reason the two calculations come out different is that when calculating $(z_1z_2)^w,$ you reduce $\theta_1+\theta_2$ mod $2\pi$ after you multiply $z_1$ and $z_2$, but before you exponentiate. Thus the problematic factor is $e^{-y(\theta_1+\theta_2)}$ which will be a factor of $e^{2\pi y}$ different between the two calculations if such a reduction takes place. Therefore the two calculations come out the same whenever $\theta_1 + \theta_2 < 2\pi$.

That's not the case in your example, since for $z_1 = z_2 = -1,$ $\theta_1 = \theta_2 = \pi$ so you reduce $\theta_1 + \theta_2 = 2\pi$ to $0$ in one of the calculations. Thus the answers are different by a factor of $e^{2\pi}.$

1

Note first that when $w=0$, then clearly the equality holds. So let $w \neq 0$. Then $$(z_1z_2)^w = z_1^wz_2^w \iff \exp(w \ln(z_1z_2)) = \exp(w \ln(z_1))\exp(w \ln(z_2)) \iff \exp(w \ln(z_1z_2)) = \exp(w(\ln(z_1)+\ln(z_2))) \iff w \ln(z_1z_2) = w(\ln(z_1)+\ln(z_2)) \iff \ln(z_1z_2) = \ln(z_1)+\ln(z_2)$$ Hence we found the required condition on $z_1$ and $z_2$ such that the equality holds, namely $$\ln(z_1z_2) = \ln(z_1)+\ln(z_2).$$

ViktorStein
  • 4,838
Schief
  • 319