Section5.2The quadratic escape criterion
Recall the escape radius stated in the escape criterion for general polynomials - namely
\begin{equation}
R = \max\left(2|a_n|,2\frac{|a_{n-1}|+\cdots+|a_0|}{|a_n|}\right).
\tag{5.2.1}
\end{equation}
We can apply this directly to the quadratic family. It turns out, though, that we can do a bit better by focusing on this particular family.
Theorem5.2.1The quadratic escape criterion
Let
\begin{equation*}
f_c(z) = z^2 + c.
\end{equation*}
Then, the orbit of \(z_0\) diverges to \(\infty\) whenever \(|z_0|\) exceeds
\begin{equation*}
R = \max(2,|c|).
\end{equation*}
The number \(R\) is called the escape radius for the quadratic.
Proof
Suppose that \(z_0\in\mathbb C\) satisfies \(|z_0| \gt 2\) and \(|z_0| \geq |c|\text{.}\) Then, by the reverse triangle inequality, \begin{align*} |z_1| = |f(z_0)| &= |z_0^2 + c| \geq |z_0|^2 - |c| \\ & \geq |z_0|^2 - |z_0| = |z_0|\left(|z_0|-1\right) = \lambda |z_0|, \end{align*} where \(\lambda \gt 1\text{.}\) Furthermore, \(z_1\) now also satisfies \(z_1 \gt R\) so that, by induction, \(|z_n| \geq \lambda^n |z_0|\text{.}\) As a result, \(z_n\to\infty\) as \(n\to\infty\text{.}\)
As with the polynomial escape criterion, this applies to any iterate \(z_i\) even if \(z_0\leq 2\text{.}\) As a result, once some iterate exceeds \(2\) in absolute value, the orbit is guaranteed to escape. Note, though, that the quadratic escape criterion is a bit sharper than the version for general polynomials, when applied to the quadratic case. For example, theorem TheoremĀ 1 implies that the escape radius for \(f_{-2}(z)=z^2-2\) is \(2\text{,}\) while equation (5.2.1) yields an escape radius of \(4\text{.}\) As we already know that the Julia set of \(f_{-2}\) is exactly the interval \([-2,2]\text{,}\) this estimate is as sharp as it can be. Furthermore, under the assumption that \(|c|\leq 2\text{,}\) the quadratic escape radius yields a uniform radius of \(2\text{.}\)