Octonionic Julia sets (Q1193660)

Besides the exhaustive studies that have been made, of quadratic mappings on the real and complex fields, there has been some recent work with quaternions. Now the authors work in the 8-dimensional real algebra of octonions, and study the mapping \[ R^ a_ z:z \to z^ 2+c+u(a,c,z), \] where \(c\) and \(a\) are fixed, and \(u=c(za)-(cz)a\). Thus \(u\) is a measure of non-associativity, and when \(a=0\) the mapping behaves like the usual quadratic mapping of the quaternions. Therefore, assuming \(a \neq 0\), the authors wish to investigate the Julia set \(J\). By suitale normalizations they express \(R^ a_ z\) as a 3-parameter mapping of \(\mathbb{R}^ 8\), with \(c\) represented by \((c_ 1, c_ 2) \in \mathbb{R}^ 2\) and \(a\) by \(\alpha \in \mathbb{R}\). Fixing \((c_ 1,c_ 2)\) as (--0.745, 0.113) and concentrating on the baisin of attraction of \(\infty\), they compute certain two-dimensional sections of \(J\) for suitably selected \(\alpha\), and their graphics include typical `sea horse' structures. They observe that `for large \(\alpha\), a concentrated region of convergence emerges which also dies away as \(\alpha\) is increased', and so define a function \(f(S, \alpha)\) to measure the `total divergence' of a region \(S\). They plot this function for some regions, with \(0\leq\alpha\leq 3\), and find that there is always a `spike' in the graph near \(\alpha=2\) which `visually' resembles a condensation process. This empirical work therefore reveals an intriguing phenomenon, but the authors offer no theoretical explanation.