(i) Linear fractals.One of the earliestreports of linear
optical fractals is diffractals(plane waves scattered by nonregular/fractal objects) [1]. Although it is intuitive that light
diffracted by complex gratings might acquire complex
structure, even a simple (regular) square-wave grating can also
produce fractal light patterns through repeated self-imaging of
the grating itself (the Talbot effect) [2]. Self-imaging is a
property of many linear opticalsystems. For instance, the
transverse empty-cavity modes of classic unstable strip
resonators have fractal character [3], where the eigenvalue
problem (a criterion for self-reproducing mode profiles)
involves an interplay between small-scale diffraction effects at
the mirror edges and successive round-trip magnifications [4].
Mode fractality was later confirmed in so-called kaleidoscope
lasers, that include non-trivial transverse boundary conditions
[5,6]. Alternative schemes for optical self-imaging, such as
multiple-reduction copiers and pixellated video feedback
setups [7], have provided further (potentially linear) contexts
for spatial fractal formation;
(ii) Soliton fractals. A range of fractal patterns in solitonsupporting systems has been identified over the last two
decades. The existence of these patterns is directly related to
nonlinear light–material coupling. Self-similarity has been
predicted during the amplification of parabolic pulses in optical
fibres [8] and also in the distributions of soliton profiles in
systems with series of abrupt material discontinuities (that can
induce individual new scale lengths through splitting
phenomena) [9];
(iii) Nonlinear phase-space fractals.Fractals can appear in
the parameter characterization of nonlinear optical phenomena
(while their real-space and time representations remain nonfractal). Examples include bifurcations in the phase-space of
chaotic pixel-pixel mappings in optical memory applications
[10] and in the properties of interacting vector solitons [11];
(iv) Spontaneous nonlinear spatial fractals.Finiteamplitude simple universal patterns (e.g., stripes, squares,
hexagons, honeycombs, etc.) may grow spontaneously from the
homogeneous states of a reaction-diffusion system that is
sufficiently stressed. Turing showed that the origin of simplepattern emergence is the existence of a singlethreshold
instability minimum whose characteristics dictate the dominant
scale length of the pattern [12]. More recently, we proposed
that any system whose threshold instability spectrum comprises
a hierarchy of comparable Turing minima may be susceptible
to truly spontaneous fractal pattern formation. The first
prediction of such patterns was made for a simple system: the
Kerr slice with a single feedback mirror [13,14]. Our
subsequent analyses of dispersive and absorptive ring cavities
[15] have offered further evidence that multi-Turing threshold
minima can be a generic signatureof a system’s innate fractalgenerating capacity.