commentary

Fractal optics and beyond

Mordechai Segev, Marin Soljai and John M. Dudley
Fractals, shapes comprised of self-similar parts, are not merely prescribed linear structures. A wide class
of fractals can also arise from the rich dynamics inherent to nonlinear optics.

RACHEL WON

n 1967, Benoit Mandelbrot published a

paper that gave birth to the study of fractals,
entitled How long is the coast of Britain?
Statistical self similarity and fractional
dimension1. According to Mandelbrot, a
fractal is a shape made of parts similar to
the whole in some way2. One particularly
spectacular example of a fractal in nature is
the Romanescu broccoli (Fig.1).
Although there are a number of different
fractal classification systems, one stands
out rather distinctly: exact (regular) fractals
versus statistical (random) fractals. An exact
fractal is an object which appears self-similar
under varying degrees of magnification, in
effect, possessing symmetry across scale, with
each small part replicating the structure of
the whole. Taken literally, when the same
object replicates itself on successively smaller
scales, even though the number of scales in
the physical world is never infinite, we call
this object an exact fractal. When, on the
other hand, the object replicates itself in only
its statistical properties, it is defined as a
statistical fractal.
Perhaps the best known example of
an exact fractal is the Cantor set fractal, a
shape that can be explained by describing
its generation. Starting with a single line
segment, the middle third is removed to leave
behind two segments, each with a length
of one-third of the original. From each of
these segments, the middle third is again
removed, and so on, ad infinitum. At every
stage of the process, the result is self-similar
to the previous stage, which is identical upon
rescaling. Of course, this triplet set is not
the only possible Cantor set. Any arbitrary
cascaded removal of portions of the line
segment may form the repetitive structure of
an exact fractal. Other famous examples of
exact fractals are the Sierpinski triangle and
the Koch snowflake.
Statistical fractals have been observed in
many physical systems, ranging from material
structures such as polymers, aggregation
and interfaces, through to biology, medicine,
electric circuits, computer interconnects,
galactic clusters and stock market price
fluctuations3. Exact fractals, on the other
hand, seem merely to be mathematical

Figure 1 | Many examples of fractal structures, such

as the Romanescu broccoli displayed here, can be
found in nature. Experiments are now showing how
self-similarity and fractals can be observed in both
linear and nonlinear optical systems.

constructs, and it is not at all apparent that

they exist in nature.
In 1993 Mandelbrot won the prestigious
Wolf Prize for the widespread occurrence of
fractals and developing mathematical tools
for describing them, which has changed our
view of nature4. This prize was presented
in the field of physics, not mathematics. It
is therefore fair to ask the question: has our
knowledge of fractals really changed the way
we view nature? The beauty of self-similarity
is clearly present in exact fractals. But can
exact fractals be found in physical systems?
Mandelbrot presented many fascinating
fractal constructs that look exactly like
pictures taken by a wonderful photographer,
or like scenes from a fantasy movie. The
very fact that mathematical constructs
can be structured in such artistic ways is
undoubtedly very pretty. But do exact fractals
have any true significance in physics?
The field of optics was one of the first
to demonstrate exact fractals in a physical
reality. Various optical settings are known
to generate fractals in space. For example, in
1979 Michael Berry showed that diffraction
from fractal structures can give rise to a
fractal diffraction pattern5. In a similar
fashion, fractal patterns emerge in many
other optical settings, ranging from the lasing

NATURE PHOTONICS | VOL 6 | APRIL 2012 | www.nature.com/naturephotonics

modes of unstable resonators6 through to

light diffracting from a binary grating 7.
Omel Mendoza-Yero and colleagues8
recently demonstrated an interesting
application of fractal optics by using an
intriguingly simple method for forming
temporal fractals from self-similar spatial
structures. Their results highlight the
interplay between time and space that is so
inherent to coherent electromagnetic fields,
and also suggest possible applications of
fractal control in the important fields of
arbitrary waveform generation and pulse
shaping. However, one should note that this is
a linear system, in which the entire evolution
is determined by the initial conditions
hence the results are not really surprising
from a fundamental viewpoint. In this sense,
what would be surprising and indeed much
more unexpected is a nonlinear dynamical
system in which the entire dynamics behaves
as an exact fractal.
This kind of nonlinear thinking takes
us back to early work on self-similarity and
fractals in nonlinear optics in the 1980s.
The first suggestion that fractals would
naturally emerge in nonlinear optical
systems was made by Sergei Manakov and
Ildar Gabitov, who studied light propagation
in an inverted two-level system9,10. A decade
later, Sunghyuck An and John Sipe suggested
that the evolution of Hill gratings is selfsimilar 11, and Curtis Menyuk, Decio Levi
and Pavel Winternitz proposed that the
transient regime of stimulated Raman
scattering gives rise to self-similar pulses12. In
all of these cases, there were hints from the
experiments that self-similarity dominates
the long-term behaviour.
Around a decade later, two of us
(Mordechai Segev and Marin Soljai, then at
Princeton), together with Menyuk, made the
connection between solitons and fractals13.
Solitons are self-trapped wavepackets that
interact with one another in a manner similar
to the way that particles do. In some cases,
solitons are self-similar; that is, their shape
is universal and their amplitude scales with
their width. This happens, for example, for
temporal solitons in optical fibres and for
one-dimensional spatial solitons, both in
209

commentary
Kerr-type nonlinear media. Our idea was
that nonlinear soliton-supporting systems
could evolve under non-adiabatic conditions
to give rise to self-similarity and fractals.
Such fractals were believed to be observable
in many systems, in which their existence
would depend on two requirements13. First,
the system should possess neither a natural
length scale (that is, the physics should be
the same on all scales) nor a natural scale in
the parameter range of interest. Second, the
system should undergo abrupt, non-adiabatic
changes in at least one of its properties. Of
course, the first requirement could physically
occur only over a finite range, but that range
should yield at least several generations of the
self-similar structure.
Immediately after the connection between
solitons and fractals was made13, we, together
with our co-workers at Princeton, proposed
experimental systems in which Cantor set
soliton fractals might be observable14. The
goal was to design a system for generating the
first exact fractals in nonlinear optics, perhaps
even the first man-made exact fractals in
nature. The idea involved launching a highpower pulse into a sequence of pre-designed
dispersion-managed fibres14. The pulse would
break up into daughter solitons that are
self-similar to one another in the sense that
they can be mapped (by a change of scale
only) onto one another because they all have
the same shape. If, however, an adequately
abrupt change is made to a property of the
medium such as the dispersion coefficient,
each of the daughter solitons would undergo
the same breakup experienced by the initial
mother pulse and generate even smaller
granddaughter solitons. Successive changes to
the medium properties thus create successive
generations of solitons on successively smaller
scales (Fig.2).The resulting structure after
every breakup is self-similar with the products
of the first breakup. Successive generations of
breakups lead to a structure that is self-similar
on widely varying scales, and each part
breaks up again in a structure that replicates
the whole. The entire structure is therefore
a fractal comprised of solitons14. What is
intriguing in this scheme is that the entire
dynamics is self-similar; the entire dynamic
process of pulse breakup repeats successively
on a smaller and smaller scale, forming a
dynamic nonlinear fractal. This concept
of linking fractal behaviour with soliton
dynamics attracted significant attention from
the wider physics community. Important early
results obtained by Jianke Yang and Yu Tan
from the University of Vermont demonstrated
that collisions between specific types of
solitons give rise to a fractal structure15.
The idea of generating soliton fractals was
so appealing that we (Segev and Soljai)
spent quite some time experimenting with
210

Figure2 | Nonlinear Cantor set generation was

first proposed from self-similar patterns of
soliton-splitting.

wavepackets in space and in time, although

generating exact fractals in nonlinear optics
remained a challenge. Success was found
8,707miles away in Auckland, New Zealand,
where the third among us (John M. Dudley),
together with his co-workers, made the first
successful demonstration of self-similar
scaling in a fibre amplifier. This development
facilitated the design and interpretation of
experiments that are now considered to be the
first example of fractal scaling in nonlinear
optics16. Subsequent experiments reported
self-similar evolution in a wider range of
fibre systems, and important work led by
Frank Wise at Cornell University in the USA
showed how self-similarity can be combined
with optical feedback to create a self-similar
laser 17. More recent research by mer Ilday
and co-workers at Bilkent University in
Turkey led to the development of a laser cavity
that combines two very different classes of
nonlinear dynamics: self-similar propagation
in one branch and soliton propagation in
another 18. The resilience of this laser to noise
(due to the attractive nature of the nonlinear
propagation in both branches) provides
greatly improved stability and represents a
beautiful example of how exploiting nonlinear
dynamics can yield new technologies that are
not possible in strictly linear systems.
The study of self-similarity and fractal
dynamics has now become an important area
of research in nonlinear fibre optics. New
classes of fibre and improved measurement
techniques are now enabling experiments to
test the early ideas of this field. For instance,
the development of gas-filled photonic crystal
fibre has finally allowed theoretical work
on self-similarity in Raman scattering 12 to
be confirmed experimentally 19. Even more
recently, studies into the higher-order regimes
of nonlinear modulation instability have
revealed cascaded pulse splitting 20 of a type
very close to that of the original prediction
in ref.14for the Cantor set fractal. Soliton
fractals have also been observed in systems
beyond optics. The first of these was the

work of Mingzhong Wu, Boris Kalinikos

and Carl Patton at Colorado State University
in collaboration with Lincoln Carr at the
Colorado School of Mines, USA, who
demonstrated exact fractals with spin-wave
solitons in magnetic films21. Soliton fractals
are now also being explored throughout
various systems in nature.
We are now 45years after Mandelbrot
published his highly influential paper and
almost 20years after his Wolf Prize. The
progress made in nonlinear optics over the
past decade adds an important flavour to
fractals because it proves that fractals
even exact fractals can emerge naturally
by virtue of interactions; that is, the
nonlinear interplay between systems that is
manifested in the plethora of lightmatter
interactions underlying nonlinear optics.
There must be many other natural nonlinear
systems in which fractals evolve naturally
with self-similar nonlinear dynamics. In this
respect, the statement made during the Wolf
Prize ceremony was not only correct but
visionary. Fractals are not merely prescribed
linear structures; rather, nonlinear systems
can give rise to dynamic fractals in which
the entire evolution is self-similar on many
scales. We are confident that the beauty of
self-similarity has many more features yet to
be discovered.

Mordechai Segev is in the Physics Department at

Technion Israel Institute of Technology, Haifa 32000,
Israel. Marin Soljai is in the Physics Department at
the Massachusetts Institute of Technology, Cambridge,
Massachusetts 02139, USA. JohnM.Dudley is at
Institut FEMTO-ST, UMR 6174 CNRS Universit de
Franche-Comt, 25000 Besancon, France.
e-mail: [email protected]
References

1. Mandelbrot, B. Science 156, 636638 (1967).

2. Mandelbrot, B.B. Enc. Phys. Sci. Tech. 5, 579593 (1987).
3. Addison, P.S. Fractals and Chaos: An illustrated course
(IOP, 1997).
4. http://www.wolffund.org.il/full.asp?id=116
5. Berry, M.V.J.Phys. A 12, 781797 (1979).
6. Karman, G.P., Mcdonald, G.S., New, G.H.C. & Woerdman, J.P.
Nature 402, 138 (1999).
7. Berry, M.V. & Klein, S.J.Mod. Opt. 43, 21392164 (1996).
8. Mendoza-Yero, O.etal. Opt. Lett. Doc. ID:156405 (2012).
9. Manakov, S.V. Sov. Phys. JETP 56, 3744 (1982).
10. Gabitov, I.R. & Manakov, S.V. Phys. Rev. Lett. 50, 495498 (1983).
11. An, S. & Sipe, J.E. Opt. Lett. 16, 14781480 (1991).
12. Menyuk, C.R., Levi, D. & Winternitz, P. Phys. Rev. Lett. 69,
30483051 (1992).
13. Soljacic, M., Segev, M. & Menyuk, C.R. Phys. Rev. E61,
R1048 (2000).
14. Sears, S.etal. Phys. Rev. Lett.84, 19021905 (2000).
15. Yang, J. & Tan, Y. Phys. Rev. Lett. 85, 36243267 (2000).
16. Fermann, M.E.etal. Phys. Rev. Lett. 84, 60106013 (2000).
17. Ilday, F.., Buckley, J.R., Clark, W.G. & Wise, F.W.
Phys. Rev. Lett. 92, 213902 (2004).
18. Oktem, B., lgdr, C. & Ilday, F.. Nature Photon. 4,
307311 (2010).
19. Nazarkin, A., Abdolvand, A., Chugreev, A.V. & Russell, P. St. J.
Phys. Rev. Lett. 105, 173902 (2010).
20. Erkintalo, M.etal. Phys. Rev. Lett. 107, 253901 (2011).
21. Wu, M., Kalinikos, B.A., Carr, L.D. & Patton, C.E.
Phys. Rev. Lett. 96, 187202 (2006).

NATURE PHOTONICS | VOL 6 | APRIL 2012 | www.nature.com/naturephotonics