The Influence of
Benoît B. Mandelbrot
on Mathematics
Edited by Michael F. Barnsley and Michael Frame
Michael F. Barnsley
Introduction
We begin this article, which deals largely with
Benoît B. Mandelbrot’s contributions to and influ-
ence upon mathematics, with a quotation from
the introduction to Fractals: Form, Chance, and
Dimension [16]
. This essay, together with many
pictures and numerous lectures in the same vein,
changed the way science looks at nature and had
a significant impact on mathematics. It is easy for
us now to think that what he says is obvious; it
was not.
Many important spatial patterns of Nature
are either irregular or fragmented to such an
extreme degree that Euclid—a term used in
this essay to denote all classical geometry—
is hardly of any help in describing their form.
The coastline of a typical oceanic island, to
take an example, is neither straight, nor
circular, nor elliptic, and no other classical
curve can serve, without undue artificial-
ity in the presentation and organization of
empirical measurements and in the search
for explanations. Similarly, no surface in Eu-
clid represents adequately the boundaries of
clouds or rough turbulent wakes.…
Michael F. Barnsley is a professor at the Mathematical Sci-
ences Institute, Australian National University. His email
address is Michael.Barnsley@anu.edu.au
.
Michael Frame is adjunct professor of mathematics at Yale
University. His email address is michael.frame@yale.edu
.
DOI: http://dx.doi.org/10.1090/noti894
In the present Essay I hope to show that
it is possible in many cases to remedy this
absence of geometric representation by us-
ing a family of shapes I propose to call
fractals—or
fractal sets. The most useful
among them involve chance, and their irreg-
ularities are statistical in nature. A central
role is played in this study by the concept
of fractal (or Hausdorff-Besicovitch) dimen-
sion.…Some fractal sets are curves, others
are surfaces, still others are clouds of discon-
nected points, and yet others are so oddly
shaped that there are no good terms for
them in either the sciences or the arts. The
variety of these forms should be sampled by
browsing through the illustrations.…
—Benoît B. Mandelbrot [16, pp. 1–2]
As with the now familiar principle that grav-
itational force tethers the earth to the sun, it
has become hard to imagine what it was like
not to know that many physical phenomena
can be described using nondifferentiable, rough
mathematical objects.
Important fractals such as the Cantor set, the
Sierpinski triangle, and Julia sets were well known
to some mathematicians, but they were neither
visible nor promoted to any practical purpose. To
me, looking back, it seems that these beautiful
things were hidden behind veils of words and
symbols with few diagrams, certainly no detailed
pictures; for example, the long text (in French)
of Gaston Julia failed to reveal to most people,
including most mathematicians, the full wonder of
1208
Notices of the AMS
Volume
59, Number 9
the endless arabesques and intricate visual adven-
tures in the boundaries of Fatou domains. It was as
though such objects were guarded by the priests
of mathematics, occasionally to be displayed, like
the monstrance at Benediction, to the inner core of
true believers. I was ritually inducted to calculus
in my first year at Oxford by Hammersley, who
took us through a full proof of the existence of
a Weierstrass nowhere differentiable continuous
curve from first principles. Half an hour with pic-
tures would have saved a lot of time and would
not have tainted our logical skills.
Benoît not only wrested these abstract objects,
these contrary children of pure mathematics, out
from the texts where they lay hidden, but he
also named them and put them to work to help
to describe the physical observable world. He
saw a close kinship between the needs of pure
mathematics and the Greek mythological being
Antaeus. In an interview [6] Benoît said, “The
son of Earth, he had to touch the ground every
so often in order to reestablish contact with
his Mother, otherwise his strength waned. To
strangle him, Hercules simply held him off the
ground. Separation from any down-to-earth input
could safely be complete for long periods—but
not forever.” He also said, “My efforts over the
years had been successful to the extent, to take an
example, that fractals made many mathematicians
learn a lot about physics, biology, and economics.
Unfortunately, most were beginning to feel they
had learned enough to last for the rest of their
lives. They remained mathematicians, had been
changed by considering the new problems I raised,
but largely went their own way.”
John Hutchinson is an example of a pure math-
ematician who was strongly influenced by Benoît’s
work.
In 1979 I was on study leave from the
Australian National University, visiting Fred
Almgren at Princeton for 6 months, as a re-
sult of my then interest in geometric measure
theory. While there, Fred suggested I read
Mandelbrot’s book Fractals: Form, Chance
and Dimension and look at putting it, or some
of it, into a unified mathematical framework.
As a result, we organised a seminar in which
I spoke about six times as my ideas de-
veloped. Participants included, besides Fred
and myself, Bob Kohn, Vladimir Schaeffer,
Bruce Solomon, Jean Taylor and Brian White.
Out of this came my 1981 article “Fractals
and self-similarity” [7] in the Indiana Uni-
versity Math. Journal, which introduced the
idea of an iterated function system (though
not with that name) for generating fractal
sets, similar ideas for fractal measures, and
Figure 1. An outlier Mandelbrot set (M-set)
(surrounded by yellow, then red) connected via a
branch of a tree-like path to the whole M-set.
The connectivity of the M-set was conjectured by
Benoît in 1980 and established by Adrien
Douady and John Hubbard in 1982.
Figure 2. Picture of F
16
(S)
F
16
(S)
F
16
(S) where S ⊂ R
2
S ⊂ R
2
S ⊂ R
2
,
F(S) = f
1
(S) ∪
f
2
(S)
F(S) = f
1
(S) ∪ f
2
(S)
F(S) = f
1
(S) ∪ f
2
(S), and f
1
, f
2
: R
2
→ R
2
f
1
, f
2
: R
2
→ R
2
f
1
, f
2
: R
2
→ R
2
are affine
contractions. The sequence
(F
n
(S))
(F
n
(S))
(F
n
(S)) converges
in the Hausdorff metric to a self-similar set, a
fractal, with Hausdorff dimension less than two.
This article has been decorated with pictures, in
the spirit of Benoît.
various structure theorems for fractals. In-
terestingly, this paper had no citations for a
few years, but now it frequently gets in the
AMS annual top ten list.
Mandelbrot’s ideas were absolutely es-
sential and fundamental for my paper. I still
October
2012
Notices of the AMS
1209