Академический Документы
Профессиональный Документы
Культура Документы
Canada Resear
h Chair, Shrum Professor of S
ien
e & Dire
tor CECM
Simon Fraser University, Burnaby, BC Canada
Prepared for Queens University Symposium on
William Blake
\And I made a rural pen,
And I stained the water
lear,
And I wrote my happy songs,
Every
hild may joy to hear."
From Songs of Inno
en
e and Experien
e
relations, pi
Introdu tion
\If my tea
hers had begun by telling me that mathemati
s was pure
play with presuppositions, and wholly in the air, I might have be
ome
a good mathemati
ian. But they were overworked drudges, and I was
largely inattentive, and in
lined lazily to attribute to in
apa
ity in
myself or to a literary temperament that dullness whi
h perhaps was
due simply to la
k of initiation." (George Santayana1)
Most resear
h mathemati
ians neither think deeply about nor are terribly
on
erned about either pedagogy or the philosophy of mathemati
s. Nonetheless, as I hope to indi
ate, aestheti
notions have always permeated (pure and
applied) mathemati
s. And the top resear
hers have always been driven by an
aestheti
imperative:
aestheti s.
238{9.
Intelligen
er
,
ompiled by Robin Wilson and Jeremy Gray, Springer-Verlag, New York, 2000.
3
The transparen
ies, and other resour
es, expanding the presentation that this paper is
based on are available at
www.
e
m.sfu.
a/personal/jborwein/talks.html,
www.
e
m.sfu.
a/personal/jborwein/math
amp00.html and
www.
e
m.sfu.
a/ersonal/loki/Papers/Numbers/.
4 An ex
ellent middle s
hool illustration is des
ribed in Nathalie Sin
lair's \The aestheti
s
is relevant," for the learning of mathemati
s, 21 (2001), 25-32.
2.1 Gauss
Carl Friedri
h Gauss (1777-1855) on
e
onfessed5 ,
\I have the result, but I do not yet know how to get it."
One of Gauss's greatest dis
overies, in 1799, was the relationship between
the lemnis
ate sine fun
tion and the arithmeti
-geometri
mean iteration. This
was based on a purely
omputational observation. The young Gauss wrote in
his diary that the result \will surely open up a whole new eld of analysis."
He was right, as it pried open the whole vista of nineteenth
entury ellipti
and modular fun
tion theory. Gauss's spe
i
dis
overy, based on tables of
integrals provided by Stirling (1692-1770), was that the re
ipro
al of the integral
2
Z
0
p dt
1
t4
5 See \Isaa
Asimov's Book of S
ien
e and Nature Quotations," Isaa
Asimov and J. A.
Shulman (eds.), Weideneld and Ni
olson", New York (1988), 115.
an+1 :=
an + bn
2
bn+1 :=
an bn
2.2 Hadamard
A
onstru
tivist, experimental and aestheti
driven rationale for mathemati
s
ould hardly do better than to start with:
(J. Hadamard6 )
log n
."
This was one of the
ulminating results of 19th
entury mathemati
s and one
that relied on mu
h preliminary
omputation and experimentation.
\All physi
ists and a good many quite respe
table mathemati
ians
are
ontemptuous about proof."
6 In E. Borel, \Le ons sur la theorie des fon tions," 1928, quoted by George Polya in Mathemati al dis overy: On understanding, learning, and tea hing problem solving (Combined
in his apologia, \A Mathemati ian's Apology". The Apology is a spirited defense of beauty over utility:
\If the theory of numbers
ould be employed for any pra
ti
al and
obviously honourable purpose, ..."
even Gauss would be persuaded.
The Apology is one of Amazon's best sellers. And the existen
e of Amazon,
or Google, means that I
an be less than thorough with my bibliographi
details
without derailing a reader who wishes to nd the sour
e.
Hardy, in his tribute to Ramanujan entitled \Ramanujan, Twelve Le
tures
: : : ," page 15, gives the so-
alled `Skewes number' as a \striking example of a
false
onje
ture". The integral
li x =
Z x
0
dt
log t
Marry theory & pra
ti
e, history & philosophy, proofs & experiments.
Mat
h elegan
e and balan
e to utility and e
onomy.
Inform all mathemati
al modalities
omputationally: analyti
, algebrai
,
geometri
& topologi
al.
`Caging' and `Monster-barring' (in Imre Lakatos' words). Two parti
ularly
useful
omponents are:
in S
ien
e
& Engineering, May/June 3 (2001), 48-53. [CECM Preprint 00:1605
9
Jonathan M. Borwein and Robert Corless, \Emerging tools for experimental mathemati
s," Ameri
an Mathemati
al Monthly, 106 (1999), 889-909. [CECM Preprint 98:110.
10 ISC spa
e limits have
hanged from 10Mb being a
onstraint in 1985 to 10Gb being `easily
available' today.
3. Extensive use of Integer Relation Methods: PSLQ & LLL and FFT.11
Ex
lusion bounds are espe
ially useful and su
h methods provide a great
test bed for `Experimental Mathemati
s'.
4. Some automated theorem proving (using methods of Wilf-Zeilberger et
).
All these tools are a
essible through the listed CECM websites.
tries (hyperboli
, ellipti
) by repla
ing Eu
lid's axiom of parallels (or something
equivalent to it) with alternative forms."
2. The Ba
onian experiment is a
ontrived as opposed to a natural happening,
it \is the
onsequen
e of `trying things out' or even of merely messing about."
3. Aristotelian demonstrations: \apply ele
trodes to a frog's s
iati
nerve,
and lo, the leg ki
ks; always pre
ede the presentation of the dog's dinner with
the ringing of a bell, and lo, the bell alone will soon make the dog dribble."
4. The most important is Galilean: \a
riti
al experiment { one that dis
riminates between possibilities and, in doing so, either gives us
onden
e in the
view we are taking or makes us think it in need of
orre
tion."
The rst three forms are
ommon in mathemati
s, the fourth is not. It is
also the only one of the four forms whi
h has the promise to make Experimental
Mathemati
s into a serious repli
able s
ienti
enterprise.1516
4
p2 : : :
4.1 Irrationality
We present graphi
ally,
Tom Apostol's lovely new geometri
proof
p
17
rationality of
2.
of the ir-
PROOF. Consider the smallest right-angled iso eles integral with integer sides.
Cir
ums
ribe a
ir
le of length the verti
al side and
onstru
t the tangent on
the hypotenuse.
4.2 Rationality
p
p2 p2
p p
( 2 2)
= 2
=
p2
2 = 2:
p2
Either 2 2 Q or 2 62 Q:
In either
ase we
an dedu
e that there are irrational numbers and
with rational. But how do we know whi
h ones? One may build a whole
mathemati
al philosophy proje
t around this. Compare the assertion that
17
as Maple
onrms. This illustrates ni
ely that veri
ation is often easier than
dis
overy (similarly the fa
t multipli
ation is easier than fa
torization is at the
base of se
ure en
ryption s
hemes for e-
ommer
e). There are eight possible
(ir)rational triples:
=
;
and nding examples of all
ases is now a ne student proje
t.
4.3
22
7
0<
(1
22
x)4 x4
dx =
1 + x2
7
;
n=1
and students asked to
onrm this typi
ally mistake numeri
al validation for
symboli
proof.
Again we see that
omputing adds reality, making
on
rete the abstra
t, and
makes some hard things simple. This is strikingly the
ase in Pas
al's Triangle:
www.
e
m.sfu.
a/interfa
es/ aords an emphati
example where deep fra
tal stru
ture is exhibited in the elementary binomial
oe
ients. Berlinski writes
\The
omputer has in turn
hanged the very nature of mathemati
al experien
e, suggesting for the rst time that mathemati
s, like
physi
s, may yet be
ome an empiri
al dis
ipline, a pla
e where things
are dis
overed be
ause they are seen."
and
ontinues
quotes I agree with from Berlinski's \A Tour of the Cal ulus," Pantheon Books, 1995
10
and friends
My resear h with my brother on also oers aestheti and empiri al opportunities. The next algorithm grew out of work of Ramanujan.
5.1
A quarti algorithm
Set a0 = 6
4 2 and y0 = 2
yk+1 =
(1)
1. Iterate
1 (1
1 + (1
yk4 )1=4
yk4 )1=4
(2)
19 Details
11
rst general algorithm was found in 1977 by Ferguson & For
ade. Sin
e '77
one has many variants: LLL (also in Maple and Mathemati
a), HJLS, PSOS,
PSLQ ('91, parallelized '99).
Integer Relation Dete
tion was re
ently ranked among \the 10 algorithms
with the greatest in
uen
e on the development and pra
ti
e of s
ien
e and engineering in the 20th
entury," by J. Dongarra and F. Sullivan in Computing in
S
ien
e & Engineering, 2 (2000), 22-23. Also listed were: Monte Carlo, Simplex, Krylov Subspa
e, QR De
omposition, Qui
ksort, ..., FFT, Fast Multipole
Method.
20 These may be
21 See also J.M.
Dis rete
12
4
1
= 4 ar
tan( )
5
ar tan(
1
):
239
4
1
1
1
= ar
tan( ) + ar
tan( ) + ar
tan( ):
2
5
8
This was used by Dase to
ompute 200 digits of in his head in perhaps
the greatest feat of mental arithmeti
ever | ` 1/8' is apparently better
than `1/239' for this purpose.
13
5.4
`Pentium farming'
Bailey, P. Borwein and Ploue (1996) dis
overed a series for (and
orresponding ones for some other polylogarithmi
onstants) whi
h somewhat dis
on
ertingly allows one to
ompute hexade
imal digits of without
omputing prior
digits. The algorithm needs very little memory and no multiple pre
ision. The
running time grows only slightly faster than linearly in the order of the digit
being
omputed.
The key, found by `PSLQ', as des
ribed above, is:
=
1
X
k=0
1
16
k
4
8k + 1
2
8k + 4
1
8k + 5
1
8k + 6
Knowing an algorithm would follow they spent several months hunting by
omputer for su
h a formula. On
e found, it is easy to prove in Mathemati
a, in
Maple or by hand | and provides a very ni
e
al
ulus exer
ise.
This was a most su
essful
ase of
REVERSE MATHEMATICAL ENGINEERING
This is entirely pra
ti
able, God rea
hes her hand deep into : in September
1997 Fabri
e Bellard (INRIA) used a variant of this formula to
ompute 152
binary digits of , starting at the trillionth position (1012 ). This took 12 days
on 20 work-stations working in parallel over the Internet.
14
6.1 De Morgan
Augustus De Morgan, one of the most in
uential edu
ators of his period, wrote:
15
THEOREM. Given N non-
ollinear points in the plane there is a proper line
through only two points.24
Sylvester's
onje
ture was it seems forgotten for 50 years. It was rst established |\badly" in the sense that the proof is mu
h more
ompli
ated | by
Gallai (1943) and also by Paul Erdos who named `the Book' in whi
h God keeps
aestheti
ally perfe
t proofs. Erdos was an atheist. Kelly's proof was a
tually
published by Donald Coxeter in the MAA Monthly in 1948! A ne example of
how the ar
hival re
ord may get obs
ured.
PROOF. Consider the point
losest to a line it is not on and suppose that line
has three points on it (the horizontal line).
The middle of those three points is
learly
loser to the other line!
23 In D. Ma
Hale,
24 Posed in
minimal onguration.
Two more examples that belong in `the Book' are aorded by:
Snell's law | does one use the Cal
ulus to establish the Physi
s, or use
physi
al intuition to tea
h students how to avoid tedious
al
ulations?
(1
qn ) =
1
X
n=
( 1)n q (3n+1)n=2 :
p(7n + 5) 0 mod 7
and
17
P (q) = 1 + q + 2 q2 + 3 q3 + 5 q4 + 7 q5 + 11 q6 + 15 q7 + 22 q8 + 30 q9
+ 42 q 10 + 56 q 11 + 77 q 12 + 101 q 13 + 135 q 14 + 176 q 15 + 231 q 16
+ 297 q 17 + 385 q 18 + 490 q 19 + 627 q 20b + 792 q 21b + 1002 q 22 + 1255 q 23 +
If introspe
tion fails, we
an re
ognize the pentagonal numbers o
urring
above in Sloane and Ploue's on-line `En
y
lopedia of Integer Sequen
es':
www.resear
h.att.
om/personal/njas/sequen
es/eisonline.html. Here we
see a very ne example of Mathemati
s: the s
ien
e of patterns as is the title
of Keith Devlin's 1997 book. And mu
h more may similarly be done.
8
serious
urri
ular insights should
ome from neuro-biology (Dehaene et al., \Sour
es
of Mathemati
al Thinking: Behavioral and Brain-Imaging Eviden
e," S
ien
e, May 7, 1999).
26 From \Where Mathemati
s Comes From," Basi
Books, 2000, p. 5.
18
\What is parti
ularly ironi
about this is that it follows from the
empiri
al study of numbers as a produ
t of mind that it is natural
for people to believe that numbers are not a produ
t of mind!"27
I nd their general mathemati
al s
hema persuasive but their spe
i
a
ounting of mathemati
s for
ed and un
onvin
ing. Compare a more traditional
view whi
h I also espouse:
A
entury after biology started to think physi
ally, how will mathemati
al
thought patterns
hange?
\The idea that we
ould make biology mathemati
al, I think, perhaps is not working, but what is happening, strangely enough, is that
maybe mathemati
s will be
ome biologi
al!" (Greg Chaitin, Interview, 2000)
Consider the metaphori
al or a
tual origin of the present `hot topi
s`: simulated annealing (`protein folding'); geneti
algorithms (`s
heduling problems');
neural networks (`training
omputers'); DNA
omputation (`traveling salesman
problems'); and quantum
omputing (`sorting algorithms').
Nunez, p. 81.
R. C. Leowontin in
and
\: : : a new s
ienti
truth does not triumph by
onvin
ing its opponents and making them see the light, but rather be
ause its opponents
die and a new generation grows up that's familiar with it."
(Albert Einstein quoting Max Plan
k31 )
5. Mathemati al obje ts are a spe ial variety of a so ial- ulturalhistori al obje t. Contrary to the assertions of ertain post-modern
The re
ognition that \quasi-intuitive" methods may be used to gain mathemati
al insight
an dramati
ally assist in the learning and dis
overy of mathemati
s. Aestheti
and intuitive impulses are shot through our subje
t, and
honest mathemati
ians will a
knowledge their role.
31 From
32 From
20
Ameri an Mathemati al
8.5 Santayana
\When we have before us a ne map, in whi
h the line of the
oast,
now ro
ky, now sandy, is
learly indi
ated, together with the winding
of the rivers, the elevations of the land, and the distribution of the
population, we have the simultaneous suggestion of so many fa
ts,
the sense of mastery over so mu
h reality, that we gaze at it with delight, and need no pra
ti
al motive to keep us studying it, perhaps for
hours altogether. A map is not naturally thought of as an aestheti
obje
t ...
This was my earliest, and still favourite, en
ounter with aestheti
philosophy.
It may be old fashioned and unde
onstru
ted but to me it rings true:
And yet, let the tints of it be a little subtle, let the lines be a little
deli
ate, and the masses of the land and sea somewhat balan
ed, and
we really have a beautiful thing; a thing the
harm of whi
h
onsists
almost entirely in its meaning, but whi
h nevertheless pleases us in
the same way as a pi
ture or a graphi
symbol might please. Give the
symbol a little intrinsi
worth of form, line and
olor, and it attra
ts
like a magnet all the values of things it is known to symbolize. It
be
omes beautiful in its expressiveness." (George Santayana33)
To avoid a
usations of mawkishness, I nish by quoting Jerry Fodor34 :
Good software pa kages an make di ult on epts a essible (e.g., Mathemati a and Sket hpad).
Progress is made `one funeral at a time' (this harsher version of Plan
k's
omment is sometimes attribute to Niels Bohr).
1896.
21
22