Reflections on the 69th

William Lowell Putnam

Mathematical Competition

Armstrong Putnam Team

The Hudson Colloquium Series

 Willaim Lowell Putnam II

Lawyer & Banker

Died in 1923 at age around 61.
wife: Elizabeth Lowell Putnam

In 1927 she established

the William Lowell Putnam Intercollegiate Memorial Fund
in order to begin a college-level mathematics competition,
the William Lowell Putnam Mathematical Competition.
The competition begun in 1935.
 Hilary Putnam

In 1970 Matiyasevich
solved Hilbert’s 10th problem
building on earlier work of
Davis, Putnam, and Robinson.

Hilbert’s 10th problem:

Is there an algorithm which determines whether or not
every Diophantine equation f (x1 , . . . , xn ) = 0 has an integer solution.
 Hilary Putnam

In 1970 Matiyasevich
solved Hilbert’s 10th problem
building on earlier work of
Davis, Putnam, and Robinson.

Hilbert’s 10th problem:

Is there an algorithm which determines whether or not
every Diophantine equation f (x1 , . . . , xn ) = 0 has an integer solution.
Ans: No, there isn’t
 Putnam Mathematical Competition
When: First Saturday in December.
Duration:
• 3 hrs (morning) for six problems
• 2 hr for lunch
• 3 hrs (afternoon) for six problems
Where: Participating universities.

The twelve problems can typically be solved

with only basic knowledge of college mathematics
but require extensive creative thinking.
 Putnam Mathematical Competition
Grading:

Each problem is graded on a basis of 0 to 10 points.

All the necessary work to justify an answer and all the neces-
sary steps of a proof must be shown clearly to obtain full credit.

Some partial credit may be given, but only when a contestant

has shown significant and substantial progress toward a solu-
tion.
 Putnam Mathematical Competition
Statistics for the 69th Competition:

• 3627 participants from 545 institutions.

• 1712 (47%) scored 0.
• 364 (10%) scored 10.
• 157 (4%) scored 20.
• Putnam Fellows:
117, 110, 108, 102, 101
Each receives $2,500.
Putnam Fellows for the 69th Competition

Brian R Lawrence Cal Tech

Seok Hyeong Lee Stanford Univ.
Arnav Tripathy Harvard Univ.
Bohua Zhan MIT
Yufei Zhao MIT
 Putnam Fellows
for the 37th Competition (1974)
 Putnam Fellows
for the 37th Competition (1974)

Grant M Roberts Waterloo

James B. Saxe Union College
Karl Rubin Princeton Univ.
Phillip Strenski AASU
Thomas G. Goodwillie Harvard
 Famous Putnam Fellows

Fields Medalists:
Milnor, Mumford, Quillen

Nobel Prize Winners in Physics:

Feynman, Wilson
 Four-time Putnam Fellows

Don Coppersmith (MIT) 1968-1971

Arthur Rubin (Purdue, CalTech) 1970-1973
Bjorn Poonen (Harvard) 1985-1988
Ravi Vakil (Toronto) 1988-1991
Gabriel Carroll (UC Berk., Harvard) 2000-2003
Reid Barton (MIT) 2001-2004
Daniel Kane (MIT) 2003-2006
 Seminar, Fall 2008
Instructors
James Brawner, Sungkon Chang

Participating Students
Alex Collins, Dan Holland, Scott King

Participating Faculty Members

Tim McMillan, Tim Ellis
 Seminar, Fall 2008
Instructors
James Brawner, Sungkon Chang

Participating Students
Alex Collins, Dan Holland, Scott King

Participating Faculty Members

Tim McMillan, Tim Ellis
Activities
• One-hour seminar per week.
• We worked on seven sets of problems.
• The students took one take-home (review) test.

Website: Department  Curriculum  Putnam Course

Seminar, Fall 2008
Problems of The 69th Competition

Show the test.

 Solutions of Problems
of The 69th Competition
Scott King will present a solution of
 Solutions of Problems
of The 69th Competition
Dan Holland will present a solution of
 Solution of Problem B3

The 4-dimensional hypercube of side length 1 centered at the origin

is the subset
{(x1, x2, x3, x4) ∈ R4 : |xi| ≤ 1/2, i = 1, 2, 3, 4}.
 Solution of Problem B3
A hyperplane in R4 is a subset
{(x1, x2, x3, x4) ∈ R4 : ax1 + bx2 + cx3 + dx4 = e},
which is 3-dimensional.
A plane in R4 is
the intersection of two hyperplanes,
which is 2-dimensional.
A line in R4 is
the intersection of three hyperplanes,
which is 1-dimensional.
 Solution of Problem B3
Parametric Vector Equations

A hyperplane in R4 is given by
{sa + tb + uc + d : s,t, u ∈ R},
which is 3-dimensional,
A plane in R4 is given by
{sa + tb + c : s,t ∈ R},
which is 2-dimensional.
A line in R4 is given by
{sa + b : s ∈ R},
which is 1-dimensional.
 Solution of Problem B3
Parametric Vector Equations

A hypersurface in R4 is given by
{ f (s,t, u), g(s,t, u), h(s,t, u), j(s,t, u) ∈ R4 : s,t, u ∈ R},

which is 3-dimensional,
A surface in R4 is given by
{ f (s,t), g(s,t), h(s,t), j(s,t) ∈ R4 : s,t ∈ R},

which is 2-dimensional.
A curve in R4 is given by
{ f (s), g(s), h(s), j(s) ∈ R4 : s ∈ R},

which is 1-dimensional.
 Solution of Problem B3
A circle in R3
 Solution of Problem B3
A circle in R4 is the intersection of
a (3-dimensional) hyper-sphere:

(x1 − c1)2 + · · · + (x4 − c4)2 = r2,

and a plane H.
 Solution of Problem B3

The 4-dimensional hypercube of side length 1 centered at the origin

is the subset
{(x1, x2, x3, x4) ∈ R4 : |xi| ≤ 1/2, i = 1, 2, 3, 4}.

A circle centered at the origin in R4 is

{(x1, x2, x3, x4) ∈ H : x12 + · · · + x42 = r2}.
where H is a (2-dimensional) plane.
 Solution of Problem B3

Lemma: Any circle C within the unit hypercube can be translated so

that it is centered on the origin and contained in the hypercube.
 Solution of Problem B3

Lemma: Any circle C within the unit hypercube can be translated so

that it is centered on the origin and contained in the hypercube.

proof: Let C be contained in the hyper-sphere:

(x1 − c1 )2 + · · · + (x4 − c4 )2 = r2 .
Let (a1 , a2 , a3 , a4 ) be a point in C. Then, the point at the opposite side is
(2c1 − a1 , . . . , 2c4 − a4 ).
Thus, −.5 ≤ ak ≤ .5 and −.5 ≤ 2ck − ak ≤ .5. This implies that
−.5 ≤ ak − ck ≤ .5.
 Solution of Problem B3

Lemma: Any circle C within the unit hypercube can be translated so

that it is centered on the origin and contained in the hypercube.

Cor: It is sufficient to consider circles centered at the origin.

 Solution of Problem B3
Prop: All circles centered at the origin in R4 are given by
{a cos θ + b sin θ : θ real}
where a and b are orthog. to e.o., and |a| = |b|.
 Solution of Problem B3
Prop: All circles centered at the origin in R4 are given by
{a cos θ + b sin θ : θ real}
where a and b are orthog. to e.o., and |a| = |b|.

proof: Let C be a circle centered at the origin in R4.

Then, by definition, C is the intersection of a plane and a
hyper-sphere: C = {s a + t b : |s a + t b| = r}. So, |a| = |b| = r.
 Solution of Problem B3
Prop: All circles centered at the origin in R4 are given by
{a cos θ + b sin θ : θ real}
where a and b are orthog. to e.o., and |a| = |b|.

proof: Let C be a circle centered at the origin in R4.

Then, by definition, C is the intersection of a plane and a
hyper-sphere: C = {s a + t b : |s a + t b| = r}. So, |a| = |b| = r.
We can choose the two vectors such that a · b = 0. Then,
r2 = |s a + t b|2 = (s a + t b) · (s a + t b) = r2 (s2 + t 2 ).
Thus, s2 + t 2 = 1 and hence,
s = cos θ and t = sin θ for some θ .
Therefore, C is given by such a parametric vector equation.
 Solution of Problem B3
Lemma: A circle C contained within the unit hypercube cannot have
√
a radius more than 2/2.
 Solution of Problem B3
Lemma: A circle C contained within the unit hypercube cannot have
√
a radius more than 2/2.

proof: Recall the prop: C = {a cos θ + b sin θ } for some a and b.

The coordinates of the points on this circle can, by the prop., be
given by
ak cos θ + bk sin θ = hak , bk i · hcos θ , sin θ i.
 Solution of Problem B3
Lemma: A circle C contained within the unit hypercube cannot have
√
a radius more than 2/2.

proof: Recall the prop: C = {a cos θ + b sin θ } for some a and b.

The coordinates of the points on this circle can, by the prop., be
given by
ak cos θ + bk sin θ = hak , bk i · hcos θ , sin θ i.
q
Hence, |ak cos θ + bk sin θ | ≤ a2k + b2k .
q Moreover, the equality holds for some θ .
Therefore, a2k + b2k ≤ 1/2.
 Solution of Problem B3
Lemma: A circle C contained within the unit hypercube cannot have
√
a radius more than 2/2.
q
proof: So, a2k + b2k ≤ 1/2. Since r = |a| = |b|,

2r2 = |a|2 + |b|2

= (a21 + · · · + a24) + (b21 + · · · + b24)
4
= ∑ (a2k + b2k ) ≤ 4 · (1/4) = 1.
k=1
 Solution of Problem B3

Lemma: Let a = h.5, .5, 0, 0i and b = h0, 0, .5, .5i.

√
Then, the following circle has radius 2/2:

{a cos θ + b sin θ : θ real}.

 Solution of Problem B3

Lemma: Let a = h.5, .5, 0, 0i and b = h0, 0, .5, .5i.

√
Then, the following circle has radius 2/2:

{a cos θ + b sin θ : θ real}.

√
Ans: The largest circle has radius 2/2.
 Problem A5

Problem A5:
Let n ≥ 3 be an integer. Let f (x) and g(x) be polynomials with
real coefficients such that the points




f (1), g(1) , f (2), g(2) , . . . , f (n), g(n)

are the vertices of a regular n-gon in counterclockwise order.

Prove that at least one of f (x) and g(x) has degree greater than
or equal to n − 1.
 Example for Problem A5
Let n = 3, and consider three points


cos(2πk/3), sin(2πk/3) for k = 0, 1, 2.
 Example for Problem A5
Let n = 3, and consider three points


cos(2πk/3), sin(2πk/3) for k = 0, 1, 2.

The x-coordinates: (1, 1), (2, −1/2), (3, −1/2).

√ √
The y-coordinates: (1, 0), (2, 3/2), (3, − 3/2).

f (x) 6= ax + b;
g(x) 6= cx + d.
 Example for Problem A5
Let n = 3, and consider three points


cos(2πk/3), sin(2πk/3) for k = 0, 1, 2.

The x-coordinates: (1, 1), (2, −1/2), (3, −1/2).

√ √
The y-coordinates: (1, 0), (2, 3/2), (3, − 3/2).

(x − 2)(x − 3) (x − 1)(x − 3)
F(x) = 1 · + (−1/2) ·
(1 − 2)(1 − 3) (2 − 1)(2 − 3)
(x − 1)(x − 2)
+ (−1/2) · .
(3 − 1)(3 − 2)
 Example for Problem A5
Let n = 3, and consider three points


cos(2πk/3), sin(2πk/3) for k = 0, 1, 2.

The x-coordinates: (1, 1), (2, −1/2), (3, −1/2).

√ √
The y-coordinates: (1, 0), (2, 3/2), (3, − 3/2).

(x − 2)(x − 3) √ (x − 1)(x − 3)
G(x) = 0 · + ( 3/2) ·
(1 − 2)(1 − 3) (2 − 1)(2 − 3)
√ (x − 1)(x − 2)
+ (− 3/2) · .
(3 − 1)(3 − 2)
 Example for Problem A5
Let n = 3, and consider three points


cos(2πk/3), sin(2πk/3) for k = 0, 1, 2.

The x-coordinates: (1, 1), (2, −1/2), (3, −1/2).

√ √
The y-coordinates: (1, 0), (2, 3/2), (3, − 3/2).

F(x) and G(x) are quadratic,

and are called Lagrange Interpolations.
 Example for Problem A5
Let n = 3, and consider three points


cos(2πk/3), sin(2πk/3) for k = 0, 1, 2.

The x-coordinates: (1, 1), (2, −1/2), (3, −1/2).

√ √
The y-coordinates: (1, 0), (2, 3/2), (3, − 3/2).

Lemma: If f (x) and g(x) are polynomials which interpolate those x-

and y-coordinates, resp., then

f (x) = (x − 1)(x − 2)(x − 3)Q1(x) + F(x);

g(x) = (x − 1)(x − 2)(x − 3)Q2(x) + G(x)

proof: f (x) − F(x) and g(x) − G(x) are divisible by

(x − 1)(x − 2)(x − 3).
 Solutions for Problem A5
Let (k, ak ) for k = 1, . . . , n be n points.

Lemma: If f (x) and F(x) are two polynomials which interpolate the
n points, then

f (x) = (x − 1) · · · (x − n)Q(x) + F(x).

Cor: There is at most one polynomial of degree ≤ n − 1 which inter-

polates the n points.
 Solutions for Problem A5
Let (k, ak ) for k = 1, . . . , n be n points.

Lemma: If f (x) and F(x) are two polynomials which interpolate the
n points, then

f (x) = (x − 1) · · · (x − n)Q(x) + F(x).

Cor: There is at most one polynomial of degree ≤ n − 1 which inter-

polates the n points.
 Solutions for Problem A5
Let (k, ak ) for k = 1, . . . , n be n points.

Lemma: If f (x) and F(x) are two polynomials which interpolate the
n points, then

f (x) = (x − 1) · · · (x − n)Q(x) + F(x).

Cor: There is at most one polynomial of degree ≤ n − 1 which inter-

polates the n points.

Cor: The Lagrange interpolation is the polynomial of the smallest

degree that interpolates the n points.
 Solutions for Problem A5
Let n = 5, and consider the n points


cos(2πk/n), sin(2πk/n) for k = 0, 1, . . . , (n − 1).

Let F(x) and G(x) be the Lagrange interpolations

for the x- and y-coordinates, resp.
Let ak = cos(2πk/n) and bk = sin(2πk/n). Then,
P1(x) P5(x)
F(x) = a1 + · · · + a5 ,
P1(1) P5(5)
P1(x) P5(x)
G(x) = b1 + · · · + b5
P1(1) P5(5)
where Pk (x) is the polynomial (x − 1) · · · (x − 5)/(x − k).
 Solutions for Problem A5
Let ak = cos(2πk/n) and bk = sin(2πk/n). Then,
P1(x) P5(x)
F(x) = a1 + · · · + a5 ,
P1(1) P5(5)
P1(x) P5(x)
G(x) = b1 + · · · + b5
P1(1) P5(5)
where Pk (x) is the polynomial (x − 1) · · · (x − 5)/(x − k).
 Solutions for Problem A5
Let ak = cos(2πk/n) and bk = sin(2πk/n). Then,
P1(x) P5(x)
F(x) = a1 + · · · + a5 ,
P1(1) P5(5)
P1(x) P5(x)
G(x) = b1 + · · · + b5
P1(1) P5(5)
where Pk (x) is the polynomial (x − 1) · · · (x − 5)/(x − k).
The coefficients of x4 in F(x) and G(x) are
5 5
ak bk
A= ∑ and B = ∑ .
k=1 Pk (k) k=1 Pk (k)
 Solutions for Problem A5
Let ak = cos(2πk/n) and bk = sin(2πk/n). Then,
P1(x) P5(x)
F(x) = a1 + · · · + a5 ,
P1(1) P5(5)
P1(x) P5(x)
G(x) = b1 + · · · + b5
P1(1) P5(5)
where Pk (x) is the polynomial (x − 1) · · · (x − 5)/(x − k).
The coefficients of x4 in F(x) and G(x) are
5 5
A= ∑ ak · ck and B = ∑ bk · ck
k=1 k=1

where ck = 1/Pk (k). Then,

 Solutions for Problem A5
Let ak = cos(2πk/n) and bk = sin(2πk/n). Then,
P1(x) P5(x)
F(x) = a1 + · · · + a5 ,
P1(1) P5(5)
P1(x) P5(x)
G(x) = b1 + · · · + b5
P1(1) P5(5)
where Pk (x) is the polynomial (x − 1) · · · (x − 5)/(x − k).
The coefficients of x4 in F(x) and G(x) are
5 5
A= ∑ ak · ck and B = ∑ bk · ck
k=1 k=1

where ck = 1/Pk (k). Then,

A + iB = c1(a1 + ib1) + · · · + c5(a5 + ib5) = c1ζ50 + · · · + c5ζ54.
 Solutions for Problem A5
Let ζ5 = cos(2π/5) + i sin(2π/5),
Pk (x) is the polynomial (x − 1) · · · (x − 5)/(x − k),
and ck = 1/Pk (k).
Then, A + iB = c1ζ50 + · · · + c5ζ54.

c1 = 1/(−1)(−2)(−3)(−4)
c2 = 1/(1)(−1)(−2)(−3)
c3 = 1/(2)(1)(−1)(−2)
c4 = 1/(3)(2)(1)(−1)
c5 = 1/(4)(3)(2)(1)
 Solutions for Problem A5
Let ζ5 = cos(2π/5) + i sin(2π/5),
Pk (x) is the polynomial (x − 1) · · · (x − 5)/(x − k),
and ck = 1/Pk (k).
Then, A + iB = c1ζ50 + · · · + c5ζ54.

c1 = 1/(−1)(−2)(−3)(−4)
c2 = c1 · (−4)/(1)
c3 = 1/(2)(1)(−1)(−2)
c4 = 1/(3)(2)(1)(−1)
c5 = 1/(4)(3)(2)(1)
 Solutions for Problem A5
Let ζ5 = cos(2π/5) + i sin(2π/5),
Pk (x) is the polynomial (x − 1) · · · (x − 5)/(x − k),
and ck = 1/Pk (k).
Then, A + iB = c1ζ50 + · · · + c5ζ54.

c1 = 1/(−1)(−2)(−3)(−4)
c2 = c1 · (−4)/(1)
c3 = c1 · (−4)(−3)/(1)(2)
c4 = 1/(3)(2)(1)(−1)
c5 = 1/(4)(3)(2)(1)
 Solutions for Problem A5
Let ζ5 = cos(2π/5) + i sin(2π/5),
Pk (x) is the polynomial (x − 1) · · · (x − 5)/(x − k),
and ck = 1/Pk (k).
Then, A + iB = c1ζ50 + · · · + c5ζ54.

c1 = 1/(−1)(−2)(−3)(−4)
c2 = c1 · (−4)/(1)
c3 = c1 · (−4)(−3)/(1)(2)
c4 = c1 · (−4)(−3)(−2)/(1)(2)(3)
c5 = c1 · (−4)(−3)(−2)(−1)/(1)(2)(3)(4)
 Solutions for Problem A5

c1 = 1/(−1)(−2)(−3)(−4)
c2 = c1 · (−4)/(1)
c3 = c1 · (−4)(−3)/(1)(2)
c4 = c1 · (−4)(−3)(−2)/(1)(2)(3)
c5 = c1 · (−4)(−3)(−2)(−1)/(1)(2)(3)(4)

Then,

A + iB = = c1ζ50 + · · · + c5ζ54
       
4 4 4 4 4
= c1ζ50 − c1 ζ5 + c1 ζ52 − c1 ζ53 + c1 ζ .
1 2 3 4 5
= c1(1 − ζ5)4 ⇒ AB 6= 0.
 Solutions for Problem A5
Let ak = cos(2πk/n) and bk = sin(2πk/n). Then,
P1(x) P5(x)
F(x) = a1 + · · · + a5 ,
P1(1) P5(5)
P1(x) P5(x)
G(x) = b1 + · · · + b5
P1(1) P5(5)
where Pk (x) is the polynomial (x − 1) · · · (x − 5)/(x − k).

We showed that
F(x) and G(x) have degree 4.
 Solutions for Problem A5
Let ak = cos(2πk/n) and bk = sin(2πk/n). Then,
P1(x) P5(x)
F(x) = a1 + · · · + a5 ,
P1(1) P5(5)
P1(x) P5(x)
G(x) = b1 + · · · + b5
P1(1) P5(5)
where Pk (x) is the polynomial (x − 1) · · · (x − 5)/(x − k).

Lemma: If f (x) and g(x) are polynomials which interpolate ak ’s and

bk ’s, resp., then their degress are greater than or equal to 4.
 Solutions for Problem A5
Lemma: Let f (x) and g(x) be polynomials which interpolate the x-
and the y-coordinates of a regular pentagon not necessarily centered
at the origin. Then, at least one has degree at least 4.
 Solutions for Problem A5
Lemma: Let f (x) and g(x) be polynomials which interpolate the x-
and the y-coordinates of a regular pentagon not necessarily centered
at the origin. Then, at least one has degree at least 4.

proof: Suppose that their degrees are both < 4. Via a shift, a
rotation, and a dialation, we have
! !
f ( f + m) cos θ − (g + n) sin θ
→t
g ( f + m) sin θ + (g + n) cos θ
The 70th William Lowell Putnam
Mathematical Competition,
December 5, 2009