CS 70 Discrete Mathematics and Probability Theory

Fall 2024 Rao, Hug HW 07 (Optional)

1 Proofs of the Combinatorial Variety

Note 10 Prove each of the following identities using a combinatorial proof.

(a) For every positive integer n > 1,

n   n−1  
n n−1
∑ k· k = n· ∑ k .
k=0 k=0

(b) For each positive integer m and each positive integer n > m,
       
n n n 3n
∑ a · b · c = m .
a+b+c=m

(Notation: the sum on the left is taken over all triples of nonnegative integers (a, b, c) such that
a + b + c = m.)

Solution:

(a) Suppose we have n people and want to pick some of them to form a special committee. More-
over, suppose we want to pick a leader from among the committee members - how many ways
can we do this?
We can do so by first picking the committee members, and then choosing the leader from
among the chosen members. We can pick a committee of size k in nk ways, and once we have

picked the committee, we have k choices for which member becomes the leader. In order to
account for all possible committee sizes, we need to sum over all valid values of k, hence we
get the expression
n  
n
∑ k· k ,
k=0
which is exactly the left hand side of the identity we want to prove.
Now, we can also count this set by first picking the leader for the committee, then choosing
the rest of committee. We have n choices for the leader, and then among the remaining n − 1
people, we can pick
any subset to form the rest of the committee. Picking a subset of size k
can be done in n−1
k ways, hence summing over k, we get the expression
n−1  
n−1
n· ∑ ,
k=0 k
which is exactly the right hand side of the identity we want to prove.

CS 70, Fall 2024, HW 07 (Optional) 1

 (b) Suppose we have n distinguishable red pencils, n distinguishable blue pencils, and n distin-
guishable green pencils (3n pencils total), and want to choose m of these pencils to bring to
class. How many ways can be do this?
We can do so by just picking the m pencils without considering color, as they are all distin-
3n

guishable. There are m ways of doing this, which is exactly the right hand side of the identity
we want to prove.
We can also count this set by picking some red pencils, the picking some blue pencils, and
then finally picking some green pencils. We can pick a red
pencils, b blue pencils, and c green
n n
n

pencils (with the tacit assumption that a + b + c = m) in a · b · c ways. Finally, in order to
account for all possible distributions of pencils, we need to sum over all valid triples (a, b, c),
which gives us the expression
     
n n n
∑ a · b · c ,
a+b+c=m

which is exactly the left hand side of the identity we want to prove.

2 Fibonacci Fashion
Note 10 You have n accessories in your wardrobe, and you’d like to plan which ones to wear each day for
the next t days. As a student of the Elegant Etiquette Charm School, you know it isn’t fashionable
to wear the same accessories multiple days in a row. (Note that the same goes for clothing items in
general). Therefore, you’d like to plan which accessories to wear each day represented by subsets
S1 , S2 , . . . , St , where S1 ⊆ {1, 2, . . . , n} and for 2 ≤ i ≤ t, Si ⊆ {1, 2, . . . , n} and Si is disjoint from
Si−1 .

(a) For t ≥ 1, prove that there are Ft+2 binary strings of length t with no consecutive zeros (assume
the Fibonacci sequence starts with F0 = 0 and F1 = 1).
(b) Use a combinatorial proof to prove the following identity, which, for t ≥ 1 and n ≥ 0, gives the
number of ways you can create subsets of your n accessories for the next t days such that no
accessory is worn two days in a row:
     
n n − x1 n − x2 n − xt−1
∑ ∑ · · · ∑ x1 ··· = (Ft+2 )n .
x1 ≥0 x2 ≥0 xt ≥0 x2 x 3 xt

a

(You may assume that b =0 whenever a < b.)

Solution:

(a) We will prove this by strong induction.

Base cases: For k = 1, the only binary strings possible are 0 and 1. Therefore, there are two
possible binary strings, and Fk+2 = F3 = 2. For k = 2, the binary strings possible are 11, 01,

CS 70, Fall 2024, HW 07 (Optional) 2

 and 10, and we have Fk+2 = F4 = 3, so the identity holds.

Inductive hypothesis: For k ≥ 2, assume that for all 1 ≤ x ≤ k, there are Fx+2 binary strings of
length x with no consecutive zeros.

Inductive step: Consider the set of binary strings of length k + 1 with no consecutive zeros. We
can group these into two sets: those which end with 0, and those which end with 1.
For those that end with a 0, these can be constructed by taking the set of binary strings of length
k − 1 with no consecutive zeros and appending 10 to the end of them. Then by the inductive
hypothesis, this set is of size Fk+1 . For those that end with a 1, these can be constructed by
taking the set of binary strings of length k with no consecutive zeros and appending a 1 to the
end of them. Then by the inductive hypothesis, this set is of size Fk+2 .
Since the union of these two subsets (those which end with 0 and those which end with 1) cover
all possible elements in the set of binary strings of length k + 1 with no consecutive zeros, the
size of this set will be Fk+1 + Fk+2 = Fk+3 . This thus proves the inductive hypothesis.
(b) We first consider the left-hand-side of the identity. To create subsets of accessories that are
consecutively disjoint with sizes xi = |Si |, 1 ≤ i ≤ n, there are xn1 ways to create S1 , the subset

of accessories you will wear on the first day. Then since S2 must be disjoint from S1 , there
n−x1

are x2 ways choose accessories to create S2 . Since S3 must be disjoint from S2 , there are
n−x2

ways choose accessories to create S3 , and so on. Thus there are xn1 n−x n−xt−1




x2 · · ·
1
x3 xt
ways to create subsets of accessories S1 , . . . , St with respective sizes x1 , . . . , xt . Then altogether,
S1 , . . . , St can be created in
     
n n − x1 n − x2 n − xt−1
∑ ∑ · · · ∑ x1 x2 x3
···
xt
x1 ≥0 x2 ≥0 xt ≥0
ways.
Now, consider the right-hand-side of the identity. Now for each accessory i ∈ {1, . . . , n}, we
will first decide which subsets S1 , . . . , St will contain accessory i, where we can’t assign item
i to consecutive subsets. For each accessory, we create a binary string of length t, where the
leading digit represents S1 , the next digit represents S2 , and so on. We will say that a 0 in digit k
means that we will wear the accessory on day k. Therefore, the number of ways we can assign
accessory i to subsets S1 , . . . , St such that no two consecutive subsets both have accessory i is
the same as the number of binary strings of length t with no consecutive zeros. Thus using
the result in part (a), there are Ft+2 ways to select the nonconsecutive subsets containing i
among S1 , . . . , St . Since we have n accessories, accessories 1, . . . , n can be placed into subsets
S1 , . . . , St in (Ft+2 )n ways.
This thus proves the identity.

3 Unions and Intersections

Note 11 Given:

CS 70, Fall 2024, HW 07 (Optional) 3

 • X is a countable, non-empty set. For all i ∈ X, Ai is an uncountable set.

• Y is an uncountable set. For all i ∈ Y , Bi is a countable set.

For each of the following, decide if the expression is "Always countable", "Always uncountable",
"Sometimes countable, Sometimes uncountable".
For the "Always" cases, prove your claim. For the "Sometimes" case, provide two examples – one
where the expression is countable, and one where the expression is uncountable.

(a) X ∩Y
(b) X ∪Y
S
(c) i∈X Ai
T
(d) i∈X Ai
S
(e) i∈Y Bi
T
(f) i∈Y Bi

Solution:

(a) Always countable. X ∩Y is a subset of X, which is countable.

(b) Always uncountable. X ∪Y is a superset of Y , which is uncountable.
S
(c) Always uncountable. Let x be any element of X. Ax is uncountable. Thus, i∈X Ai , a superset
of Ax , is uncountable.

(d) Sometimes countable, sometimes uncountable.

Countable: When the Ai are disjoint, the intersection is empty, and thus countable. For exam-
ple, let X = N, let Ai = {i} × R = {(i, x) | x ∈ R}. Then, i∈X Ai = ∅.
T

Uncountable: When the Ai are identical, the intersection is uncountable. Let X = N, let Ai = R
T
for all i. i∈X Ai = R is uncountable.
(e) Sometimes countable, sometimes uncountable.
S
Countable: Make all the Bi identical. For example, let Y = R, and Bi = N. Then, i∈Y Bi = N
is countable.
Uncountable: Let Y = R. Let Bi = {i}. Then,
S
i∈Y Bi = R is uncountable.
T
(f) Always countable. Let y be any element of Y . By is countable. Thus, i∈Y Bi , a subset of By ,
is also countable.

CS 70, Fall 2024, HW 07 (Optional) 4

 4 Countability Proof Practice
Note 11 (a) A disk is a 2D region of the form {(x, y) ∈ R2 : (x − x0 )2 + (y − y0 )2 ≤ r2 }, for some x0 , y0 , r ∈
R, r > 0. Say you have a set of disks in R2 such that none of the disks overlap (with possibly
varying x0 , y0 , and r values). Is this set always countable, or potentially uncountable?
(Hint: Attempt to relate it to a set that we know is countable, such as Q × Q.)
(b) A circle is a subset of the plane of the form {(x, y) ∈ R2 : (x − x0 )2 + (y − y0 )2 = r2 } for some
x0 , y0 , r ∈ R, r > 0. Now say you have a set of circles in R2 such that none of the circles
overlap (with possibly varying x0 , y0 , and r values). Is this set always countable, or potentially
uncountable?
(Hint: The difference between a circle and a disk is that a disk contains all of the points in its
interior, whereas a circle does not.)

Solution:

(a) Countable. Each disk must contain at least one rational point (an (x, y)-coordinate where x, y ∈
Q) in its interior, and due to the fact that no two disks overlap, the cardinality of the set of disks
can be no larger than the cardinality of Q × Q, which we know to be countable.
(b) Possibly uncountable. Consider the circles Cr = {(x, y) ∈ R2 : x2 + y2 = r} for each r ∈ R. For
r1 ̸= r2 , Cr1 and Cr2 do not overlap, and there are uncountably many of these circles (one for
each real number).

CS 70, Fall 2024, HW 07 (Optional) 5