Intro to Combinatorics

ft. Graphs
Kazi Aryan Amin

INMOTC Karnataka, December 26th, 2023

This handout is supposed to provide a reader with an introduction to some

preliminary ideas/themes in Olympiad Combinatorics.

Contents
1 Preliminaries 2

2 Induction 3

3 Algorithms and Invariants 4

4 Local Ideas 5

5 Global Ideas 7

6 Problems 9

7 Solutions to Examples 14
7.1 Example 2.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
7.2 Example 2.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
7.3 Example 3.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
7.4 Example 3.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
7.5 Example 4.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
7.6 Example 4.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
7.7 Example 4.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
7.8 Example 4.4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
7.9 Example 5.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
7.10 Example 5.2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
7.11 Example 5.3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

1
Kazi Aryan Amin (INMOTC Karnataka, December 26th, 2023) Intro to Combinatorics

1 Preliminaries
The intention of this handout is to familiarize the reader with some common themes in
Olympiad Combinatorics. A recurring theme in the examples we cover, will be problems in
Graph Theory. One of the reasons I did this, is in the hopes that it would help take the fear
of Graphs out of the newbies. When I was starting out in Oly combo, I was pretty scared
of graphs at first and always thought they required a lot of "theory". While that is partially
true, most of Olympiad graph theory is "really just any other combo problem". I hope after
reading the handout, the reader feels comfortable working with graphs while having learned
some basic ideas in Oly combo.
Here are some of the type of arguments we cover:

• Induction

• Algorithms and Invariants

• Local ideas: ft. Greedy Algorithms and Discrete Continuity, et al.

• Global Ideas: ft. Double Counting, et al.

Here are some basic graph terms, we will define the rest as we move forward.

Notation
– A graph G = G (V, E) is a pair of sets such that E ⊆ V × V. Thus the elements
of E are 2-element subsets of V. We shall always assume V ∩ E is empty.
– The elements of V are called the vertices, while the elements of E are called the
edges. The usual way to visualize a graph is to draw a dot for every vertex,
and draw a line between 2 dots, if the corresponding 2 vertices form an edge.
– A vertex v is incident to an edge e, if v is one of the endpoints of e. Two vertices
u, v ∈ V are said to be adjacent, if uv ∈ E, that is if uv is an edge.
– A subgraph of G = (V, E) is a graph G = (V ′ , E′ ), if V ′ ⊆ V and E′ ⊆ E. The
subgraph is induced, if E′ contains all edges xy ∈ E such that x, y ∈ V ′ .
– A graph is complete, if there is an edge between any two pair of vertices.
– We may assign a direction to every edge of the graph, thus making it into a
directed graph.

2
Kazi Aryan Amin (INMOTC Karnataka, December 26th, 2023) Intro to Combinatorics

2 Induction
Some things to keep in mind while inducting:

• Most of the times the main difficulty of the problem is "finding a way to induct". So
most often we might need to choose what we are inducting on with some care.

• There are a number of ways induction shows up. The most simple idea is that you will
somehow try to a problem in n to a problem in n − 1. One of the ways to do this is by
deleting items carefully.

• Induction is often helpful in specifying algorithms. Suppose you already know how to
do something for n, and sometimes it turns out you just need to add a bit more to make
it work for n + 1. (of course, most often the point of the problem is to figure out that
"bit more"). We will see more examples of this in the next section.

• Sometimes instead of simply going from n 7→ n − 1, we might need to rather "partition"

n into two components a, b such that a + b = n and then use the induction hypothesis
on both of them to get something interesting.

• A slightly advanced idea: Suppose you have a family F of subsets of {1, 2, · · · , n}.
Suppose it turns out that you have two elements 1 ≤ a, b ≤ n such that for all sets
S ∈ F , either S contains both a, b, or neither. Then we can combine a, b into a single
element c, and induct down.

Now we move on to some contest examples:

Example 2.1 (USAMO 2002 Problem 1)

Let S be a set with n ≥ 1 elements, and let N be an integer with 0 ≤ N ≤ 2n . Prove that it
is possible to color every subset of S either black or white so that the following conditions
hold:

• The union of any two white subsets is white.

• The union of any two black subsets is black.

• There are exactly N white subsets.

Walkthrough.

1. WLOG S = [n]. First, let us figure out what to induct. We have two options n and N.
Note that if we decide to induct on N, since union of sets are much harder to control, it
becomes very unruly. So we will induct on n.

2. Let P (n) denote the set of all subsets of [n]. Note that there is a natural way to get P (n)
from P (n − 1). Every set S ∈ P (n) is of the form T ∪ {n} or T for some T ∈ P (n − 1).
This gives us a way to induct.

3. With the above setup, solve the problem for N ≤ 2n−1 .

4. Now we deal with the case N > 2n−1 , so now 2n − N ≤ 2n−1 . Finish using the last step.

3
Kazi Aryan Amin (INMOTC Karnataka, December 26th, 2023) Intro to Combinatorics

Example 2.2 (USA EGMO TST 2020 Problem 5)

Let G = (V, E) be a finite simple graph on n vertices. An edge e of G is called a bottleneck if
one can partition V into two disjoint sets A and B such that

• at most 100 edges of G have one endpoint in A and one endpoint in B; and

• the edge e is one such edge (meaning the edge e also has one endpoint in A and one
endpoint in B).

Prove that at most 100(n − 1) edges of G are bottlenecks.

Walkthrough.

1. Assume WLOG that there are no non-bottleneck edges. Start with a bottleneck edge e
and consider the corresponding sets A and B that partition the vertices of G.

2. The "partition" naturally makes way for us to induct! Induct on vertices of G, and use
the hypothesis on A and B. The finish should be direct.

3 Algorithms and Invariants

In a process, an invariant is some special quantity that always stays constant during the
process.
Conversely, a monvariant is quantity that either always increases, or always decreases
during a process.

In a lot of problems you will be asked to find all values of x for which some task is possible.
This usually has two parts:

• First, we figure out which values of x work. This comes from experimenting/playing
with small cases. Then to prove that those values actually work, you need to come up
with an algorithm.

• Now we need to show the other values of x do not work. Usually, we will try to find a
mono/in-variant to show impossibility.

Mono/in-variants can come up in various ways:

• Suppose you have a quantity associated to a process that decreases every step. If the
quantity is supposed to be a positive integer, then you’re done.

• Suppose I want to prove to prove certain configurations aren’t possible while doing
some process. One way to do this is to tag a quantity to every configuration, which
either stays constant or changes predictably. Show that for a configuration that isn’t
possible, something absurd happens to the quantity.

• Sometimes, our invariants are error terms, measuring how much the current configura-
tion "deviates" from the desired one.

4
Kazi Aryan Amin (INMOTC Karnataka, December 26th, 2023) Intro to Combinatorics

Example 3.1 (RMM SL 2018 C1)

Call a point in the Cartesian plane with integer coordinates a lattice point. Given a finite
set S of lattice points we repeatedly perform the following operation: given two distinct
lattice points A, B in S and two distinct lattice points C, D not in S such that ACBD is a
parallelogram with AB > CD, we replace A, B by C, D. Show that only finitely many such
operations can be performed.

Walkthrough. This is very direct use of mono-variants. To figure out what’s the correct mono-
variant to use, try to think about which lengths are affected... (Note that ACBD being a
parallelogram allows for some nice length conditions, etc). (I don’t really know how to write
a walkthrough for this problem.)

Example 3.2 (USAJMO 2019 Problem 1)

There are a + b bowls arranged in a row, numbered 1 through a + b, where a and b are
given positive integers. Initially, each of the first a bowls contains an apple, and each of the
last b bowls contains a pear. A legal move consists of moving an apple from bowl i to bowl
i + 1 and a pear from bowl j to bowl j − 1, provided that the difference i − j is even. We
permit multiple fruits in the same bowl at the same time. The goal is to end up with the
first b bowls each containing a pear and the last a bowls each containing an apple. Show
that this is possible if and only if the product ab is even.

Walkthrough.
1. The problem has two parts. First, we show that ab even works by specifying an
algorithm for the same. Then we show that both a, b odd doesn’t work.
2. First assume a is even. Show that it suffices to consider the case a = 2. (In other words,
find a way to go a 7→ a − 2.)
3. Solve the cases ( a, b) = (2, 1), (2, 2), (2, 3) by hand, and figure out a way to generalize.
4. Now we show that when a, b are both odd, it doesn’t work. Let Ao , Ae denote the
number of apples in odd and even positions respectively. Let Po , Pe be defined analo-
gously. Note that after every legal move, an even apple becomes odd and vice-versa
and similarly for pears.
5. What does the condition i − j even imply? Say i is odd, then so is j. This implies that
after the legal move, an odd apple becomes an even apple and similarly for the pears.
Similarly for i, j both even. Can you think of an invariant involving Ao , Ae , Po , Pe ?
6. What are the initial and final values of the invariant when ab is odd? Conclude.

4 Local Ideas
The main idea of this section is to make to slight changes/perturbations in a small part of
the problem. The idea is to look at some small substructure, try to modify it slightly, and see
what happens. Keep going until done.

• Greedy Algorithms: Keep grabbing the best thing possible one at a time till you can’t
anymore.
5
Kazi Aryan Amin (INMOTC Karnataka, December 26th, 2023) Intro to Combinatorics

• Smoothing/Making Small perturbations: Change the configuration slightly (for eg,

swap two items and see what changes happen. Often you can move from configuration
1 to configuration 2 by making a lot of these swaps. See how some specific quantity
changes with the swaps).

• Local Optimizations: Make some operations such that the problem "remains true"
after making these operations. In a sense after making these operations, there are more
restrictions on the problem, and often this gives us more structure to work with.

• Discrete Continity: Suppose a quantity increases by ±1. If it was −3 at some point and
+5 at some other point, it must have been 0 at some intermediate point.

Example 4.1
Every graph G with at least one edge has a subgraph H with δ( H ) > ε( H ) > ε( G ). Where
| E( G )|
δ( H ) denotes the minimal degree of a vertex in H, and ε( G ) = |V (G)|

Walkthrough. This is a pretty classic greedy algorithm problem.

1. The idea is to get H by deleting vertices of small degree until only the vertices ofa large
degree remain.

2. Call a vertex to be "bad" if it has a small degree. What vertices should be considered
bad? All those for which d(v) ≤ ε. Keep deleting bad vertices.

3. However there is some dilemma. Is it possible that some good vertices may become
bad after some of the deletions? Show that this cannot happen by showing ε doesn’t
decrease after every deletion.

4. When can we not delete any more vertices?

Example 4.2
There are 100 points on the circumference of a circle, arbitrarily labeled by 1, 2, . . . , 100. For
each of the three points, call their triangle clockwise if their increasing order is in clockwise
order. Prove that it is impossible to have exactly 2017 clockwise triangles.

Walkthrough.

1. Swap any two adjacent points. How many triangles change orientation?

2. Keep swapping till you get to a configuration where it is easy to determine the number
of clockwise triangles. Conclude.

Example 4.3
A tournament is a directed complete graph. Show that in any tournament on n vertices,
there exists a directed path v1 7→ v2 7→ · · · 7→ vn . (Such a path is called a Hamiltonian path)

Walkthrough. Classic discrete continuity example

6
Kazi Aryan Amin (INMOTC Karnataka, December 26th, 2023) Intro to Combinatorics

1. We will induct. Suppose v1 7→ v2 7→ · · · 7→ vn is a Hamiltonian path given to us by

induction. We need to fit in vn+1 somewhere. If vn+1 7→ v1 or vn 7→ vn+1 , then we are
done.

2. So assume WLOG that we have v1 7→ vn+1 and vn+1 7→ vn . Look at the direction of
the edge vn+1 to vi for 1 ≤ i ≤ n. It is • 7→ vn+1 for i = 1 and vn+1 7→ •, for i = n.
So by discrete continuity, there is some index i, such that vi 7→ vn+1 and vn+1 7→ vi+1 .
Conclude.

Example 4.4
A positive integer n is called special if there exist integers a > 1 and b > 1 such that
n = ab + b. Is there a set of 2014 consecutive positive integers that contains exactly 2012
special numbers?

Walkthrough.

1. Let f (n) denote the number of special numbers in the set {n, n + 1, · · · , n + 2013}. How
does f (n) change as n changes?

2. Find a x such that f ( x ) < 2012 and f (y) = 2014. Then by discrete continuity, we must
have hit f (n) = 2012 somewhere in between.

5 Global Ideas
In contrast to the previous situation, when we took things one at a time/ looked at a specific
part of a structure, here we will look at the entire problem at once. Some common tropes:

• Pigeonhole Principle

• Averaging/Linearity of Expectation: Suppose you have n objects with average value A.

Then someone must have value ≥ A and someone ≤ A.

• Probabilistic Method: (This might not be the best way to put this oops) Suppose you
want to show some set of objects has a "good" object amongst them. Show that the
probability of finding a bad object is < 1.

• Double Counting: Often you will be working with double sums related to some specific
quantity. When you have double sums, there’s really a few thing you can do: one of
them is swapping!
∑∑=∑∑
a∈ A b∈ B b∈ B a∈ A

Here are a few ways to identify problems with global ideas from local ones. One of the
main features would be that the problem has a "high degree of symmetry", that is there are
non real "special objects" in the structure of the problem. (If the latter was true, you could
consider deleting and inducting down, etc).

Example 5.1 (Mantel’s Theorem) j 2k

Let G be a graph on n vertices with no triangles. Show that G has atmost n4 edges.

7
Kazi Aryan Amin (INMOTC Karnataka, December 26th, 2023) Intro to Combinatorics

Walkthrough.

1. Let uv ∈ E, note that u, v has no common neighbors. Using this, find a bound for
deg u + deg v.

2. Show that ∑uv∈E deg u + deg v = ∑u∈V (deg u)2 . Bound the latter term by Jensen and
finish.

Example 5.2 (ARMO 2014 9.8)

In a country of n cities, an express train runs both ways between any two cities. For any
train, ticket prices either direction are equal, but for any different routes these prices are
different. Prove that the traveler can select the starting city, leave it and go on, successively,
n − 1 trains, such that each fare is smaller than that of the previous fare. (A traveler can
enter the same city several times.)

Walkthrough.

1. Start with the obvious greedy algorithm. Show that it doesn’t always work

2. Thus we resort to a global approach. We start every person at a vertex. We hope that
the average length is large enough (namely n − 1). So we want the total length to be
n(n − 1) = 2(n2 ).

3. This strongly suggests that the algorithm we choose for how every person traverses
should ensure that in the end, we have a directed edge (meaning a traveler went through
that edge in that direction) u 7→ v for every 2 vertices u, v.

4. Show that the above claim is true and conclude. Think greedily.

Example 5.3 (RMM 2012 Problem 1)

Given a finite number of boys and girls, a sociable set of boys is a set of boys such that
every girl knows at least one boy in that set; and a sociable set of girls is a set of girls such
that every boy knows at least one girl in that set. Prove that the number of sociable sets
of boys and the number of sociable sets of girls have the same parity. (Acquaintance is
assumed to be mutual.)

Here’s an example of how to use global in ” (mod 2)” situtations.

Walkthrough.

1. Take the obvious graph interpretation. Let f (b, g) be 0 if bg ∈ E, and 1 otherwise. Find
a "indicator" function for sociable sets of boys. That is find a function χ (in terms of
f (b, g)), such that if X is a set of boys, then χ( X ) = 1 if X is sociable, and 0 otherwise.

2. Consider the sum ∑ X sociable 1. Write it as a sum of χ(S) for any arbitary subset S of the
boys.

3. Consider this sum (mod 2) and show its symmetric in the boys and girls sets.

8
Kazi Aryan Amin (INMOTC Karnataka, December 26th, 2023) Intro to Combinatorics

6 Problems
Problem 1 (ISL 2015 C1). In Lineland there are n ≥ 1 towns, arranged along a road running
from left to right. Each town has a left bulldozer (put to the left of the town and facing left)
and a right bulldozer (put to the right of the town and facing right). The sizes of the 2n
bulldozers are distinct. Every time when a left and right bulldozer confront each other, the
larger bulldozer pushes the smaller one off the road. On the other hand, bulldozers are quite
unprotected at their rears; so, if a bulldozer reaches the rear-end of another one, the first one
pushes the second one off the road, regardless of their sizes.

Let A and B be two towns, with B to the right of A. We say that town A can sweep town
B away if the right bulldozer of A can move over to B pushing off all bulldozers it meets.
Similarly town B can sweep town A away if the left bulldozer of B can move over to A
pushing off all bulldozers of all towns on its way.

Prove that there is exactly one town that cannot be swept away by any other one.

Problem 2 (IMO SL 2020 C1). Let n be a positive integer. Find the number of permutations
a1 , a2 , . . . an of the sequence 1, 2, . . . , n satisfying

a1 ≤ 2a2 ≤ 3a3 ≤ · · · ≤ nan

Problem 3 (IMO SL 2017 C1). A rectangle R with odd integer side lengths is divided into
small rectangles with integer side lengths. Prove that there is at least one among the small
rectangles whose distances from the four sides of R are either all odd or all even.

Problem 4 (IMO SL 2019 C2). You are given a set of n blocks, each weighing at least 1; their
total weight is 2n. Prove that for every real number r with 0 ≤ r ≤ 2n − 2 you can choose a
subset of the blocks whose total weight is at least r but at most r + 2.

Problem 5 (USA TSTST 2020 P1). Let a, b, c be fixed positive integers. There are a + b + c
ducks sitting in a circle, one behind the other. Each duck picks either rock, paper, or scissors,
with a ducks picking rock, b ducks picking paper, and c ducks picking scissors. A move
consists of an operation of one of the following three forms:

1. If a duck picking rock sits behind a duck picking scissors, they switch places.

2. If a duck picking paper sits behind a duck picking rock, they switch places.

3. If a duck picking scissors sits behind a duck picking paper, they switch places.

Determine, in terms of a, b, and c, the maximum number of moves which could take place,
over all possible initial configurations.

9
Kazi Aryan Amin (INMOTC Karnataka, December 26th, 2023) Intro to Combinatorics

Problem 6 (RMM SL 2016 C1). We start with any finite list of distinct positive integers. We
may replace any pair n, n + 1 (not necessarily adjacent in the list) by the single integer n − 2,
now allowing negatives and repeats in the list. We may also replace any pair n, n + 4 by
n − 1. We may repeat these operations as many times as we wish. Either determine the most
negative integer which can appear in a list, or prove that there is no such minimum.

Problem 7 (USA TSTST 2023 P4). Let n ≥ 3 be an integer and let Kn be the complete graph
on n vertices. Each edge of Kn is colored either red, green, or blue. Let A denote the number
of triangles in Kn with all edges of the same color, and let B denote the number of triangles in
Kn with all edges of different colors. Prove

n ( n − 1)
B ≤ 2A + .
3
(The complete graph on n vertices is the graph on n vertices with (n2 ) edges, with exactly one
edge joining every pair of vertices. A triangle consists of the set of (32) = 3 edges between 3 of
these n vertices.)

Problem 8 (IMO SL 2017 C3). Sir Alex plays the following game on a row of 9 cells. Initially,
all cells are empty. In each move, Sir Alex is allowed to perform exactly one of the following
two operations:

1. Choose any number of the form 2 j , where j is a non-negative integer, and put it into an
empty cell.

2. Choose two (not necessarily adjacent) cells with the same number in them; denote that
number by 2 j . Replace the number in one of the cells with 2 j+1 and erase the number in
the other cell.

At the end of the game, one cell contains 2n , where n is a given positive integer, while the
other cells are empty. Determine the maximum number of moves that Sir Alex could have
made, in terms of n.

Problem 9 (2022 Thailand MO Day 1 P3). Let Ω be a circle in a plane. 2022 pink points
and 2565 blue points are placed inside Ω such that no point has two colors and no two points
are collinear with the center of Ω. Prove that there exists a sector of Ω such that the angle at
the center is acute and the number of blue points inside the sector is greater than the number
of pink points by exactly 100. (Note: such sector may contain no pink points.)

Problem 10 (IMO SL 2016 C3). Let n be a positive integer relatively prime to 6. We paint
the vertices of a regular n-gon with three colours so that there is an odd number of vertices of
each colour. Show that there exists an isosceles triangle whose three vertices are of different
colours.

10
Kazi Aryan Amin (INMOTC Karnataka, December 26th, 2023) Intro to Combinatorics

Problem 11 (IMO SL 2021 C2). Let n ≥ 3 be a fixed integer. There are m ≥ n + 1 beads on
a circular necklace. You wish to paint the beads using n colors, such that among any n + 1
consecutive beads every color appears at least once. Find the largest value of m for which
this task is not possible.

Problem 12 (IMO SL 2022 C4). Let n > 3 be a positive integer. Suppose that n children are
arranged in a circle, and n coins are distributed between them (some children may have no
coins). At every step, a child with at least 2 coins may give 1 coin to each of their immediate
neighbors on the right and left. Determine all initial distributions of the coins from which it
is possible that, after a finite number of steps, each child has exactly one coin.

Problem 13 (Balkan MO SL 2021 C3). In an exotic country, the National Bank issues coins
that can take any value in the interval [0, 1]. Find the smallest constant c > 0 such that the
following holds, no matter the situation in that country:
Any citizen of the exotic country that has a finite number of coins, with a total value of no more
than 1000, can split those coins into 100 boxes, such that the total value inside each box is at most c.

Problem 14 (USAMO 2017 P4). Let P1 , P2 , . . . , P2n be 2n distinct points on the unit circle
x2 + y2 = 1, other than (1, 0). Each point is colored either red or blue, with exactly n red
points and n blue points. Let R1 , R2 , . . . , Rn be any ordering of the red points. Let B1 be the
nearest blue point to R1 traveling counterclockwise around the circle starting from R1 . Then
let B2 be the nearest of the remaining blue points to R2 travelling counterclockwise around
the circle from R2 , and so on, until we have labeled all of the blue points B1 , . . . , Bn . Show
that the number of counterclockwise arcs of the form Ri → Bi that contain the point (1, 0) is
independent of the way we chose the ordering R1 , . . . , Rn of the red points.

Problem 15 (RMM 2012 Problem 5). Given a positive integer n ≥ 3, colour each cell of an
n × n square array with one of ⌊(n + 2)2 /3⌋ colours, each colour being used at least once.
Prove that there is some 1 × 3 or 3 × 1 rectangular subarray whose three cells are coloured
with three different colours.

Problem 16 (USA TSTST 2019 Problem 4). Consider coins with positive real denominations
not exceeding 1. Find the smallest C > 0 such that the following holds: if we have any 100
such coins with total value 50, then we can always split them into two stacks of 50 coins each
such that the absolute difference between the total values of the two stacks is at most C.

Problem 17 (USA TSTST 2016 P5). In the coordinate plane are finitely many walls; which
are disjoint line segments, none of which are parallel to either axis. A bulldozer starts at an
arbitrary point and moves in the + x direction. Every time it hits a wall, it turns at a right
angle to its path, away from the wall, and continues moving. (Thus the bulldozer always
moves parallel to the axes.)
Prove that it is impossible for the bulldozer to hit both sides of every wall.

11
Kazi Aryan Amin (INMOTC Karnataka, December 26th, 2023) Intro to Combinatorics

Problem 18 (IMO SL 2014 C1). Let n points be given inside a rectangle R such that no two
of them lie on a line parallel to one of the sides of R. The rectangle R is to be dissected
into smaller rectangles with sides parallel to the sides of R in such a way that none of these
rectangles contains any of the given points in its interior. Prove that we have to dissect R into
at least n + 1 smaller rectangles.

Problem 19 (USA TSTST 2021 P7). Let M be a finite set of lattice points and n be a positive
integer. A mine-avoiding path is a path of lattice points with length n, beginning at (0, 0) and
ending at a point on the line x + y = n, that does not contain any point in M. Prove that if
there exists a mine-avoiding path, then there exist at least 2n−| M| mine-avoiding paths.
Note: A lattice point is a point ( x, y) where x and y are integers. A path of lattice points
with length n is a sequence of lattice points P0 , P1 , . . . , Pn in which any two adjacent points in
the sequence have a distance of 1 from each other.

Problem 20 (IMO SL 2010 C5). n ≥ 4 players participated in a tennis tournament. Any two
players have played exactly one game, and there was no tie game. We call a company of four
players bad if one player was defeated by the other three players, and each of these three
players won a game and lost another game among themselves. Suppose that there is no bad
company in this tournament. Let wi and li be respectively the number of wins and losses of
the i-th player. Prove that
n
∑ (wi − li )3 ≥ 0.
i =1

Problem 21 (USA TSTST 2020 P5). Let N2 denote the set of ordered pairs of positive integers.
A finite subset S of N2 is stable if whenever ( x, y) is in S, then so are all points ( x ′ , y′ ) of N2
with both x ′ ≤ x and y′ ≤ y.
Prove that if S is a stable set, then among all stable subsets of S (including the empty set
and S itself), at least half of them have an even number of elements.

Problem 22 (IMO SL 2008 C5). Let S = { x1 , x2 , . . . , xk+l } be a (k + l )-element set of real

numbers contained in the interval [0, 1]; k and l are positive integers. A k-element subset
A ⊂ S is called nice if
1 1 k+l
∑
k x ∈A
xi − ∑
l x ∈S\ A
xj ≤
2kl
i j
 
2 k+l
Prove that the number of nice subsets is at least .
k+l k

Problem 23 (IMO SL 2019 C6). Let n > 1 be an integer. Suppose we are given 2n points in
the plane such that no three of them are collinear. The points are to be labelled A1 , A2 , . . . , A2n
in some order. We then consider the 2n angles:

∠ A1 A2 A3 , ∠ A2 A3 A4 , · · · , ∠ A2n−2 A2n−1 A2n , ∠ A2n−1 A2n A1 , ∠ A2n A1 A2 .

We measure each angle in the way that gives the smallest positive value (i.e., between 0◦ and
180◦ ). Prove that there exists an ordering of the given points such that the resulting 2n angles

12
Kazi Aryan Amin (INMOTC Karnataka, December 26th, 2023) Intro to Combinatorics

can be separated into two groups with the sum of one group of angles equal to the sum of
the other group.

Problem 24 (IMO 2005 P6). In a mathematical competition, in which 6 problems were

posed to the participants, every two of these problems were solved by more than 25 of the
contestants. Moreover, no contestant solved all the 6 problems. Show that there are at least 2
contestants who solved exactly 5 problems each.

Problem 25 (RMM 2017 Problem 3). Let n be an integer greater than 1 and let X be an n-
element set. A non-empty collection of subsets A1 , ..., Ak of X is tight if the union A1 ∪ · · · ∪ Ak
is a proper subset of X and no element of X lies in exactly one of the Ai s. Find the largest
cardinality of a collection of proper non-empty subsets of X, no non-empty subcollection of
which is tight.
Note: A subset A of X is proper if A ̸= X. The sets in a collection are assumed to be distinct.
The whole collection is assumed to be a subcollection.

13
Kazi Aryan Amin (INMOTC Karnataka, December 26th, 2023) Intro to Combinatorics

7 Solutions to Examples
7.1 Example 2.1
Solution. Induct on n. Assume the result is true for n − 1. Consider the cases:

• N ≤ 2n−1 : Take the construction for n − 1 with N white sets and color all those elements
white. Color everything else black.

• N > 2n−1 . Then note that 2n − N ≤ 2n−1 . So we again take the construction for n − 1
with 2n − N white sets, recolor those sets to be black, and color everything else to be
white.

7.2 Example 2.2

Solution. (by alfenix- on aops First, remove all the non-bottleneck edges from G, so that every
remaining edge is now a bottleneck. We use strong induction on n; consider the base case
n = 1. Obviously there are no more than 100n − 100 = 0 edges.
Now consider a graph G with n vertices, and arbitrary bottleneck e with corresponding
sets A, B that partition the vertices of G. Observe that there are at most 100 edges between A
and B, and since A and B also have the property that all edges within them are bottleneck,
the number of edges in A is at most 100VA − 100, where VA is the number of vertices in A.
Similarly, the number of edges in B is at most 100VB − 100, so the total number of edges in G
is at most

100 + (100VA − 100) + (100VB − 100) = 100(VA + VB ) − 100 = 100n − 100,

as desired.

7.3 Example 3.1

Solution. We claim that the value ∑ AB∈S ( AB)2 strictly decreases. This obviously finishes.
Note that we obviously have ( AB)2 > (CD )2 . Next, note that by Appolonius theorem , we
have that for any point P ∈ S (Denote by M the midpoint of AB.) :

CD2 AB2 AB2 CD2

 
2 2
PC + PD − = PM2 = PA2 + PB2 − =⇒ PC2 + PD2 = PA2 + PB2 − −
4 4 4 4
| {z }
>0

Hence the claim holds and we are done.

7.4 Example 3.2

Solution. (by yayyups on aops) We first show that it’s possible is ab is even. WLOG, suppose
a is even. Note that we can turn

AAPn 7→ ( )( AAP)( ) Pn 7→ PAAPn−1 ,

14
Kazi Aryan Amin (INMOTC Karnataka, December 26th, 2023) Intro to Combinatorics

so by repeating this process we can turn AAPn 7→ Pn AA. Thus, by shoving the As down in
pairs of 2, we finish.
Now we show that ab odd doesn’t work. Let Ao , Ae , Po , and Pe be the numbers of apples
and pears in odd and even bowls, respectively. Then the quantity ( Ao − Ae ) − ( Po − Pe )
is invariant. Now if a, b are both odd, then it’s easy to see that this quantity is originally
1 − (−1) = 2. However, at the end, it is −1 − (1) = −2. So we arrive at a contradiction.

7.5 Example 4.1

Solution. Construct a sequence G = G0 ⊇ G1 ⊇ · · · of induced subgraphs of G as fol-
lows:

If Gi has a vertex vi of degree d(vi ) ≤ ε( Gi ), then let Gi+1 := Gi − vi . If such a vertex does
not exist, then we terminate our algorithm and set H := Gi

Note that by design, we have ε( Gi+1 ) ≥ ε( Gi ), thus ε( H ) ≥ ε( G ). The inequality δ( H ) > ε( H )

follows from the fact that the algorithm stopped at H; if the inequality wasn’t true, we could
again take a vertex of small enough degree, delete it, and keep going.

7.6 Example 4.2

Solution. (by Ankan Bhattacharya) Switching two adjacent labels changes the orientation of
exactly 98 triangles, so the parity of clockwise triangles is invariant under such a switch. By
repeatedly performing switches we may arrive at the labelling with 1, ..., 100 in counterclock-
wise order, which has no clockwise triangles.

Thus the number of clockwise triangles is always even; in particular, it can’t be 2017.

7.7 Example 4.3

Solution. We proceed by induction on n. Suppose the result holds for n vertices, and we will
prove it for n + 1 vertices. Let v1 7→ v2 7→ · · · 7→ vn be a directed path. If vn+1 7→ v1 or
vn 7→ vn+1 , then we are done, as we can extend the path in the previous line naturally. So
assume WLOG that we have v1 7→ vn+1 and vn+1 7→ vn . Look at the direction of the edge
vn+1 to vi for 1 ≤ i ≤ n. It is • 7→ vn+1 for i = 1 and vn+1 7→ •, for i = n. So by discrete
continuity, there is some index i, such that vi 7→ vn+1 and vn+1 7→ vi+1 . Then the path

v 1 7 → · · · 7 → v i 7 → v n +1 7 → v i +1 7 → · · · 7 → v n

works.

7.8 Example 4.4

Solution. Let f (n) denote the number of special numbers in the set {n, n + 1, · · · , n + 2013}.
Note that f (n + 1) − f (n) = {−1, 0, +1}.

Now note that f (1) < 2012, as 1, 2, 3, 4, 5 aren’t special. Also note that if L = 2lcm(2,3,4,··· ,2015) +
2, then f ( L) = 2014. Thus as we "slide" from 1 to L, by discrete continuity, there exists some
1 < n < L, such that f (n) = 2012.
15
Kazi Aryan Amin (INMOTC Karnataka, December 26th, 2023) Intro to Combinatorics

7.9 Example 5.1

Solution. Let | E| = m. We begin by making the observation that for any u, v ∈ V, such that
uv ∈ E, we have that u, v have no common neighbors. This leads to the inequality :

(deg u − 1) + (deg v − 1) ≤ n − 2 =⇒ deg u + deg v ≤ n

Summing over all edges gives us :
!2  2
Jensen deg v 2m
mn ≥ ∑ deg u + deg v = ∑ (deg v) 2
≥ n ∑ n
=n
n
uv∈ E v ∈V v ∈V

n2
This rearranges to give m ≤ 4 as desired.

7.10 Example 5.2

Solution. Suppose n people start at each of the n vertices, and suppose they follow the
following algorithm :

Suppose we start at A0 . For every k, consider all edges of the form Ak X, which haven’t
been already chosen, and choose Ak+1 = X such that the weight of Ak X is lower than that
of Ak−1 Ak and its value is maximal.

We claim that for every edge ab, there is exactly one trail containing a 7→ b, and similarly,
exactly one trail containing b 7→ a. Suppose A0 = A and A1 = B. For all k ≥ 1, consider
all edges of the form Ak X such that their edge weights are greater than that of Ak−1 Ak and
choose Ak+1 = X to be such that Ak X is minimal. Note that this process must end at some
point. Then we have the trail Ak 7→ Ak−1 7→ · · · 7→ A1 7→ A0 , so the claim holds.
Hence note that the total path length covered by the n people is 2 · (n2 ), hence by PHP,
n
2( )
someone has a path length of at least n2 = n − 1. We are done. □.

Remark 7.1. Note that the presence of a Kn is a sort of a distraction here. In particular the problem
holds with n − 1 replaced by average degree for any generic graph G.

7.11 Example 5.3

Solution. Let B and G denote the set of boys and girls respectively. For boys b and girls g,
define the function f (b, g) to be 0 if b knows g and 1 otherwise. Then notice that for a subset
X of the boys, the quantity !
∏ 1− ∏ f (b, g)
g∈ G b∈ X

equals 1, iff X is sociable, and equals 0 otherwise. Now we have:

!
∑ 1= ∑ ∏ 1− ∏ f (b, g)
X sociable X ⊆ B g∈ G b∈ X

16
Kazi Aryan Amin (INMOTC Karnataka, December 26th, 2023) Intro to Combinatorics

The RHS equals the following quantity (mod 2):

!
∑ ∏ 1+ ∏ f (b, g)
X ⊆ B g∈ G b∈ X

But this equals

∑ ∑ ∏∏ f (b, g)
X ⊆ B Y ⊆ G b ∈ X g ∈Y

However, this is symmetric in B, G, so we are done.

17