Combinatorics on the Chessboard

Interactive game:

1. On regular chessboard a rook is placed on a1 (bottom-left corner). Players

A and B take alternating turns by moving the rook upwards or to the right by
any distance (no left or down movements allowed). Player A makes the rst
move, and the winner is whoever rst reaches h8 (top-right corner). Is there a
winning strategy for any of the players?

Solution:
Player B has a winning strategy by keeping the rook on the diagonal.

Knight problems based on invariance principle:

A knight on a chessboard has a property that it moves by alternating through

black and white squares: if it is on a white square, then after 1 move it will
land on a black square, and vice versa. Sometimes this is called the chameleon
property of the knight. This is related to invariance principle, and can be used
in problems, such as:

2. A knight starts randomly moving from a1, and after n moves returns to
a1. Prove that n is even.

Solution: Note that a1 is a black square. Based on the chameleon property

the knight will be on a white square after odd number of moves, and on a black
square after even number of moves. Therefore, it can return to a1 only after
even number of moves.

3. Is it possible to move a knight from a1 to h8 by visiting each square on

the chessboard exactly once?

Solution:
Since there are 64 squares on the board, a knight would need 63 moves to
get from a1 to h8 by visiting each square exactly once. However, based on the
chameleon property, after 63 moves it has to be on a white square, while h8 is
a black square, therefore, such path is impossible.

Problems related to placing pieces on the chessboard:

4. Find the maximum number of specic chess pieces you can place on a
chessboard such that none of them is under attack. Solve it for:
a) rooks
b) queens
c) bishops
d) kings

1
 e) knights

In this class we will discuss solutions for a and e. Whoever is interested, try
solving b,c and d using similar approach. We can go over those later if there is
any interest.

For these problems rst we need to nd a cap, then need to show an example
of a conguration using that number.

In many board problems an important technique is to consider the problem

for a smaller board rst. If we can divide the board into smaller regions such
that there is a clear cap for each subregion, then by adding those numbers for
each subregion we can get a total cap.

Solution for a):

Consider each vertical line of the board (a1 to a8, b1 to b8, etc). There
are 8 lines, and clearly you can place at most 1 rook on each line. Therefore,
you can place at most 8 rooks on the chessboard. We still need to show that
it is possible to place 8 rooks on a chessboard without attacking each other.
Indeed, one such example is placing 8 rooks on a main diagonal. Therefore, the
maximum number is 8.

Solution for e):

Divide the board into eight 4 × 2 regions. Each region can have at most
4 knights, therefore, full chessboard can have at most 32 knights. And as an
example, place 32 knights on all 32 black squares (or all 32 white squares), then
none of them will attack each other. So the answer is 32. It turns out that
placing knights on same color squares are the only two ways that you can place
32 knights without attacking each other.

Below is another take home problem (harder version of 4a):

5. What's the maximum number of rooks you can place on a chessboard
such that each rook is under attack by no more than one other rook?

6. Find a minimum number of specic chess pieces needed to place on a

chessboard in a way that all free squares of the board are under attack. Solve
it for:
a) rooks
b) queens
c) bishops
d) kings
e) knights

Note that in those problems the requirement is to keep under attack only
empty squares; if there is a piece on it, then it doesn't have to be attacked.
There are versions of those problems with that extra requirement as well.

2
 In this class we will discuss b and e (a, c and d to think about at home).

Solution for b):

Consider 5 queens placed at a2, c4, d5, e6 and g8. Then note that every
square of the board will be under attack. Therefore, 5 queens are sucient for
the problem requirement. In order to show 5 is the minimum number, one needs
to show that 4 queens cannot cover the whole board. It turns out that's actually
very hard to show by hand; I wasn't able to nd any proof for this. Instead, it is
done by writing computer code that looks at various permutations of 4 queens
on a chessboard and in each case nds a square that is not under attack.

Solution for e):

Consider the 12 squares with a green line: a1, a2, b2, a7, a8, b7, g2, h1, h2, g7, h7, h8.
Note that a single knight cannot attack or cover more than 1 of those squares,
thus you need at least 12 knights to attack or cover those 12 squares. And with
12 knights you can actually attack the whole board (empty squares), illustrated
with brown rectangles. Thus the answer is 12.

Now let's look at some counting problems related to rooks.

7.
a) What's the number of ways to place 8 rooks on a chessboard such that
no rook is under attack?
b) What's the number of ways to place 8 rooks on a chessboard such that
no rook is under attack and there are no rooks placed on the main diagonal
(a1 : h8)?
c) What's the number of ways to place 8 rooks on a chessboard such that
each square of the board is under attack?

Solution for a):

Note that each row contains one rook. The rst rook (on the rst row) can
be placed on any of the 8 columns; second rook can be placed on 7 columns, etc,
and last rook can be on a single column. The total number of ways is, therefore,

3
8!. In combinatorics terms, this is equivalent to number of permutations of a
set with 8 elements.

Solution for b):

In this case we are dealing with number of derangements of a set, i.e. per-
mutations where no element appears in its original position. The number of
derangements satises the following recursive equation:
D(n) = (n − 1)(D(n − 2) + D(n − 1))
In case of rooks and a chessboard (n = 8), this can be explained as follows:
consider the rst rook; it can be placed on 7 dierent columns. Assume it is on
column k. Then consider the k-th rook. If it is placed on the rst row, then
we're left 6 rooks, which gives us D(6) derangements. If k-th rook is not on the
rst column, then we're left with D(7) derangements. Thus:
D(8) = 7 ∗ (D(6) + D(7))
By induction it can be easily proved that D(n) also satises equation:
Pn (−1)i
D(n) = n! i=0 i!
Using that formula we can get D(8) = 14833.

Solution for c):

Note that in order for each square of the board to be under attack, there
must be a rook on either every row or every column. Indeed, if there was a row
without a rook, and a column without a rook, then their intersection would not
be under attack. The number of congurations where there is 1 rook on each
row is 88 (because each one of the 8 rooks can be placed in 8 dierent places).
Similarly, the number of congurations covering each column is also 88 . And
the number of congurations where there is a rook on each row and each column
is the same as 7a, i.e. 8!. Thus, the answer is 2 ∗ 88 − 8!.

Below is another take home problem:

8. Is it possible to cover all of the chessboard squares with knight moves by
visiting each square exactly once?

Second part of the class is related to chessboards in general, without any

chess pieces involved.

Many board problems are related to tiling and coloring. Domino tiles are
pretty common:
9. Corners a1 and h8 have been removed from a chessboard. Can you cover
the rest with 1 × 2 dominoes?

Solution:
Here we use an approach similar to problems related to knight movements.
Here an important property of a domino is that it covers 1 white square and
1 black square. This is also a version of invariance principle (the dierence
between white and black cells remains constant). Note that by removing a1
and h8, we're left with 30 black squares and 32 white squares. Therefore, the

4
board cannot be covered with dominoes, because dominoes can cover only equal
amount of black and white squares.

10. Two arbitrary squares of dierent colors (i.e. 1 black square and 1 white
square) have been removed from a chessboard. Prove that the rest of the board
can always be covered with 1 × 2 dominoes.
Solution:
Below is an elegant proof by Ralph E. Gomory. Consider the following closed
path on a chessboard (the green lines):

If two squares of dierent colors have been removed, then we can travel from
the rst removed square to the second one in two directions, and each path will
contain even number of squares, thus they can be covered with dominoes, and
as a result the full board will be covered.

Now let's consider arbitrarily shaped boards (not necessarily rectangular).

It is interesting to see which boards can be covered with dominoes, and which
cannot. As we saw, if a board contains unequal number of black and white
squares, then it cannot be tiled with dominoes. So it is more interesting to
consider boards with equal number of black and white squares.
11. Can the following board be tiled with dominoes (the red square has
been removed)1 ?

5
 Solution:
Although there is an equal number of black and white squares on the board,
the answer is still no. Consider the rst two black squares of top three rows (6
squares total). We need at least 6 dominoes to cover these. However, note that
combined they have only 5 dierent neighbor squares. Thus, those 6 dominoes
would have to overlap, meaning the tilling is impossible.

In general, if on a board we can nd k squares of one color with less than
k total neighbors, then the board cannot be covered with dominoes. The proof
follows from the Pigeonhole principle.
It turns out the opposite is also true: any board or region that cannot be
tiled with dominoes contains k cells of one color with fewer than k neighbors
(proof by Philip Hall; this is a special case of marriage theorem). Thus, in order
to show that a domino tiling is impossible for a given board, we can try looking
for such cells.

Take-home problem:
12. Corner a1 has been removed from the chessboard. Can you tile the rest
with 1 × 3 "trominoes"?

Now let's consider tiling boards of rectangular shape with other rectangles.
13. Is it possible to tile2 :
a) 7 × 10 board with 2 × 3 rectangles
b) 17 × 28 board with 4 × 7 rectangles
c) 10 × 15 board with 1 × 6 rectangles

Solution:
a) There are 70 squares on the board, while each rectangle covers 6 squares,
so clearly the covering is impossible.
b) Note that the previous argument doesn't apply here. We use a dierent
approach: consider the rst row of board; it contains 17 squares. Each 4 × 7
rectangle can cover either 4 or 7 of those. However, 17 cannot be written as a
sum of fours and sevens. Therefore, even the rst row cannot be covered, which
makes the board covering impossible.

6
 c) Here the previous two arguments don't work anymore. It turns out that
tiling is still impossible, because neither 10 nor 15 is divisible by 6. This follows
from a theorem by de Bruijn and Klarner (and so do parts a and b):
An m × n board can be tiled with a × b rectangles if and only if:
• mn is divisible by ab
• both m and n can be written as ax + by where x and y are non-negative
integers
• either m or n is divisible by a, and either m or n is divisible by b

Below is another problem related to chessboard tiling:

14. A random square has been removed from a chessboard. Prove that you
can cover the rest with angled trominoes:

Solution:
As noted in some of the chess problems (e.g. 4e), it is often useful to solve
a board problem for smaller regions rst. In fact, this problem can be solved
using mathematical induction for any board with dimensions 2n ×2n (for regular
chessboard, n = 3).
- if n = 1, then we have a 2 × 2 board, and the solution is trivial
- assume the statement is true for n = k, i.e. it is possible to cover a 2k × 2k
board by removing any square. Consider a 2k+1 × 2k+1 board. By cutting it
through the middle vertically and horizontally, we get four 2k ×2k boards. When
we remove a square from the original board, it has to be in one of those smaller
boards. Based on the induction hypothesis, that board can be covered with
angled trominoes. For the remaining 3 subregions, note that you can place a
tromino on the center of the original board such that it covers 1 corner from each
of those 3 subregions. That gives us another 3 boards with 1 square missing.
Those can also be covered based on the induction hypothesis, hence the original
board can be covered too.

Below are several chessboard related problems from various AMC stages,
leading to IMO3 . Note that there is a jump in diculty level; USAMO and
IMO are generally harder than earlier stages of AMC, and there is more time
allocated for each problem.

Easy:
15. From 2009 AMC 8:
On a regular chess board, what is the probability that a randomly chosen
unit square does not touch the outer edge of the board?
16. From 2012 AMC 10A:
How many black squares are on a 31 × 31 chessboard?

7
 Hard:
17. From 1998 USAMO:
Consider a 98 × 98 chessboard on a computer screen. One can select with a
mouse any rectangle with sides on the lines of the chessboard, and switch the
colors in the selected rectangle (black becomes white, white becomes white).
Find, with proof, the minimum number of mouse clicks needed to make the
chessboard all one color.
18. From 1976 USAMO:
a) Suppose that each square of a 4 × 7 chessboard is colored either black or
white. Prove that with any such coloring, the board must contain a rectangle
whose four distinct unit corner squares are all of the same color.
b) Show that the above statement is false for a 4 × 6 chessboard, i.e. it can
be colored in a way that the four corner squares of every rectangle are not all
of the same color.
19. From 2014 IMO:
Consider an n × n chessboard, where n ≥ 2. A conguration of n rooks is
called peaceful if every row and every column contains exactly one rook. Find
the greatest positive integer k such that, for each peaceful conguration of n
rooks, there is a k × k square which does not contain a rook.

If there is any interest, we can go through the solutions of those problems

in other classes.

References
[1], [2]: Federico Ardila, Richard P. Stanley: Tilings
[3] artofproblemsolving.com

Literature
Evgeny Gik. Math and Chess