Combinations and Permutations

Tom Davis
[email protected]
http://www.geometer.org/mathcircles
October 25, 2000

Permutations: A permutation  is an arrangement of things. The number of ordered ways to choose things from a set of
is


  

   
   . This is because the first object can be chosen in any of
ways, the second in any of

ways, and so on.
Example: How many ways are there to choose an ordered set of  numbers from a set of  ? Answer:     . Suppose
the numbers are      and  . The possibilities are listed below. Note that it is very convenient to organize a list like this in
some order to be certain you haven’t left anything out. In this case, they’re arranged in a sort of “alphabetical order”, as if
      were     in the alphabet:
                                                                      

Combinations: Combinations are like permutations except that they are unordered. In the example above, the combinations
of 4 things taken two at a time would not include both  and   . To get the number of combinations of
things taken at a

time, we must divide the number of permutations by to get rid of duplicate permutations.
Thus, the number of combinations of
things taken at a time is:
  "


  
!
    


#     

  


This formula works fine for
%$ and/or %$ , as long as you remember that $ & .
Example: The number of ways to choose  numbers from the set '       (  ) * is + ,- ./
 $ . The possibilities include.
Notice that this list is also in “alphabetical order”:

      (   )       (    )    (    )   ( )      (   )   (   )  ( )    (    )   ( )   ( ) 

Relationship with Pascal’s Triangle: Here is Pascal’s triangle, written both in the usual way, and written with its terms
expressed as combinations.

+0 . 
+ 1 .2 0 +1.  3 
0 + 4 .2 1
- + 04 .5 - 1 - + 44 . - 6 7
+ .5+ .2+ .8+ - . 639
+ : .20 + : .5 1 + : .24 + :- .8+ : . ;<)=>
0 1 4 :
In other words, the entries in Pascal’s triangle equal to the corresponding entry in the triangle of combination coefficients on
the left.

Relationship of Pascal’s Triangle with the binomial theorem: Note that in the following expansion, the coefficients of the

1
terms correspond exactly to the fourth row (starting at row zero) of Pascal’s Triangle:
? @A
B C DFE&G @ HA!I @ J B KAL @ M B MNA!I @ K B JNA%G B H O

The general expansion of a binomial to a power is given by:

? @AB C DFE/P Q R S@ DATP Q S @ D UVK BAWP Q X S@ D UM B MNA%Y Y Y ATPZQ S@ B D UVK\ATP Q SB DO
G G
Q[ Q
? @A!B C J ? @A
B C
? @AB never
If you’ve C H done it, a good exercise is to start with, say, the expansion of and multiply that by hand by to
get . You’ll then see exactly why the rule for calculating entries in Pascal’s Triangle corresponds to the combination
coefficients.

Counting Tricks: If you have no idea how to proceed, it’s often a good idea to count or calculate (very carefully) the first
? C
few cases, and then try to work from that. Here’s a simple example of that, together with a new technique for analysis. The
problem is to find a formula for the sum of the first Q squares. In other words, find a formula for ] Q , where
X
? Q CNE&G MA  M A^ MA%Y Y Y A Q M O
]
R R X
? CE ? G CNE_G ?  C E&R GA!IE%` ? ^ CNE_GA!IA
aE&G I
In this case, it’s easy to work out the first few cases: ] ,] ,] E ,] , and so
on. Here’s a list of the first few (be sure to start your row with the value for Q ):
R Gb`_G Ic^ R ` `da GG I RcX R I X e `

Now take differences of the numbers in this row, forming a second row. Then take differences in the second row, forming a
third, and so on:
R f
G `gG Ih^ R X ` `ia GjG I RiX R I e X e `
GkIlamG L `i^ LnI a3L I G
^o q
` r
p s
a
XcX=X<X3XuXvX G i
G G 9
^ G t
` G p
RwR=R<R3RxR

If you’re lucky and theR last

y G y ^ row
y X is all zeroes, it’s obvious that you can generate the entire
X top row from the numbers on the
diagonal (in this case, ). You can write them on the diagonal, extend the final forever, and fill in any other term by
adding the term to its left to the term below and to the left.
But there’s also a formula. In this case, it is:
R P Q R SA%GVP Q SA
^P Q X SA X P Q S
G ^
X X X AG C
R ^ ? G C ? G C ? C ? %
A G C ?
E AG Q A Q QXz [ A Q Q[ L Q[ EWQ Q
L Q
O

E&G y X y —
Counting Mountains ^ y Catalan
I Numbers: Using Q pairs of the characters { and | , build as many chains of mountains as
you can. For Q and here are the possibilities:

} ~ { €

2
 ‚
ƒ „ … †‚‡ ‚  ‚

‚
 ‚ˆ ‚
†  ‚ˆ ‚  ‚
ƒ „‰ ˆ
 †
‚ ‡
 ‚  †
‚  ‚ †
 ‚ ˆ‚† ‚  ‚  ‚
†

‚
 ‚
† ‚‚ ‚ ‚ ‚
ˆ‚ˆˆ‚ˆ†‚  ‚ˆ†‚  ‚ †ˆ ‚  ‚  ‚  ‚Š ‚  ‚
ƒ „‹  ‡
‚  ‡
‚  ‚ ‡ ˆ
 ‚  †
‚  ‚ ‡  ‚ Š‚  ‚
‡
‚
 ‚† ‚ˆ ‚ ‡‚ ‚‚ ‚ ‚
†‚ †‚‡†‚  ‚  ‚† ‚ Š‚‡ ‚ Š‚‡ ‚ †‚  ‚† ‚  ‚ Œ‡ ‚ ‚‚‚‚

So there are 1, 2, 5, and 14 ways to construct legal mountain chains for 1, 2, 3, and 4 pairs of the slash characters. How do
we count these?
The answer is  Ž  
Ž
  
 ‘  ’“
n
  
way 
The
F ! to” seeŽ Fthis
  is to imagine all possible arrangements
 of up-strokes and only down-strokes. From among these
 › them to be up-strokes, so there are • – — —b
strokes, we need to choose ”%of ˜V™ choices. But imagine putting
š
these stroke sequences next to each other. For , look at the example “up down down up up up down”:
‚
 ‚  ‚†
 ‚  ‚†ˆ‚ 
 ‚  ‚†ˆ‚ 
 ‚  ‚Œˆ‚ 
 ‚†ˆ‚ 
‚


  as above, and all will have aŽ lowest
All of these will form a sort of staircase
  point marking the lowest edge. Since the low
point can occur at any of the positions, we must divide by .
These are known as the Catalan numbers, and they come up in many counting problems.”%Is› it obvious why the number of
 
valid arrangements of pairs or parentheses is also given by the Catalan numbers? For :
œ œ œ   žœ œ  œ žœ œ œ  Ÿœ œ œ  žœ œ œ 