Spring, 2012
1. PRELIMINARIES
Consider the binary operation g defined on the set S { A, B , C , D, E , F } as follows.
Number of
tables
Number of
assoc tables
2
16
3
19683
4
5
4294967296
113
3492
Notice also that the semigroup above has one right identity, namely F but no left identity.
Exercise 2. Use the program CHECKER.exe to find a semigroup having 3 elements (i.e. a 3x3
Caley table) that has exactly 3 left zeros.
Solution to Exercise 2. The following table satisfies the requirements of the exercise.
Exercise 3. Show that the table above is associative directly (without using the checker
program).
Solution to Exercise 3. The expression xg( y gz ) always simplifies to x since x is a left zero.
Similarly, the expression ( xgy )gz simplifies to ( x)gz and then to x again since x is a left zero.
Proposition 3. If S is a semigroup in which all elements are left zeros, then all elements are
also right identities.
Proof: Suppose the hypothesis is true. [WTS if x S then x is a right identity, i.e., y gx y for
all y S .] Suppose that x is an arbitrary element of S . In order to show that x is a right
identity, let y S . Then y gx y since y is a right zero. Hence x is a right identity.
Exercise 4. Use the program CHECKER.exe to find a semigroup of order 3 (i.e. a 3x3 Caley
table) that has exactly 2 left zeros.
2. SUBSEMIGROUPS
5
Consider the following Caley table (Table #1 from the original handout).
We can highlight a portion of that table as follows, showing that under the given operation, the
set T { A, B} is itself a semigroup, since T is closed under that restricted operation.
Definition 7. Suppose that ( S ,g) is a semigroup and that T S . Then (T ,g) is a subsemigroup
of ( S ,g) iff the set T is closed under the operation g. (Frequently, we will simply say that T is a
subsemigroup of S .)
Note that T is associative under the operation g since S is associative and T S . Note also
that if ( S ,g) is commutative then so is (T ,g) . It is customary to use the word order to describe
the number of elements in any particular semigroup. So, for example, the semigroup given in
Figure 7 above has order 4 and the subsemigroup T has order 2.
Exercise 5. Use the homework handout to find an example of a noncommutative semigroup that
has a commutative subsemigroup the subsemigroup should contain at least 2 elements.
In what follows, we create the subsemigroup tree for the Caley table given in Figure 7.
6
Definition 8. Suppose that ( S ,g) is a semigroup and x S . Then x is called an idempotent of
S iff xgx x .
Note that in the semigroup of Figure 8 (#18), the idempotents are A and B .
Exercise 6. Use the original handout to list all idempotents for each given semigroup.
Proposition 4. Suppose that ( S ,g) is a semigroup and I is the set of all idempotents of S . If
S is commutative, then I is a subsemigroup of S .
Proposition 5. Suppose that is a semigroup with subsemigroups M and N . Then M I N is
also a subsemigroup of ( S ,g) .
Definition 9. Suppose that ( S ,g) is a semigroup and x S . Then
xS {z : z xy for some y S } .
Proposition 6. Suppose that is a semigroup and x S . Then xS is a subsemigroup of S .
Definition 10. Suppose that ( S ,g) is a semigroup and x S . Then x n is defined inductively as
follows
x1 x
If x k has been defined for some k 1 , then x k 1 x k gx1 .
Proposition 7. Suppose that ( S ,g) is a semigroup and x S . Then x n gx1 x1 gx n for all n 1 .
Proof (by induction on n):
Since x1 gx1 x1 gx1 , the conclusion holds for k 1 .
Suppose now that x k gx1 x1 gx k for some k 1 , and consider x k 1 gx1 . By associativity
and the definition of x k 1 , x k 1 gx1 ( x1 gx k )gx1 x1 g( x k gx1 ) , and by the induction
hypothesis, x1 g( x k gx1 ) x1 g( x1 gx k ) x1 gx k 1 .
Proposition 8. Suppose that ( S ,g) is a semigroup and x S . Then x i gx j xi j for all positive
integers i, j .
Proof: Suppose the hypothesis is true. [WTS if x S and i, j are positive integers, then
x i gx j xi j ] Write j n i . Then, we want to show that x i gx n i xi ( n i ) x n . Our proof is
by induction on i .
By Definition 9 and Proposition 4, x1 gx n 1 x1 ( n 1) x n , and the conclusion holds for
k 1.