| 17#17 | (1) |
The union of all the cores 10#10
may reasonably be called the core of 1#1
and written 19#19.
In fact,
A group element 1#1 of Tn is one that lies in maximal subgroup of Tn. It is well-known that 1#1 is a group element if and only if the image 22#22 is a transversal of the equivalence 23#23 ( = 24#24. Equivalently, 1#1 is a group element of Tn if and only if 25#25is one-to-one.
The concept of a group element is related to that of 1#1-height by the following result.
Proof: It is clear that 26#26 and that 27#27 is one-to-one. If every x in Xn has 1#1-height 0 or 1, then 28#28; hence, 25#25 is one-to-one and so 1#1 is a group element.
Conversely, suppose that there exists x in Xn with 1#1-height
29#29.
Let x be in a component 9#9
(say), where
30#30.
Then
Thus, 25#25 is not one-to-one and so 1#1 is not a group element. 32#32
The symmetric group Sn is of course contained in Tn, but we shall here be concerned with 33#33, the subsemigroup of Tn consisting of all singular self-maps of Xn. We shall denote this semigroup by Singn.
For 1#1 in Singn we define 34#34 to be the number of cyclic components in 7#7 and 35#35 to be the number of fixed points of 1#1. In fact 35#35 ( 36#36 ) is the number of acyclic components plus the number of singleton components.
Let E denote the set of idempotents in Singn and let E1be the subset of E consisting of all idempotents 1#1
such that
37#37.
It has been known for some
time[2] that E, and even E1, is a set of
generators for Singn:
In [5] and [2] the function
39#39
given by
Much more recently Tatsuhiko Saito has shown[6] that 3#3
is relevant also to the more general problem of expressing 1#1
as a product of elements of E. Let 45#45
(the defect of
1#1)
be defined by
46#46,
and let 19#19
be defined by the property that
47#47,
48#48Saito's theorem states that
In this paper we begin by exploring some arithmetical consequences of Saito's results. Then we examine the function 3#3 more closely, obtaining some formulae in relatively special cases.