As you see, I did a finite semigroup and then try to find its possible ideals in a very basic way (with the hand of a friend):

  > f:=FreeSemigroup("a","b");;  
    a:=f.1;;  b:=f.2;; 
    w3 := Combinations(w2);;
    for i in [1..Size(w3)] do
      w3[i] := DuplicateFreeList(Flat(w3[i]));;
    w3 := DuplicateFreeList(w3); 

 > [ [  ], [ b*a*b, b, b^2, b*a, b*a^2 ], [ a, a^2, a*b, a^2*b ] ]

Do you think that above codes suffice to make them visible appropriately? Thanks for your attention!

  • $\begingroup$ I am a bit confused but maybe I am misintrepreting the Gap code. It seems that $ba = ab^3 = ab^2b = aa^2b = a^3b = ab$. Thus your semigroup is commutative. On the other hand, you have both ab and ba in the final line, which seems to indicate that these elements are distinct. $\endgroup$ – J.-E. Pin Dec 5 '15 at 19:57
  • $\begingroup$ Yes! That's right. Maybe it happens cause of using an old version of GAP. I usualy use Alexander Hulpke's windows frendly one[math.colostate.edu/~hulpke/CGT/education.html]. Thanks for the point. :-) $\endgroup$ – mrs Dec 6 '15 at 19:02
  • $\begingroup$ @J.-E.Pin GAP recognises here that $ab=ba$. If you call w4:=DuplicateFreeList(Flat(w3)); it will return [ b*a*b, b, b^2, b*a, b*a^2 ]. The code in question simply have $ab$ and $ba$ in different sublists. $\endgroup$ – Alexander Konovalov Dec 11 '15 at 20:56
  • $\begingroup$ @s-snape Please do so, but I might be a bit slow to answer if the question is too difficult... $\endgroup$ – J.-E. Pin Mar 28 '16 at 13:30
  • $\begingroup$ @s-snape Please delete this comment. $\endgroup$ – J.-E. Pin Mar 28 '16 at 21:56

A bit of theory, for finite semigroups:

  • every ideal in a semigroup is a union of principal ideals (the least ideal containing a given single element)
  • two elements in a finite semigroup generate the same principal ideal if and only if they are $\mathscr{D}$-related (this is the definition of being $\mathscr{D}$-related).
  • the relation $\mathscr{D}$ is an equivalence relation, and its equivalence classes are called $\mathscr{D}$-classes
  • if $A$ and $B$ are $\mathscr{D}$-classes, then we say that $A \leq_{\mathscr{D}} B$ if the the principal ideal generated by any (and all) elements in $A$ is contained in the principal ideal generated by any (and all) elements in $B$.

So, if you know the $\mathscr{D}$-classes of a finite semigroup, and the partial order $\leq_{\mathscr{D}}$, then you know everything about the ideals of the semigroup, and how they are interrelated.

In GAP, you can do:

T := Range(IsomorphismTransformationSemigroup(s));

you will see that your semigroup has two $\mathscr{D}$-classes and hence two ideals. If you have the Semigroups package, then you can also do:


and obtain the following picture of the $\mathscr{D}$-classes and their partial order:

D-class diagram

You can also see that your semigroups is a copy of the direct product of the cyclic group of order 2 with itself, with an additional non-regular element on the top. The ideals are: the whole semigroup, and the group $C_2\times C_2$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.