Another bug in ideals triggered by trying to find maximal subsemigroups

[james-d-mitchell]

Originally reported by: James Mitchell (Bitbucket: james-d-mitchell, GitHub: james-d-mitchell)

---

#!GAP

gap> S:=[SymmetricInverseMonoid(2)];
[ <symmetric inverse semigroup on 2 pts> ]
gap> S[2]:=MaximalSubsemigroups(S[1]);
[ <partial perm monoid on 2 pts with 3 generators>, <partial perm semigroup 
   on 2 pts with 2 generators> ]
gap> S[3]:=List(S[2], MaximalSubsemigroups);
Error, no method found! For debugging hints type ?Recovery from NoMethodFound
Error, no 1st choice method found for `RhoOrb' on 1 arguments called from
RhoOrb( d ) called from
RClassReps( d[i] ) called from
PartialOrderOfDClasses( I ) called from
GeneratorsOfSemigroup( SemigroupIdeal( S, reps ) ) called from
func( C[i] ) called from
...  at line 3 of *stdin*
you can 'quit;' to quit to outer loop, or
you can 'return;' to continue
brk> 

The problem seems to be that the D-classes of I don't know that they belong to `IsInverseOpClass' and so the wrong method is used to find the RClassReps.

---

• Bitbucket: https://bitbucket.org/james-d-mitchell/semigroups/issue/107

[james-d-mitchell]

Original comment by James Mitchell (Bitbucket: james-d-mitchell, GitHub: james-d-mitchell):

---

Removing milestone: 2.3 (automated comment)

[james-d-mitchell]

Original comment by James Mitchell (Bitbucket: james-d-mitchell, GitHub: james-d-mitchell):

---

resolved Issue 107, added tests etc.

[james-d-mitchell]

Original comment by James Mitchell (Bitbucket: james-d-mitchell, GitHub: james-d-mitchell):

---

This was closed incorrectly from the content of a related commit message.

[james-d-mitchell]

Original comment by James Mitchell (Bitbucket: james-d-mitchell, GitHub: james-d-mitchell):

---

I've sort of fixed this. I have to do some more debugging.

[james-d-mitchell]

Original comment by James Mitchell (Bitbucket: james-d-mitchell, GitHub: james-d-mitchell):

---

attempting to fix Issue 107, not yet working, the ideals.tst badly broken.

[james-d-mitchell]

Original comment by wilfwilson (Bitbucket: wilfwilson, GitHub: wilfwilson):

---

I have encountered a similar (identical) bug too:

#!gap

gap> gens := [ PartialPerm( [ 1, 2, 3, 4 ], [ 3, 2, 5, 4 ] ), 
>  PartialPerm( [ 1, 2, 4 ], [ 3, 5, 4 ] ), 
>  PartialPerm( [ 1, 2, 3, 4 ], [ 5, 2, 3, 1 ] ), 
>  PartialPerm( [ 1, 3, 4, 5 ], [ 5, 3, 4, 1 ] ), 
>  PartialPerm( [ 1, 2, 3, 4, 5 ], [ 5, 4, 3, 2, 1 ] ) ];;
gap> S := InverseSemigroup(gens);
<inverse partial perm semigroup on 5 pts with 5 generators>
gap> S := Semigroup(S);
<partial perm semigroup on 5 pts with 8 generators>
gap> MaximalSubsemigroups(S);
#I  finding an irredundant generating set...
#I  element has larger rank than any element of semigroup.
at  8 of    8 with  2 redundant,    5 non-redundant
#I  find the D-class of each generator...
#I  finding maximal subsemigroups arising from maximal D-classes...
#I  considering D-class 3.
#I  calculating the ideal S minus D...
Error, no method found! For debugging hints type ?Recovery from NoMethodFound
Error, no 1st choice method found for `RhoOrb' on 1 arguments called from
RhoOrb( d 
 ) on line 1254 of file /Users/Wilf/Documents/GAP/pkg/semigroups/gap/greens.gi\
 called from
RClassReps( d[i] 
 ) on line 1088 of file /Users/Wilf/Documents/GAP/pkg/semigroups/gap/inverse.g\
i called from
PartialOrderOfDClasses( I 
 ) on line 153 of file /Users/Wilf/Documents/GAP/pkg/semigroups/gap/ideals-act\
ing.gi called from
GeneratorsOfSemigroup( V 
 ) on line 663 of file /Users/Wilf/Documents/GAP/pkg/semigroups/gap/maximal.gi\
 called from
<function "unknown">( <arguments> )
 called from read-eval loop at line 35 of *stdin*
you can 'quit;' to quit to outer loop, or
you can 'return;' to continue