New difference family (25,4,1) and cleanup in bibd.py

[vdelecroix]

In the construction of Stinson book the BIBD(25,4,1) comes from a difference family so we can just add it to the database. Moreover the function PBD_4_5_8_9_12 previously failed for v=4,5,8,9,12 where the trivial answer [range(v)] just fits.

[vdelecroix]

All right, this is rather trivial, but it cleans a bit the bibd stuff.

[vdelecroix]

Last 10 new commits:

​f0474bb	trac #16464: be more careful with errors
​21cb1a8	trac #16464: handle correctly arguments for cyclotomic_cosets
​b07ad01	trac #16464: Second review
​136e870	trac #16464: tiny improvements for is_a_splitting
​965f550	trac #16464: replace bool(S1&S2) by S1.isdisjoint(S2)
​9b199c7	trac #16461: merge #16464
​2d12cc0	trac #16461: adapt the implementation with #16464
​d3d6a31	trac #16461: reset finite_field_base.pyx to its previous state
​2638697	trac #16461: always check the difference family
​8b1e6ce	trac #16461: get rid of _nonzero_and_have_distinct_images

[vdelecroix]

Last 10 new commits:

​90f828f	trac #16391: use OrthogonalArrayBlockGraph instead of OrthogonalArrayGraph
​6e3c685	trac #16391: reviewer's comments
​52c2de8	trac #16391: Removing the holes
​d0885f7	trac #16391: bugfix
​c1de597	trac #16391: doc clean + remove some list construction
​41a9a24	trac #16391: Typo
​42239cc	trac #16391: Undoing stuff
​24ac3a8	trac #16391: Broken doc
​70c9805	trac #16391: Merged with #16460
​ee3b035	trac #16461: Merged with #16391

[git]

Branch pushed to git repo; I updated commit sha1. New commits:

​868e0b0

trac #16476: new (25,4,1) DF and clean bibd.py

[ncohen]

Good to go ! I wondered about that though... Do we care ? Should we check that the set of block sizes contains integer > 1 ?

sage: from sage.combinat.designs.bibd import _check_pbd
sage: _check_pbd([[1],[]],1,[1,0])
...
RuntimeError: This is not a nice honest PBD from the good old days!

Nathann

[vdelecroix]

Replying to ncohen:

Good to go ! I wondered about that though... Do we care ? Should we check that the set of block sizes contains integer > 1 ?

sage: from sage.combinat.designs.bibd import _check_pbd
sage: _check_pbd([[1],[]],1,[1,0])
...
RuntimeError: This is not a nice honest PBD from the good old days!

Actually, it would make more sense to have is_pairwise_balanced_design in designs_pyx.pyx... and testing that blocks are non-empty certainly needed! What do you want do be added to that ticket?

[ncohen]

Actually, it would make more sense to have is_pairwise_balanced_design in designs_pyx.pyx... and testing that blocks are non-empty certainly needed! What do you want do be added to that ticket?

I will add a commit in a second to check this, before we move it to another file.

In the meantime could you check the doc of #16461 ? It does not build on my computer and Volker will set it back to needs_work.

Nathann

[ncohen]

New commit at public/16476.

Nathann

[vdelecroix]

Block of size 1 are not allowed? The Handbook says "Let K be a subset of positive integers"...

[ncohen]

Block of size 1 are not allowed? The Handbook says "Let K be a subset of positive integers"...

What do we do with the PBD with v=1 or v=0 and many sets of size 0,1 ?

Nathann

[vdelecroix]

Replying to ncohen:

Block of size 1 are not allowed? The Handbook says "Let K be a subset of positive integers"...

What do we do with the PBD with v=1 or v=0 and many sets of size 0,1 ?

Yup. You are right. Makes more sense to exclude them.

Are we happy with this?

[ncohen]

Are we happy with this?

Are you willing to add my commit ?

Nathann

[vdelecroix]

Of course, we take it!

---

New commits:

​032ee82

trac #16476: Additional check