Brouwer's separable design construction of OA

[nathanncohen]

Once more, this wouldn't have been possible without Julian R. Abel's help.

Oh, and...

This code is "commented". To say the least :-P

Depends on #16863

CC: @videlec @KPanComputes @dimpase @brettpim

Component: combinatorial designs

Author: Nathann Cohen

Branch/Commit: 63ca3d7

Reviewer: Vincent Delecroix

Issue created by migration from https://trac.sagemath.org/ticket/16780

[nathanncohen]

Branch: u/ncohen/16780

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. Last 10 new commits:

355ac2a	trac #16662: OA for n=1046,1059,2164,3992,3994
15b449c	trac #16665: New OA for n=408,600,792,856,1368,2328,...
a515dee	trac #16673: Three factors construction of MOLS
845de7a	trac #16716: OA for n=262,950
a9608c5	trac #16722: OA(17,560)
e5fc881	trac #16722: Merged with 6.3.beta8
698b704	trac #16757: Organize the V(m,t) vectors into a dictionary
0598e46	trac #16763: New OA for n=189, plus some others through Vmt vectors
792ed9d	trac #16763: Merged with 6.3.rc1
dca31d3	trac #16780: Brouwer's separable design construction of OA

[sagetrac-git]

Commit: dca31d3

[sagetrac-git]

Changed commit from dca31d3 to 5400e2a

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:

5400e2a

trac #16780: Brouwer's separable design construction of OA

[nathanncohen]

Changed dependencies from #16763 to #16863

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. This was a forced push. Last 10 new commits:

9a57f13	trac #16757: Organize the V(m,t) vectors into a dictionary
57c00f0	trac #16757: doctest simplication
424e229	trac #16763: New OA for n=189, plus some others through Vmt vectors
5ba165e	trac #16763: Complete bibliographical references
f3f644d	trac #16763: code simplification
04936aa	trac #16802: database of difference family
4bd8d69	trac #16802: Review
4434d61	trac #16802: review the review
33b4171	trac #16863: twin prime difference set
deee069	trac #16780: Brouwer's separable design construction of OA

[sagetrac-git]

Changed commit from 5400e2a to deee069

[videlec]

comment:7

Hi,

A first commit at u/vdelecroix/16780.

In the documentation of brouwer_separable_design case i) it would be nice to say before that you want to build a resolvable OA(k-1,N).

I let you do the rebase over #16859.

Vincent

[nathanncohen]

comment:8

Yoooooo !

I agree with your modifications. This patch, however, will have to be rebased not on #16859 but on #16863 (which only needs a final check). Plus I will also have to merge it with #16859 before removing the "incomplete+resolvable OA" blocks.

Nathann

[nathanncohen]

Changed branch from u/ncohen/16780 to u/vdelecroix/16780

[nathanncohen]

Changed commit from deee069 to fac1b83

[nathanncohen]

New commits:

fac1b83

trac #16780: review 1

[nathanncohen]

Changed commit from fac1b83 to 52c4fa0

[nathanncohen]

Changed branch from u/vdelecroix/16780 to public/16780

[nathanncohen]

Last 10 new commits:

2297d31	trac #16780: Brouwer's separable design construction of OA
9b449fb	trac #16780: review 1
b2f1c8f	trac #16655: resolvable OA/TD
9103ee9	trac #16655: Merged with 6.4.beta1
cffb31d	trac #16722: A note about GLPK's "performances", new arguments to change the solver and the verbosity level
5245ef6	trac #16655: review
0bad9ed	trac #16859: Resolvable incomplete orthogonal arrays
e6deb83	trac #16859: doc
db53470	trac #16780: Merged with #16859 (need the resolvable incomplete OA)
52c4fa0	trac #16780: resolvable incomplete OA are built with 2 lines

[videlec]

comment:11

In my last commit (at public/16780):

• added in the doc: in case i) x=0, we construct a resolvable OA(k-1,N).
• the l.extend([f(x) for x in R]) are replaced by l.extend(f(x) for x in R) (we win at least 2ms... incredible, isn't it?)
• Exception -> RuntimeError
• the size of a matrix is given as nb_rows x nb_cols and not the contrary

I would not say that I carefully checked all constructions, but I am confident since there is a lot of doctest. I think it could go to positive review.

Vincent

[sagetrac-git]

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

7fe173a

trac #16780: review 2

[sagetrac-git]

Changed commit from 52c4fa0 to 7fe173a

[sagetrac-git]

Changed commit from 7fe173a to 931a33f

[sagetrac-git]

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

931a33f

trac #16780: repeat "parallel classes"

[nathanncohen]

comment:14

Hello !

• added in the doc: in case i) x=0, we construct a resolvable OA(k-1,N).

Good, good !

• the l.extend([f(x) for x in R]) are replaced by l.extend(f(x) for x in R) (we win at least 2ms... incredible, isn't it?)

Perhaps even twice that. Crazy.

• Exception -> RuntimeError

Good good !

• the size of a matrix is given as nb_rows x nb_cols and not the contrary

I object, but I have everybody against me. So I will not object aloud.

I would not say that I carefully checked all constructions, but I am confident since there is a lot of doctest. I think it could go to positive review.

Well, each case is tested and the code cannot return anything wrong anyway as all results are checked before that. Somehow while the design code is always very tricky there is no risk to return anything wrong.

I added a small commit to repeat "parallel classes" in a sentence, but short of this everything is cool, and it can go. Thank you very much for this review !

Nathann

[nathanncohen]

Reviewer: Vincent Delecroix

[vbraun]

comment:16

Merge conflict in src/sage/combinat/designs/database.py

[videlec]

comment:17

Replying to @vbraun:

Merge conflict in src/sage/combinat/designs/database.py

It is hardly believable... the combinatorial design tickets were linearly ordered! And I just check that it is up to date with #16802. Could you point the relevant ticket for the conflict?

Vincent

[vbraun]

comment:18

Conflict is 424e229 (#16763)

[sagetrac-git]

Changed commit from 931a33f to 63ca3d7

[sagetrac-git]

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

bb21dc0	trac #16763: Double citation
63ca3d7	trac #16780: merge #16763

[videlec]

comment:20

done...

Vincent

[vbraun]

Changed branch from public/16780 to 63ca3d7