iOA of type OA(k,n)-OA(k,u)-2.OA(k,1)

[ncohen]

This branch extends a previous case that only worked for three holes of size 1.

The algorithm is greedy but the implementation is a bit painful :-/

Nathann

[git]

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

​28038c2	trac #17149: iOA with big holes through product of OA
​478c658	trac #17149: y->sum_of_holes, x->number_of_holes, holes_sizes->holes
​a9e7c28	trac #17149: Code cleaning
​c3f7ddf	trac #17149: OA(10,1620)
​2330e61	trac #17149: OA(9,1078)
​da42199	trac #17149: OA(9,1612)
​6746c80	trac #17149: Merged with 6.5.beta5
​e4d1622	trac #17594: Holes of size OA(k,n)-OA(k,u)-2.OA(k,1)

[git]

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

​cfd7171

trac #17594: Add an URL

[vdelecroix]

Error building the documentation.
Traceback (most recent call last):
...
OSError: [combinat ] sage/combinat/designs/orthogonal_arrays.py:docstring of sage.combinat.designs.orthogonal_arrays.OAMainFunctions.is_available:6: WARNING: Block quote ends without a blank line; unexpected unindent.

[vdelecroix]

I found the pb (there was an unexpected "x" at the begining of a line!!). I am on it.

Vincent

[git]

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

​b30b420	trac #17594: fix doc build
​311cd17	trac #17594: simplification in iOA

[vdelecroix]

Hello Nathann,

It seems to work but we do have very few examples of OA(k,n) - OA(k,u), right?

sage: for n in range(2,1000):
....:     if designs.incomplete_orthogonal_array(4,n,[2],existence=True):
....:         print n
9

Vincent

[ncohen]

Hello !

Sorry for this 'x': actually, I already fixed that in this ticket's dependency #17149 (it is closed but not yet in a beta)

It seems to work but we do have very few examples of OA(k,n) - OA(k,u), right?

sage: for n in range(2,1000):
....:     if designs.incomplete_orthogonal_array(4,n,[2],existence=True):
....:         print n
9

With this code you only test OA(k,n)-OA(k,2). Try this:

sage: for i in range(2,1000):
....:     for j in range(2,i):
....:         if designs.incomplete_orthogonal_array(4,i,[j],existence=True):
....:             print "OA({},{})-OA({},{})".format(4,i,4,j)
....:             
OA(4,9)-OA(4,2)
OA(4,16)-OA(4,3)
OA(4,17)-OA(4,4)
OA(4,25)-OA(4,6)
OA(4,26)-OA(4,5)
OA(4,30)-OA(4,5)
OA(4,36)-OA(4,5)
OA(4,36)-OA(4,7)
OA(4,37)-OA(4,6)
OA(4,41)-OA(4,10)
OA(4,42)-OA(4,7)
...

If you have a hole of size 3 in an OA, you do not necessarily have a hole of size 2. It's not a -2.OA(k,1), it's a -OA(k,2)!

Nathann

[ncohen]

Ahahahah. I am ashamed. Yeah, your code is much simpler :-P

[ncohen]

Do we get this in ? I see nothing wrong with your commit, so if mine is fine for you ?...

Nathann

[vdelecroix]

It is ok for me.

No conflict to expect with #17149?

Vincent

[git]

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

​89bd369	trac #17149: Merged with 6.5.beta5
​9fd2e69	trac #17149: Broken doc
​b1ef60c	trac #17594: Holes of size OA(k,n)-OA(k,u)-2.OA(k,1)
​4f2cd65	trac #17594: Add an URL
​5934845	trac #17594: simplification in iOA

[vdelecroix]

... no more conflicts ;-)

[ncohen]

Rebasing is so cool. Thanks for the review!

Nathann