New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
OA for n=1046,1059,2164,3992,3994 #16662
Comments
Branch: u/ncohen/16662 |
Branch pushed to git repo; I updated commit sha1. Last 10 new commits:
|
Commit: |
comment:5
Hello
Are a,b,c allowed to be 1? In the examples of the article a,b,c are never 1 (but d might be).
For the other one, the existence of an OA(9,21) should not be necessary for the construction
Vincent |
comment:6
Hello !
Is it me or is the git server slow ? It takes MINUTES to download a branch
How, not "valid" ? The input is valid but in order to build the new designs you need some sub-designs, and among the subdesigns there is a
I don't think that there is anything wrong with =1 values.
No, it is just that there exists no
The exception I added for the previous one is also triggered here.
Nathann |
comment:9
Hi, You add a check at the begining but there is no associated specification in the doc... Vincent |
comment:10
Very bad... I was not able to build the potentially new
|
comment:11
Well you can't build a This deepcopy is a crime. Nathann |
comment:12
I said that because this one is smaller than the largest you add to the database: an OA(12,3994). I will create a ticket to remove all deepcopy around (and in that case I think it is more like a Vincent |
comment:13
Yoooooooo !
Please don't, I fixed that in at least two Nathann |
comment:14
All right... I put constructions for OA(10,1048), OA(11,1524) and OA(12,3362) at Vincent |
comment:15
Yo !
Oh. Are they the only other MOLS that we can build ? If there are more it may be worth writing a find function after all... Nathann |
comment:16
Replying to @nathanncohen:
Seems to be the only one. If you want to write this find function that's also an option. But I am not sure you will ever get the
Vincent |
comment:17
Yo !
I tried several times and it really was complicated. And required a lot of copy/paste.
Well, you need to fill the cache at first, like for all constructions.
Why do you care so much about not having useless constructions ? It is good to have several ways to generate the same design. Some paths may be better than others, some paths may result in designs with different properties...
The problem with that is that it is true with the current OA that Sage can build. Wait a bit and new one will appear that may change that. Nathann |
comment:18
Replying to @nathanncohen:
So let us not do that.
All right. Then there are thousands of thwarts construction that you may want to consider...
This is cool. If a ticket creates new OA that allows more of this construction, then it will create an error in a doctest and the database will be automatically updated. Nevertheless, filling the cache up to 3000 is too long for a doctest. So I can keep my file under my pillow and run it from time to time. Vincent |
comment:30
Helloooooooo Vincent ! Okay... So let's do that:
What it stores: a set of integers defined by a boolean function Of course we want to minimize the number of boolean function queries. Even though it takes spaces, I am thinking of something like that: An array which associates to (any) integer n: a) if f(n) is computed: the smallest integer n'>=n such that f(n') is True or has not been computed yet. b) if f(n) is not computed yet: None This way, if you want to get all solutions to
This may be cool if we ever implement the And we could use it in conjunction with a real function to solve the partition problem, i.e. "try to find integers from a set S whose sum is equal or lequal to C". Anyway.
Your code is good otherwise. Nathann |
comment:32
Hello, Replying to @nathanncohen:
+1
And:
The problem with your approach is that you store a lot of data. It is perhaps not as bad as what I imagine. Questions:
So the bound should be
I edit my commit at Vincent |
comment:33
Yooooooo !
That's easy to do. It would be the result of all_n(min=current_value,max=None).next()
This would be harder, but I'm not sure it is needed either
it bothered me at first but we are thinking of k lists (and k<50 usually) or size around 2000 at most. How bad can that be ? If it were stored in C it would represent 500ko
Argg... Because I thought we could assume
I just downloaded your branch but did not see anything new. I just merged it with the beta0 but I will add a commit in a second ! Nathann |
comment:34
Replying to @nathanncohen:
I edited my commit. I did not add a commit. My last commit in my branch is 68a2d99 and not 2f936c5. Vincent |
comment:35
Yo !
Oh, I see. I think that this N-2 is correct after all, however. If n=N-2, then it means that there are at most two points in the last 4 columns, i.e. that at most two columns are used. And because all pairs of points are equivalent, this is already handled by the wilson construction with two truncated columns. I had written some code which I hoped would improve the performances of this function, but well... It produced no difference whatsoever. And I don't know why yet, so let's get this in. Perhaps the cached "all_n" function will change something Nathann |
comment:36
Replying to @nathanncohen:
I believe it but please write it in a comment!
I hope it will. Vincent |
comment:38
Great great... and of lot of rebase for the follow-up tickets! Vincent |
comment:39
Thanks ! Nathann |
comment:40
There is a (simple) conflict with #16604. Vincent |
comment:41
see at Vincent |
Branch pushed to git repo; I updated commit sha1. Last 10 new commits:
|
Reviewer: Vincent Delecroix |
Changed branch from u/ncohen/16662 to |
As the title says ! New MOLS from another paper.
Nathann
Depends on #16604
CC: @videlec
Component: combinatorial designs
Author: Nathann Cohen
Branch/Commit:
7a73e74
Reviewer: Vincent Delecroix
Issue created by migration from https://trac.sagemath.org/ticket/16662
The text was updated successfully, but these errors were encountered: