Very careless typoes in strongly_regular_db

[ncohen]

This branch fixes very bad typoes in that file, that made several constructions useless or broken. Mostly missing characters (emacs macros...)

Nathann

[git]

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

​a414c36	trac #19018: More SRGs using Regular Symmetric Hadamard matric with Constant Diagonal
​05b2fa8	trac #19019: Very careless typoes in strongly_regular_db

[jmantysalo]

Not relating to just this patch, but why functions like SRG_100_44_18_20(), SRG_100_45_20_20() etc. instead of something like SRG_from_db(100, 44, 18, 20)? There might be some rationale explained in some other ticket.

[git]

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

​c430b4b	trac #19018: Link the new constructions with graphs.strongly_regular_graph
​0e533bb	trac #19018: Bibliography
​4abbce4	trac #19019: Very careless typoes in strongly_regular_db

[ncohen]

Not relating to just this patch, but why functions like SRG_100_44_18_20(), SRG_100_45_20_20() etc. instead of something like SRG_from_db(100, 44, 18, 20)? There might be some rationale explained in some other ticket.

I do not understand what you mean, and I do not want the conversation to happen on this unrelated ticket. Please send me an email or write to sage-devel.

[git]

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

​c5e24be

trac #19019: Missing space

[git]

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

​6893ad1	trac #19018: Typo
​e18bee7	trac #19019: Very careless typoes in strongly_regular_db
​3f76966	trac #19019: Missing space

[git]

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

​d66f0b6	trac #19018: Specify which graph is built
​1cae2dc	trac #19028: Table 8.1
​7045124	trac #19018: Merged with 6.9.beta2
​99fa6ce	trac #19019: Very careless typoes in strongly_regular_db
​cb071e2	trac #19019: Missing space
​19b40d9	trac #19019: Sort the list of SRGs

[dimpase]

rebased over the latest developments on #19018. LGTM

[jmantysalo]

Volker will reject this if you don't add your name to Reviewers-field.

[git]

Branch pushed to git repo; I updated commit sha1 and set ticket back to needs_review. New commits:

​323416f

added a doctest

[dimpase]

Replying to git:

Branch pushed to git repo; I updated commit sha1 and set ticket back to needs_review. New commits:

​323416f

added a doctest

rebased over the latest #19018 branch

[ncohen]

What you did is not a rebase, nor a merge. You apparently cherry-picked your commit on top of that branch (note that its hash changed). As a result your commit appears twice in the history.

Please remove this latest commit: it is not because the branch are not exactly the same that they do not merge trivially. I don't think that your last commit from #19018 caused the slightest problem with this branch.

Nathann

[vbraun]

Reviewer name missing

[dimpase]

I cannot find the extra commit in question.

[ncohen]

It is the last commit of this branch.

[dimpase]

In the local branch I do not have 2 copies of this commit. Naturally, if I were to remove it and push the result then it would not be there at all.

[ncohen]

In the local branch you have this commit. In the branch of #19018 you have another copy of it. Because you cherry-picked it, when the two branches will be merged you will have two copies of it.

This commit belongs to the other branch, not to that one. If you remove it from this branch, it will be in Sage when #19018 will be merged.

You should have merged the branches, not cherry-picked the commit.

Nathann

[git]

Branch pushed to git repo; I updated commit sha1 and set ticket back to needs_review. This was a forced push. New commits:

​6ed716d	added a doctest
​cb0b3fb	Merge branch 'reg_symm_hadamard' into t19019

[dimpase]

please review these git things. I'll be offline for the coming 10 hours.

[ncohen]

Looks good, thanks.

Nathann

[dimpase]

I overlooked

+    When `\epsilon\in\{-1,+1\}`, we say that `M` is a `(n,\epsilon)-RSHCD` if
+    `M` is a regular symmetric Hadamard matrix with constant diagonal
+    `\delta\in\{-1,+1\}` and row values all equal to `\delta \epsilon
+    \sqrt(n)`. For more information, see [HX10]_ or 10.5.1 in
+    [BH12]_.

• what are "row values"? row sums?
• the definition of RSHCD here does not match one in [BH12], where the diagonal entries are all 1, and (n,e)-RSHCD has row sums e\sqrt{n}.
• it should be \sqrt{n}, not \sqrt(n)

And another thing - RSHCD should be mentioned in the module description in the beginning of the corr. file.

[ncohen]

• what are "row values"? row sums?

Arg... Yes. Row sums.

• the definition of RSHCD here does not match one in [BH12], where the diagonal entries are all 1, and (n,e)-RSHCD has row sums e\sqrt{n}.

So what do we do? With two papers that say contradictory things?

• it should be \sqrt{n}, not \sqrt(n)

Right.

And another thing - RSHCD should be mentioned in the module description in the beginning of the corr. file.

I don't understand.

Nathann

[jmantysalo]

Replying to ncohen:

So what do we do? With two papers that say contradictory things?

IMO select one and (almost) always document these cases. See chain_polynomial() on posets: "Warning: This is not what has been called the chain polynomial in [St1986]. The latter is - -".

[ncohen]

IMO select one and (almost) always document these cases. See chain_polynomial() on posets: "Warning: This is not what has been called the chain polynomial in [St1986]. The latter is - -".

Jori, this was a rethoric question. The two definitions agree with each other.

[dimpase]

Replying to ncohen:

• the definition of RSHCD here does not match one in [BH12], where the diagonal entries are all 1, and (n,e)-RSHCD has row sums e\sqrt{n}.

So what do we do? With two papers that say contradictory things?

I was comparing your definition with the one from ​http://homepages.cwi.nl/~aeb/math/ipm/ipm.pdf and your definition is more complicated. I did not check the other one.

And another thing - RSHCD should be mentioned in the module description in the beginning of the corr. file.

I don't understand.

If look at combinat/matrices/hadamard_matrix.py there is a paragraph saying

The module below implements the Paley constructions (see for example
[Hora]_) and the Sylvester construction. It also allows you to pull a
Hadamard matrix from the database at [HadaSloa]_.

But there is no word about RSHCDs. Should it instead get a list of methods, etc, as more modern Sage modules have?

As well, you're not in the list of authors.

As I am gearing up for skew-Hadamard matrices, it can be done on the corresponding ticket.

[ncohen]

I was comparing your definition with the one from ​http://homepages.cwi.nl/~aeb/math/ipm/ipm.pdf and your definition is more complicated. I did not check the other one.

I don't see how. In the paper you give the matrix is requested to have 1 on the diagonal, that's the only difference I see.

If look at combinat/matrices/hadamard_matrix.py there is a paragraph saying But there is no word about RSHCDs.

Oh, right. I'll do that.

Should it instead get a list of methods, etc, as more modern Sage modules have?

Probably. I am working on a patch related to index of functions at the moment, though. Something to make building a thematic index easier.

As well, you're not in the list of authors.

How I hate this "I am the one who did it" bullshit...

As I am gearing up for skew-Hadamard matrices, it can be done on the corresponding ticket.

Skew-hadamard? *sigh*.... Someday we will have all kind of designs in Sage. But today is not this day.

Nathann

[dimpase]

Replying to ncohen:

I was comparing your definition with the one from ​http://homepages.cwi.nl/~aeb/math/ipm/ipm.pdf and your definition is more complicated. I did not check the other one.

I don't see how. In the paper you give the matrix is requested to have 1 on the diagonal, that's the only difference I see.

yes, this is the difference that makes your definition different. I suppose this is very close to the original one, as all is needed is multiplication by -1 to convert one into the other. But still not the same...

As well, you're not in the list of authors.

How I hate this "I am the one who did it" bullshit...

well, if only for uniformity...

As I am gearing up for skew-Hadamard matrices, it can be done on the corresponding ticket.

Skew-hadamard? *sigh*.... Someday we will have all kind of designs in Sage. But today is not this day.

we need them to build "skewhad*" and your humble servant srg's, as I already mentioned.

[ncohen]

yes, this is the difference that makes your definition different. I suppose this is very close to the original one, as all is needed is multiplication by -1 to convert one into the other. But still not the same...

For all practical purposes it is the same. One says that the diagonal has to be 1, the other says that it can be -1 but adapts the value of row sums. If you get one which does not have a 1 on the diagonal, then -your_matrix does the job, and all results of existence/non-existence apply.

Nathann