Import Multipartition functionality from JuLie

[mjrodgers]

Import Multipartition functionality from JuLie.
Fixes #4656

[fingolfin]

Thank you for getting this started! Now we need to clean this up some more

[mjrodgers]

Couple of spare things to clean up, but this should hopefully pass tests now and be ready for more serious review.

[thofma]

I vaguely remember that we used artificial iterators to make the API "stable" (even if we don't have a proper iterator yet). Maybe @joschmitt remembers?

[joschmitt]

I vaguely remember that we used artificial iterators to make the API "stable" (even if we don't have a proper iterator yet). Maybe @joschmitt remembers?

Yes, before the OSCAR 1.0 release, I turned all the return some_array into return (x for x in some_array).

[mjrodgers]

Perfect, thanks; I will use this for Multipartition until there is demand to convert to a real iterator.

…

On 31. Mar 2025, at 3:13 PM, Johannes Schmitt ***@***.***> wrote: joschmitt left a comment (oscar-system/Oscar.jl#4746) I vaguely remember that we used artificial iterators to make the API "stable" (even if we don't have a proper iterator yet). Maybe @joschmitt remembers? Yes, before the OSCAR 1.0 release, I turned all the return some_array into return (x for x in some_array). — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread. <https://github.com/joschmitt> <#4746 (comment)> <https://github.com/notifications/unsubscribe-auth/AANWIWPSIVJGI67TRPKCNGL2XE5PNAVCNFSM6AAAAABZNMGE2CVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDONRWGE4TAMJWG4> joschmitt left a comment (oscar-system/Oscar.jl#4746) <#4746 (comment)> I vaguely remember that we used artificial iterators to make the API "stable" (even if we don't have a proper iterator yet). Maybe @joschmitt <https://github.com/joschmitt> remembers? Yes, before the OSCAR 1.0 release, I turned all the return some_array into return (x for x in some_array). — Reply to this email directly, view it on GitHub <#4746 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AANWIWPSIVJGI67TRPKCNGL2XE5PNAVCNFSM6AAAAABZNMGE2CVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDONRWGE4TAMJWG4>. You are receiving this because you authored the thread.

[fingolfin]

Is this perhaps like this?

Suggested change

	5-element Vector{Multipartition{Int64}}:
	5-element Vector{Partition{Int64}}:

[fingolfin]

But CI test pass... It seems I am misunderstanding something... huh

[mjrodgers]

I agree that it seems like this should be Vector{Multipartition}, this definitely passes CI but that doesn't mean that it's what we want... I'm looking at it now.

[mjrodgers]

Ah okay, I understand the confusion. multipartitions returns ALL the multipartitions of n into r parts, so we get a Vector{Multipartition}. While each individual Multipartion looks like a Vector{Partition}. So this is all correct.

[fingolfin]

Ahah I get it now:

julia> foo = collect(MP)
5-element Vector{Multipartition{Int64}}:
 Partition{Int64}[[], [2]]
 Partition{Int64}[[], [1, 1]]
 Partition{Int64}[[1], [1]]
 Partition{Int64}[[2], []]
 Partition{Int64}[[1, 1], []]

julia> typeof(foo)
Vector{Multipartition{Int64}} (alias for Array{Multipartition{Int64}, 1})

julia> foo[1]
Partition{Int64}[[], [2]]

julia> typeof(foo[1])
Multipartition{Int64}

So my confusion is that Multipartition is printed as a Vector{Partition}. Which makes some sense, but at least for me it would be easier to understand what is going on if there was a custom show` method...

[mjrodgers]

I can do this but I'm not sure what it should look like. We have the same issue with Partition printing as Vector{Int}:

julia> P = partition(6, 4, 4, 2)
[6, 4, 4, 2]

[mjrodgers]

If I compare to what is detailed about the Combinatorics API, I should maybe have a dedicated Multipartitions type that is a wrapper around the iterator returned by multipartitions, with a dedicated show method. I think this should be pretty straightforward.

[fingolfin]

@mjrodgers are you working on that then?

[mjrodgers]

@mjrodgers are you working on that then?

After our last conversation, I thought we decided to merge this and work on the printing in a separate pull request; but I can take care of this before merging if you think that's better.

[fingolfin]

@mjrodgers ok I forgot about that, I am happy to do it like you said :-)