@rabbitgrowth - I should really review the whole paper, but I have a few preliminary fixes to my tree renderings here.

@doug The link doesn't seem to work. It asks me to select the language, and when I do, I just get an empty "Code" field.

Grrr…. maybe something changed on ATO. How is this

Works now. Thank you!

is there a nice way to write dzaima's ⍁ (map diagonals of a matrix) in dyalog?

@RubenVerg ⎕UCS 9025??

@RubenVerg 1 1∘⍉

Ah I see that ⍁ is more complicated than that,

@B.Wilson lmao

I have {↑⍺⍺¨(1 1⍉⌽)¨(⊢∪⍥(↑∘⍵¨)⌽∘-)(⍴⍴⍵)/¨⍳⌈/⍴⍵} but it's far from optimal

@j63nu ye so like ⊂⍁3 3⍴⍳9 is (,1)(2 4)(3 5 7)(6 8)(,9)

@j63nu nice thanks, will try to improve it

{((⍺⍺1 1⍉⌽)↑∘⍵)⍤¯1⊢(⍴⍴⍵)/⍪(⊢,1↓-∘⌽)⍳≢⍵}

I believe {((⍺⍺(R/1)⍉⌽)↑∘⍵)⍤¯1⊢(R←⍴⍴⍵)/⍪(⊢,1↓-∘⌽)⍳≢⍵} works on n-dimensions but can't verify ATM

'Hello'(∊,⍤(∘.,)⌷⍨∘⊂(⍋∘∊∘(∘.+/)⍳¨⍤,)⍥≢)'World'

alright, officially published TinyAPL Part 1: Introduction & Arrays, now with some more explanation between code

(cc @Adám)

Diagonals: (,⊂⍤⊢⌸⍨∘(∊∘(∘.+⌿)⍳¨)⍴)5+5 3⍴⍳15

Nice insight!

Can also be written as ,⊢∘⊂⌸⍨∘∊∘.+⍥⍳⌿∘⍴

Neat

Something similar can probably golf the previous solution: codegolf.stackexchange.com/q/263716/111844

@doug How'd you come up with that?

It's fairly well known that when you add the coordinates of a 2x2 matrix that the numbers along the (anti-)diagonals are constant. It's even described in that problem I linked.

I see, thanks

@RubenVerg Oh, you actually meant perform the operation. lol. I'm an idiot. Anyway, here's a thing: {(∊⍵)⊢∘⊂⌸⍨∊+/¨⍳⍴⍵}

Oh, @j63nu already went there. Nevermind.

Oh, right. I forgot that iota over longer vectors works like that. I think mine can be fixed with this too.

Oh, sorry. That's exactly what @j63nu did.

(By 2x2, I meant 2d...)

@Adám That's helpful to know.

@RubenVerg Thanks, added to the news.

@Adám There's a missing opening quote mark on this entry.

@Silas no, stray closing, but fixed. Thanks.

ah, yes that makes more sense - thought the section name was meant to be in quotes

welcome