How come +/ is not done in place for a vector of matrices?

@rak1507 example?

@rak1507 Because the (cumulative) result is a scalar pointer array, while the argument is a vector pointer array.

@Adám not sure why that means it can't be done in place but I'll take your word for it tbh

@rak1507 creating the resulting scalar in-place would be bad for memory usage too - you'd have a long tail of unused elements

how so?

you are storing (mat1 mat2 mat3 mat4 …) as a vector, taking up N bytes, and creating the scalar you get (matRes unused unused unused …)

oh right

and since you've just done a complicated O(n) reduction, a single constant-size allocation isn't that bad

(the separate matrix additions could be done in-place)

is there a way to do (⊣ f g) (dyadic) using only composition?

⍺ f ⍺ g ⍵

No.

There's no way to use the same argument twice.

yeah, thought so

@rak1507 You don't want a return variable?

oh, is that done just by doing f←something (no, found it, thanks)

Well, something←f but yeah.

looks very suitable for apl

something like

      {(3/⍵)⍴⊂6⍴0}3
┌───────────┬───────────┬───────────┐
│0 0 0 0 0 0│0 0 0 0 0 0│0 0 0 0 0 0│
├───────────┼───────────┼───────────┤
│0 0 0 0 0 0│0 0 0 0 0 0│0 0 0 0 0 0│
├───────────┼───────────┼───────────┤
│0 0 0 0 0 0│0 0 0 0 0 0│0 0 0 0 0 0│
└───────────┴───────────┴───────────┘
┌───────────┬───────────┬───────────┐
│0 0 0 0 0 0│0 0 0 0 0 0│0 0 0 0 0 0│
├───────────┼───────────┼───────────┤
│0 0 0 0 0 0│0 0 0 0 0 0│0 0 0 0 0 0│
├───────────┼───────────┼───────────┤
│0 0 0 0 0 0│0 0 0 0 0 0│0 0 0 0 0 0│

might work

but I'm not sure how to paint the faces correctly