I was trying to port xnor's algorithm for caird's Stirling number question (TIO), but I get different results from xnor's solution

I'm trying to use the same formula they did - $\frac{3^{n−1}−2^n+1}{2}$

Can someone help me figure out what I'm doing wrong?

² is n ^ 2

afaict, ’3*_²‘:2 decrements ⍵, 3* raises 3 to the power of that, then _² subtracts the square of ⍵, ‘ increments it, and :2 finally halves it

@hyper-neutrino Oh I'm dumb

you did not read the formula correctly :p

H is halve, by the way

you don't need the floor division

I thought I needed it because of the n^2 instead of 2^n thing :P

lol

Thanks, it works now

You can inline the helper link for -1 byte

With chain separators?

I thought those would still be two bytes because of the starting and ending thingy

nope, there is a better way of combining a few links into one link

Found it, $ and friends?

Yep

If you can generate the list of partitions in 5 bytes or less, then ...Ẉċ3 should work for 8 or less

The issue is that it's set partitions (so unordered), rather than list partitions which is a builtin

Wouldn't <all permutations+<list partitions>+<remove duplicates> work, though?

Actually, you shouldn't need to remove duplicates afaict

No, that fails. Consider S(4, 2). That counts [[1,2],[3,4]] and [[2,1],[4,3]] as different, but a set wouldn't

Oh right

I guess sort each then? That's quite verbose, though

And it's Œ!ŒṖ€Ẏ (6 bytes) to even get all partitions of all permutations

@user Guess I was wrong here

Which brings you to Œ!ŒṖ€ẎṢ€€Ṣ€QẈċ3 for 15 bytes :/

€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€€

(I meant for that to be pronounced "EEEEEEEEEEEEEEEEEEEEEEEEEEEE!!!!11111")