« first day (1473 days earlier)      last day (1478 days later) » 

Jonah
Jonah
03:03
Can someone explain what precisely is happening with the encoding behavior of box characters here? When I get them from a. by index, 2 2 shapes them correctly. But if copy/paste that output back, I need to shape by 2 6 to get the same result.
Presumably something to do with the number of bytes in the characters, but I don't understand exactly what.
Marshall
Marshall
@Jonah J is using control characters to encode box drawing characters. But these characters don't display as boxes in the IDE (because they're not), so it prints out different characters. J string literals are UTF-8, and the actual box drawing characters displayed are three bytes long.
Jonah
Jonah
@Marshall "so it prints out different characters" So J is internally doing conversion from control characters representing the boxes to the actual box characters? Where in J does this actually occur? Also your list sentence implies that J strings are basically just byte arrays and not "character aware" -- is that correct?
Marshall
Marshall
@Jonah I'm not going to dig through the J source for that conversion—it might even be in the IDE. As for J's relation with Unicode, see u:. It's not good.
Jonah
Jonah
03:23
@Marshall Thanks.
Dyalog APL
Dyalog APL
03:59
<moon-child> anybody have thoughts on a good glyph for a cap (as in capped forks)?
<moon-child> overstruck ⊣ seems appropriate, but I don't think unicode has it
<moon-child> (or crossed out|inverted ⊣, indicating the absence of a left argument. Maybe there should be a colourapl a la colorforth.github.io)
Razetime
Razetime
04:12
@DyalogAPL you mean a monadic fork right
hmm
Was trying this for the t square thing
Dyalog APL
Dyalog APL
<moon-child> @Razetime yeah, like jsoftware.com/help/dictionary/d502.htm
<moon-child> there are ⍅ and ⍡, which are somewhat warm, so I'm holding out hope
<moon-child> actually ⍆ might be good, showing the (uni-directional) direction of application
Razetime
Razetime
⍆ does look good, but it's very detailed
 
2 hours later…
Razetime
Razetime
06:38
50 rep for the shortest APL answer on this by 13 Feb 12PM IST
Dyalog APL
Dyalog APL
07:32
<kritixilithos> for this bounty codegolf.stackexchange.com/q/217690, i think a port of chrispn's k answer might prove shorter than the recursive apl one
 
1 hour later…
Razetime
Razetime
08:52
@DyalogAPL razetime.github.io/APLgolf/… this was my original attempt, pretty similar
maybe k is just more optimized for it
Dyalog APL
Dyalog APL
09:08
<kritixilithos> @razetime, can you shorten ⍸¨↓⍵∘.=⍺? what about (⊂⍬)? try also for ⍸~(≢⍺)↑⍸⍣¯1
<kritixilithos> @razetime what apl does that website run - extended?
Kamila Szewczyk
Kamila Szewczyk
09:30
okay, i'm finished with my load balancer for isolates 
bal←{{(¯1↓0,+\⍵)+⍳¨⍵}⍣(⍺>0)⊢(1↓q),⍨(⍵-+/q)+⊃q←(|⍺){+/↑(⍺⍴⌈⍺÷⍨≢⍵){(-⍺)↑¨↑∘⍵¨+\⍺}⍵,(⍺-⍺|≢⍵)⍴0}({⍵↓⍨-⊥⍨0=⍵}(⌊⊢+2÷⍨0.5≠2|⊢)⍵×⌽v÷+/v←⍺⍺¨⍳⍵)}
if i have a function with the O(n^2) complexity, and i want to run it in 4 different threads on the dataset consisting of 30 elements in ascending order with the same gap between them, then 
      ¯4(×⍨bal)30
13 9 5 3
computes the dataset layout, and after i pass a positive core number to bal, then generates indices of element to be passed to the function 
      ¯4(×⍨bal)16
7 5 3 1
      4(×⍨bal)16
┌─────────────┬────────────┬────────┬──┐
│1 2 3 4 5 6 7│8 9 10 11 12│13 14 15│16│
└─────────────┴────────────┴────────┴──┘
i used it with my PE72 solution: 
* Benchmarking "+/{_←'pco'⎕cy'dfns'⋄+/{⍵××/(1-1÷⊢)¨∪3pco⍵}¨⍵}IÏ 1+16(×⍨bal)999999"
             (ms)
 CPU (avg):  1256
 Elapsed:    8677
Razetime
Razetime
@DyalogAPL you can choose from 17, 18, dzaima, extended in the options
@DyalogAPL I'm not sure how to shorten those yet
If I have any more ideas I'll post links
 
2 hours later…
Razetime
Razetime
11:31
It's a bit hard to find where to split on for explanations.
Razetime
Razetime
12:04
APLgolf now has a simple explanation assistant.
rak1507
rak1507
13:03
@Razetime nice solution to t square
dzaima
dzaima
@Razetime ⍸¨↓⍵∘.=⍺ → ⍸¨⍺∘=¨⍵; no need for the ⊢ after ⍸⍣¯1; in extended (⊂⍬), could be ⍬⊂⍛,
⍸~(≢⍺)↑⍸⍣¯1 → (⍳≢⍺)~
Razetime
Razetime
14:08
@dzaima woah
@rak1507 I decided to actually read the question lol
rak1507
rak1507
lmao
Razetime
Razetime
@dzaima this saves a ton
@dzaima 43 bytes: {'*'@((⍳≢⍺)~⊃{⍵,⊃⍺/⍨⍺>⊃⌽⍵}/⌽⍬⊂⍛,⍸¨⍺∘=¨⍵)⊢⍺} in extended
not sure what "blank char" suffix is aaaa
 
1 hour later…
Razetime
Razetime
15:30
@rak1507 why answer in dyalog classic?
rak1507
rak1507
Oh for some reason it didn't work in extended
Couldn't be bothered figuring out what the issue was
Martmists
Martmists
15:53
Can someone explain to me why enclose returns an array with shape zilde? (Dyalog APL 18.0)

      m ← ⍳5
      m
1 2 3 4 5
      ⊂m
┌─────────┐
│1 2 3 4 5│
└─────────┘
      ⍴m
5
      ⍴⊂m
sorry for the monospace on the question, never really used SE chat before ^^'
rak1507
rak1507
It's a scalar
Marshall
Marshall
@Martmists The result is a scalar. Maybe the APL Wiki page on the concept makes it clearer?
See also bound.
Martmists
Martmists
but why would it have an empty shape rather than simply a shape of 1
Marshall
Marshall
@Martmists A shape of 1 (,1 really) indicates that it has one axis of length one, but the result of Enclose doesn't really have any axes.
One strange result you would get if you made it a vector instead of a scalar is that the outer product (⊂'abc') ∘., 'ef' 'ghi' 'jk' would have shape 1 3 (the shapes of the two results together) instead of just 3. So you have this extra axis that appears and can get mixed into other arrays where you don't want it.
ngn
ngn
@Marshall i've never heard "bound" before. does it come from j?
Marshall
Marshall
16:05
@ngn Dyalog implementers used it regularly. I think it's an old APL term but I don't know when it originated.
ngn
ngn
@Marshall maybe it just happens to be the name of an internal function or field with no significance outside of dyalog's implementation?
Marshall
Marshall
@ngn I believe I've seen it in documentation for some old APL dialect, but I have no hope of tracking it down now. In any case, it's featured prominently in my outer product talk so there's some precedence now.
ngn
ngn
:)
Martmists
Martmists
a follow-up question about scalars: How does it decide when to apply certain functions and when to error? i.e. ⍳⊂10 works but ⍳⊂1 2 does not.
ngn
ngn
@Martmists ⊂10 is the same as 10 but ⊂1 2 is not the same as 1 2
Razetime
Razetime
16:14
@Martmists ⊂10 doesn't enclose 10
Marshall
Marshall
See floating arrays.
Razetime
Razetime
@Martmists try ⍳⊂,10
Martmists
Martmists
Alright, I think I get it
time to overhaul my code again lol
ngn
ngn
beware of ⍳,10 which will never be fixed
rak1507
rak1507
⋄ (⍳⍴∘1)¨⍳5
bot is taking the day off?
ngn
ngn
16:25
@rak1507 could be jewish :)
rak1507
rak1507
lol
Razetime
Razetime
@ngn oh no what's wrong with that
@rak1507 use userscript
rak1507
rak1507
effort
Razetime
Razetime 
      (⍳⍴∘1)¨⍳5
┌─┬─────┬───────┬─────────┬───────────┐
│1│┌───┐│┌─────┐│┌───────┐│┌─────────┐│
│ ││1 1│││1 1 1│││1 1 1 1│││1 1 1 1 1││
│ │└───┘│└─────┘│└───────┘│└─────────┘│
└─┴─────┴───────┴─────────┴───────────┘
rak1507
rak1507
It's more profound with ⎕IO←0 
      (⍳⍴∘1)¨⍳5
┌──┬─┬─────┬───────┬─────────┐
│┌┐│0│┌───┐│┌─────┐│┌───────┐│
││││ ││0 0│││0 0 0│││0 0 0 0││
│└┘│ │└───┘│└─────┘│└───────┘│
└──┴─┴─────┴───────┴─────────┘
Razetime
Razetime
16:27
yay userscript
ngn
ngn
@Razetime it should return the ,¨ of what ⍳10 returns
Razetime
Razetime
@ngn oh that does make sense
 
1 hour later…
user41805
user41805
17:37
@Razetime ⍬⊂⍛, do you really need it to be a ⍬? is there anything else that could take its place?
maybe it's worth porting the newer ngn/k solution? idk, feels weird seeing a recursive solution win over in apl
 
4 hours later…
rak1507
rak1507
21:40
I want to make a function that simply maps a given value a to a given value b, but it needs to be invertible
so when I give it b, it needs to give back a
what is the easiest way to do this?
it doesn't matter what it gives for other outputs that are neither a nor b
dzaima
dzaima
@rak1507 what are a and b - some values you've chosen? arguments?
rak1507
rak1507
some values that I've chosen
dzaima
dzaima
are they arrays? numbers?
rak1507
rak1507
Ideally it should work for anything, but in particular character vectors
dzaima
dzaima
this works
@dzaima -2
rak1507
rak1507
21:45
That's smart, thanks
 
1 hour later…
rak1507
rak1507
23:11
came up with imo a pretty decent answer to problem 57 (@KamilaSzewczyk) dzaima.github.io/paste#009avftS71e5R71KNRz1dhgaPeuc/…
ngn
ngn
23:51
@rak1507 challenge: solve #66 without web searching :)
rak1507
rak1507
Challenge accepted :)

« first day (1473 days earlier)      last day (1478 days later) »