The APL Orchard

« first day (1523 days earlier) ← previous day next day → last day (1428 days later) »

Razetime

02:50

@rak1507 in Dyalog local it should prolly say "INPUT LIMIT"

@rak1507 github.com/Dyalog/ride/issues/634 logged it here

rak1507

hmm, but it should be fine reading from stdin

that's frustrating

maybe it'll work with scripting...

Razetime

03:13

files generally work better for enormous input

rak1507

sure, but it shouldn't just give up and fall over if you give it large stdin

8 hours later…

Morten Kromberg

11:19

@Razetime Old users are a pain because they tie you down, new users because they go where no user has ever gone before :-). Thanks for the bug report!

3 hours later…

rak1507

13:58

CMC: given a vector of floats representing probabilities return booleans

1 hour later…

Elias Mårtenson

15:25

I'm stuggling to come up with a concise way of implementing ⊤.

I mean, I can implement it with a loop and stuff but that's pretty ugly. What's a nice APL way of doing it?

dzaima

@EliasMårtenson by the nature of it, it must be iterative

Elias Mårtenson

@dzaima so you're saying that my failure to come up with a nice one-liner is because it's not possible, not become I'm incompetent?

dzaima

@EliasMårtenson well, there definitely are some one-liners for it, but they must involve some function being applied for each item of ⍺ separately, depending on the previous result

Elias Mårtenson

@dzaima Yeah, something with ⍣ perhaps.

dzaima

it really asks for both a reduce and an each in the same place

{c←⍵ ⋄ ⌽{r←⍵|c ⋄ c⊢←⌊c÷⍵ ⋄ r}¨⌽⍺ is one impl; doesn't handle 0s in ⍺

{a←⍬ ⋄ ⌽a⊣{a,←⍺|⍵ ⋄ ⌊⍵÷⍺}/⍺,⍵} is another. you can really tell that this one wants to have an ¨ and the previous one wants a /

(this is a problem with functional things in many languages - you either have a map, a reduce, a scan, but never multiple of them)

Morten Kromberg

15:46

What about:
{⍉⌽a|⍤1⌊⍵∘.÷d←×\¯1↓1,a←⌽⍺} ⍝ ⍺ integers only

Should be OK on any rank ⍵, I think

dzaima

@MortenKromberg for some reason i thought that must fail on some ⍺s, but i guess it doesn't. has 2 divisions per item of ⍺ though

Morten Kromberg

@dzaima Can you explain? I guess it always begins with an unnecessary ÷1

Wish I could remember why I defined d :)

dzaima

@MortenKromberg In a statically compiler language, the ⍺|⍵ and ⌊⍵÷⍺ in the iterative solution would become one instruction. No chance of that happening in yours

it's a pretty good APL solution though

Morten Kromberg

Ah, statically compiled languages is not my department :D

Of course, in a modern APL system both ⍺ and ⍵ can contain fractional values and it must all depend on ⎕CT as well.

2 hours later…

Elias Mårtenson

18:04

Thanks. I'll take a look at these tomorrow.

« first day (1523 days earlier) ← previous day next day → last day (1428 days later) »

all rooms