The APL Orchard

@cairdcoinheringaahing should Adams request get nuked? if i click on that chat.stackexchange.com link i can see the message that he requested to be nuked.

caird coinheringaahin g

Huh, it should say "(removed)", if it even links you to a message

Aiden Chow

@gwa000 well the link didnt one box so its fine

gwa000

looks like it went away.

caird coinheringaahin g

If you click on the link now, can you see the message?

Aiden Chow

@cairdcoinheringaahing no

gwa000

5:51 AM

maybe it was still in some sort of cache?

Aiden Chow

@gwa000 probably

when i click the link it doesnt link to any msg

huh i can still see the history of the msg, is that available to everyone or just me?

Adám

@AidenChow I believe only you and ROs. (I can see it.)

But all ROs are disqualified from competing, so it doesn't matter.

Aiden Chow

@Adám oh nice, thats an interesting rule

ro's of this room right? not any room?

idek who the ro's of this room are here other than u lol

Adám

Only this room, yes.

@AidenChow chat.stackexchange.com/rooms/info/52405/…

Aiden Chow

ok ig other ro's cant see it

cool

@Adám ah cool, ive never seen the other three before

@Adám oh wait i wasnt paying attention... i mean a matrix of that list repeated

or a string

strings are just char arrays right

Adám

6:01 AM

yes

@AidenChow did you aplcart it?

Aiden Chow

@Adám it doesnt seem to pop up anything useful, i might be wrong tho

maybe im using the wrong search terms...

Adám

@AidenChow What did you search for?

Aiden Chow

@Adám "repeat list"

gives 3 results

wait i think i just find it by scrolling through the entire list

lemme see

oh yep it looks like its just ⍴

not even too sure what that is lol

Adám

Hm, I don't see ↑⍴∘⊂ or ⌿∘⍉∘⍪ there.

Aiden Chow

@Adám wat do those even do...

like how do they even work

Adám

6:12 AM

n⍴∘⊂v creates a list of n vectors v, which ↑ mixes into a matrix.

⍪v makes a column matrix out of the vector, ⍉ transposes it to a 1-row matrix, and n⌿ repeats the rows n times.

gtg, back in around 45 mins

Aiden Chow

@Adám oh bye!

might not be here 45 mins later but cya

caird coinheringaahin g

@Adám And mods, but I guess that's something we'd disclose if competing

Aiden Chow

@Adám how do i know if smth is a matrix or a list of vectors, like why do we need to mix into a matrix

is there a huge difference between them or r they interchangeable?

also if im understanding rotate correctly if i give a matrix and a vector of nums to rotate, then it will rotate each column of the matrix by the corresponding value in the vector? or is it each row?

hyper-neutrino

6:28 AM

@Adám using the power of magic you should no longer be able to see it either :)

(I can't either)

Aiden Chow

yo damn i cant either

wow

11'ed

Adám

6:53 AM

@AidenChow You can see how it prints, ask for its shape, and/or depth.

@AidenChow They are rarely completly interchangable. If you need to treat each row as a unit, use a vector of vectors. If the elements all need to be addressed equally, or if you need performance, use a matrix.

@AidenChow v⌽m rotates the rows while v⊖m rotates the columns.

Aiden Chow

7:18 AM

@Adám what would be the shape of a matrix vs list of vectors

anyways i gave up on using rotate and solved 1:3 without it lol

Adám

@AidenChow A matrix's shape has two elements; a vector's shape has one element.

Aiden Chow

@Adám oh yep thats obvious in hindsight lol

wait is both phase being judged by actual ppl or just phase 2

the problem is that i have no idea whether what im doing is "apl-like" (so more likely to score better) or just pure garbage

lyxal

@AidenChow without rotate? That sounds painful

Adám

7:33 AM

@AidenChow Phase 1 has extensive set of test cases that are immediately checked against, and the best participants are then judged by actual people. Phase 2 only does a syntax sanity check upon submission, and the test case set is run at judgement time, after which the judges look over the code and score clarity and use of APL.

@AidenChow @lyxal Let's be careful, alright?

Aiden Chow

@lyxal i have given up trying to understand how it works lol

lyxal

@Adám I am, I'm just saying if it's a recommended hint, and it isn't included, the answer must use some weird workaround to replace it :p

Aiden Chow

yea well idk how rotate works apparently cuz i couldnt get it to work

Adám

@AidenChow Does this help?

Aiden Chow

@Adám how are the "best participants" determined in phase 1 before going to judges?

Adám

7:38 AM

Correctness. IIRC, 1 point for passing basic cases and −0.2 for failing edge cases.

lyxal

Is there a way to see how well a submission scores without submitting it?

Aiden Chow

@Adám if u get a gold trophy does that mean u pass every test case even edge ones, or is that not guaranteed?

Adám

@lyxal No, but what's the problem with submitting?

@AidenChow yes, that's what gold means.

Aiden Chow

@Adám ok just making sure

Adám

it is all described under the sample problem.

lyxal

7:42 AM

@Adám I was just curious how well an answer of mine would perform.

Aiden Chow

on the off chance i make it to judging (which is probably not gonna happen tbh), will not using the hints affect my score in any way

Adám

@lyxal Oh, wait, no, you can test without submitting. Sorry.

@AidenChow No, not at all.

lyxal

@Adám but does it show the score?

Adám

Yes.

lyxal

Oh cool

Where do I see the score without submitting?

Aiden Chow

7:45 AM

i dont remember seeing a score when test, but i may be remembering wrong

actually i dont even remember seeing a score at all, even after submitting

i swear im probably writing some of the most horrendous apl code ever lmfao

lyxal

Can't be any worse than some of the stuff I've written

Aiden Chow

@lyxal how many u solve so far

lyxal

I mean apl I've written in the past

Adám

@lyxal You get a message below the input box.

lyxal

It just says "passed all basic tests"

@AidenChow I'm not competing :p

Adám

7:51 AM

@lyxal yes, that's the full score.

Aiden Chow

yea for me its saying "Passed all basic and edge cases"

lyxal

Oh

Okay good to know

Adám

@AidenChow Oh, trust me, I've seen stuff…

Aiden Chow

just curious, r u one of the judges adam?

Adám

yes.

Aiden Chow

7:52 AM

@lyxal oh so u didnt make an acct?

lyxal

No

I don't have one

Aiden Chow

which ones u solve so far??

@Adám lol true, uve probably seen all the nasty stuff that beginners write lmfao

lyxal

@AidenChow just 1:3

Only for fun

Adám

Every year, some people ask that we blanket publish all solutions after the competition. One of the reasons we don't do it, is to avoid the risk of embarrasing people.

Aiden Chow

@Adám maybe release the solutions but make it anon?

Adám

7:56 AM

We've of course considered that, but it might still come back at someone, and especially phase 2 can be easy to identify based on style. Safer to let people self-publish (and add an entry on the APL Wiki for discoverability) if they feel like it.

Aiden Chow

@lyxal lol nice, u probably have some elegant answer in less than 10 bytes lmfao, meanwhile i have some monstrosity of a 30 bytes answer...

lyxal

I, uh, wouldn't call it elegant

Aiden Chow

@Adám how many submissions yall get on average each year anyways?

@lyxal if u used rotate anywhere in ur code its most likely better than mines...

i wont say what, but i did some wacky stuff for 1:3 lol

lyxal

Mine does sound comparatively sane then :p

Aiden Chow

exactly lol

mines is insanity

like if a judge saw it they would actually just mark it down so much...

Adám

8:01 AM

@AidenChow Hm, I'm actually not sure, as I only see the top scoring entries that need judging.

Aiden Chow

ah i see... how many u judge usually (totally not tryna see my chances of making it to judging lmfao)

Adám

About 45 or so. At least it was around that in the random year's set that I just dug up.

Aiden Chow

damn only 45, there goes my chances

are all the other submissions just hidden from u or can u access anyones submission?

Adám

I think I can access everything, at least with some effort, but not "anyones" in particular, as I don't know the identity of the submitter until chosen as a winner.

We want to avoid bias on people we know, so competitors are only identified to the judges using ID numbers.

Aiden Chow

ah yep that makes sense

lyxal

8:16 AM

@Adám was there any beta testing for this year's problems? I didn't see any mention of it in TNB like there usually is this year.

Aiden Chow

bruh my 1:5 is 51 bytes

this is so sad

at least i got gold trophy but damn this is horrendous

is there a builtin to get the last element of a list

O_O what is this in 1:6... string parsing... how to do that lol

f*** how am i going to fix this

┌─────────────┬────────────┬───────────┐
│┌────┬──────┐│┌────┬─────┐│┌────┬────┐│
││name│Morten│││name│Brian│││name│Adám││
│└────┴──────┘│└────┴─────┘│└────┴────┘│
└─────────────┴────────────┴───────────┘

thats what my code return....

how am i going to turn that into whats expected

:(((((

Adám

Obviously, we can't really help you there.

Aiden Chow

wait.... i just fixed it

cool

didnt expect that to work but nice

1:6 done yay!

Adám

@AidenChow No, but I'm suggesting that monadic ⊇ be that. Meanwhile, "last element of list" on APLcart gives the alternative.

Aiden Chow

8:32 AM

@Adám ah ok, is there anything for tail (everything except first element)?

Adám

@AidenChow I have a perfectly sensible solution that doesn't use rotate :-)

@AidenChow 1∘↓

Aiden Chow

@Adám yea obviously, u r adam. im just some random beginner who doesnt know anything about apl lol

@Adám oh yea duh

u probably have some crazy 5 byte answer for 1:3 too

lyxal

@AidenChow no need to be so harsh on yourself - I've been doing apl on and off for 4 years now and I'm still really horrible at it and get headaches after a while :p

Aiden Chow

@lyxal lol ive been doing it on and off for about 3 months, mostly off tho :P

Adám

I've been doing it on and off for about 3 decades…

Aiden Chow

8:36 AM

or maybe it was a little more than 3 months idk lol

@Adám lmaooo

bruh what is this 1:7

Adám

@AidenChow No, not at all.

Aiden Chow

@Adám ok well its probably like 5 times better than my 30 byte answer lol

mines is actually not pleasant to look at

Adám

By "byte", I assume you mean "character". Mine isn't much shorter, but let's compare once the competition closes.

Aiden Chow

@Adám which is months from now lol

how do i convert a number to a string

Adám

What did you APLcart?

Aiden Chow

8:47 AM

decimal to string

ok wait num to string seems to give a promising result

Adám

@AidenChow That does give a usable result, no?

Aiden Chow

errr im not sure

xpqz

Is there a way in the session to test if CEF is enabled?

Adám

@xpqz {0::0 ⋄ ≢⎕NEW⊂'HTMLRenderer'} ?

We have a "Verify .NET Interface" I-beam. Maybe we should have one for CEF too?

Aiden Chow

are the only submissions that have a chance of being judged are the ones that solve all the problems

xpqz

8:58 AM

I've not needed it until today, but it struck me that it would be nice to be able to see/query the session's view of its config state -- if you write an application, it might be good to be able to tell if a feature you rely on has been disabled by the user. I suspect it's already possible; I've not checked.

@Adám Does the job nicely; thanks!

Is it possible to pin the Dyalog version the Jupiter kernel should use? It seems to always pick the one with the largest version number if there are several.

Adám

@xpqz github.com/Dyalog/dyalog-jupyter-kernel/issues/74

@xpqz There aren't many disableable features, are there?

xpqz

@Adám I've not checked; I'll take your word for it. But there are quite a few aspects that are configurable, at least.

Adám

Configuration can be read with ⎕SE.Dyalog.Utils.Config

xpqz

Nice! Perfect.

Wow, I can hack the language bar? .../Dyalog/languagebar.json

Master Quiz

9:14 AM

Hi! I was interesting to learn about multithreading in APL. I've tried to run this:
{⍎'(⍎&''solver ''''',⍵,''''''')'}¨'02-swarming-ant' '02-swarming-ant'
solver '02-swarming-ant', is executed in about 30s and the line above is executed in 1min, so apparently it starts two threading but they aren't executed at the same time.

Adám

& launches so called green threads, i.e. threads that APL manage and divides its processor time between them, but there's still only one OS thread. You need isolates if you want multiple OS threads.

Master Quiz

@Adám So it's possible to execute them at the same time and obtain a result in 30s?

Adám

Yes. see Parallel Language Features for details.

Also, there's no need to (double) wrap in executes.

⎕CY'isolate' ⋄ solver IÏ '02-swarming-ant' '02-swarming-ant' should do the trick.

fyi, IÏ is a model for a proposed ∥¨

Master Quiz

This gives an error
FUTURE ERROR: 2: SYNTAX ERROR: {⍺←⊢ ⋄ ⍺ solver ⍵}: iEvaluate[13] (,⍕(⍕rc),': ',(0⊃res),{(⍵∨.≠' ')/': ',⍵}1⊃res,'' '')iSpace.qsignal rc

But I'll read about it, thanks

Adám

Does solver¨ '02-swarming-ant' '02-swarming-ant' work?

Master Quiz

9:29 AM

@Adám Yes

Adám

Does it try to access globals?

Master Quiz

@Adám Yes

Adám

Well, read the doc :-)

Master Quiz

It's a tradfn that creates some local variables that are accessed by an inner function

@Adám Yes, I'll do it later

Adám

11:23 AM

Reaction to “Change the way you write. Change the way you think.” — part 1 (The APL Show ep. 6)

►

Adám

12:22 PM

Unconditionally Shifty (APL Quest 2018-7)

►

Adám

1:00 PM

Welcome to APL Quest 2018-7! Today's quest is Unconditionally Shifty:

> Write an APL function that given a right argument Y of any array and a numeric scalar or vector left argument X returns a Boolean indicating if the left argument is a valid argument for X⍉Y, like the result of {0::0 ⋄ 1⊣⍺⍉⍵} but does not use ⍉ (to test the arguments).

Note: From next week, March 17, these sessions will begin two hours later, at 15:00 UTC — also beware of daylight savings time changes.

Richard

{(⌽⍣(⍺<0))(-|⍺)↓(⍺/0),(⌽⍣(⍺<0))⍵}

Adám

@Richard Wrong paste?

Richard

{⍺≡⍬:1 ⋄ (r=⍨≢⍺)∧(∧/⍺∊⍨⍳⌈/⍺)∧∧/⍺∊⍳r←⍴⍴⍵}

Adám

:63151738 Oops, this is ofc 2018-8: Making a Good Argument

Richard

Not very exciting this one. Just did the tests as asked

Maybe something smarter is possible

Adám

1:03 PM

Why do you need the special case for the empty vector?

Richard

the ⍬ gives an error on ⌈

Adám

Ah, not so much ⌈/ but ⍳ on negative infinity…

Richard

yes

couldn't think of another way to avoid it

ovs

⍳0⌈⌈⌈/⍺ should do it

Richard

ah yes :)

Adám

1:08 PM

I have an elegant one in the works.

Richard

@ovs ⍳0⌈⌈/⍺

Adám

∧/=⍥≢∘⍴,{⍺(∊⍨,∊)⍳≢∪⍺} but I have a feeling that the inner fork can be simplified.

⍳≢ can be ⍋

Richard

Still looking at the first part with ⍥. The rest is clear to me.

Adám

=⍥≢∘⍴ just says that the left argument length must match the right argument rank.

Richard

but ≢∘⍴ on ⍺ also gives rank, does it? And not the length of ⍺

Adám

1:16 PM

Operators have long left scope: (=⍥≢)∘⍴

=⍥≢∘⍴∧{(⊢≡∪∘⍺)⍋∪⍺}

Richard

ok, still confusing for me. But that explains it

ovs

=⍥≢∘⍴∧{∧/∊∘⍋⍨∪⍺} seems to work. Or =⍥≢∘⍴∧{r≡⍋⍋r←∪⍺}

Adám

Ah, nice.

ovs

Ah the second one makes a lot of sense, it's asking if ∪⍺ is a permutation

Richard

@ovs why ⍋⍋ and not just ⍋?

Adám

1:23 PM

⍋ inverts a permutation, so ⍋⍋ will be identity — but only for permutations.

=⍥≢∘⍴∧{≡∘⍋∘⍋⍨∪⍺} if you want to avoid the variable, and =⍥≢∘⍴∧≡∘⍋∘⍋⍨⍤∪⍤⊣ if ou want to go all-tacit.

Richard

I do not understand
⍋⍋ 2 3 2 gives the same as ⍋ 2 3 2

Adám

Bad test case.

Slightly shorter but repeats (trivial amounts of) work: =⍥≢∘⍴∧⊣≡∘⍋∘⍋⍥∪⊣

Richard

Could you give another test case in which this is indeed different to make it clear for me

Adám

3 1 4 1 5

Richard

ok!

Adám

1:29 PM

20.0 will be able to match the length without duplicate work: =⍥≢∘⍴∧⍋⍤⍋⍛≡⍤∪⍤⊣

But =⍥≢∘⍴∧{r≡⍋⍋r←∪⍺} is probably easier to read.

So we can read this as "length of left argument and shape of right argument must match, and the left argument must be a permutation vector but where duplicates are allowed".

So there are really only two rules, and the first one is actually artificial. Dyadic ⍉ could limit itself to transpose the ≢⍺ leading axes.

Silas

So you could provide a longer ⍺ and the ⍉ would still work?

Adám

No, a shorter.

E.g. 2 1⍉rank3array would be equivalent to current 2 1 3⍉rank3array

1 1⍉rank4array would be equivalent to current 1 1 2 3⍉rank4array

Silas

ok, so a shorter ⍺ would leave non-specified axes and only modify the ones given?

Adám

Yes, and the non-specified ones would be the trailing ones.

If you want to transpose only trailing axes, you just use Rank.

Have we exhausted the topic?

Richard

yes, for me we did

Adám

1:42 PM

OK, see you next week for 2018-9: Earlier, Later, or the Same?

Stack Exchange Feeds

4:43 PM

0

Q: Two APL functions give different results, but their boxing is identical

I am trying to create an APL function which returns differences (non-identical items with identical index) between two vectors. I have come up with the following and it works fine, subtracting two vectors and discarding zeroes from the result: func ← {((⍺-⍵)≠0)/(⍺-⍵)} However, I was not satisfie...