The APL Orchard

Elias Mårtenson

2:58 AM

Making an APL implementation is a great way to understand APL itself actually.

Adám

5:04 AM

@EliasMårtenson Yeah, quite a few people here have gone that route.

Dyalog APL

12:06 PM

<phantomics> Worked for me

Dyalog APL

12:17 PM

<phantomics> On that note, has anyone considered implementing rational numbers in an APL? I get them for free in Common Lisp so I was thinking of using J's 1r2 syntax

Elias Mårtenson

GNU APL does it.

But it has to be activated using a compile flag and I think it's disabled by default.

I've considered implementing it in KAP.

Marshall

@DyalogAPL @phantomics NARS2000 has them too.

Elias Mårtenson

The thing is that rational numbers are usually only useful if you have bignum support, which GNU APL doesn't do so on that platform they are not overly useful.

rak1507

Damn just solved a challenge locally but doesn't work on TIO :(((

Dyalog APL

<phantomics> Cool, do they use the r-notation like 1r2?

<phantomics> So I can be consistent

Elias Mårtenson

12:25 PM

If I implement bignums in KAP, then I'll do rationals as well.

But yeah, I don't know of any language that does numbers better than Common Lisp.

Marshall

@phantomics NARS2000 uses r for rationals, like J.

Elias Mårtenson

Maxima might be a bit better but it piggy-backs on top of Common Lisp so it's not fair :-)

Dyalog APL

<phantomics> Yeah, I'm spoiled with rationals, bignums and complex for free

Elias Mårtenson

At least APL does Complex right.

rak1507

Does anyone have any idea why {10|⍵,⍨⊂⍺+⊃⍵} might error in dyalog extended

a domain error

Adám

12:33 PM

@rak1507 TIO link?

RGS

@EliasMårtenson Mathematica?

rak1507

Never mind, got it

Or not

Adám

⍨

rak1507

tio.run/##SyzI0U2pTMzJT9dNrShJzUtJTfn/P@1R24Tqgkc92x/… @Adám

ignore the horrible code (for now)

Oh now it doesn't work locally either

Ah it does but with ⍣¯1

Got a chemistry test now :(( will have to fix this later

Razetime

1:42 PM

@rak1507 whats this do?

rak1507

Stuff

48

Q: Combination bike lock

The scenario After a long day's work slogging in the office and browsing stackexchange.com, I finally walk out the door at 16:58, already weary with the day. Because I am still only an intern, my current mode of transportation is on bicycle. I head over to my trusty Peugeot Reynolds 501, but bef...

Razetime

woooaaaah

rak1507

f←{p⌷⍨⍸⍵≡¨⍥⊆⊃¨p←({⍵/⍨≠⊃¨⍵}⊢,⍥,((+,-)↓⍉2⊥⍣¯1⊢⎕A⍳'ABCDFGHLN')∘.{10|⍵,⍨⊂⍺+⊃⍵}⊢)⍣≡⊂⊂4⍴0}

works locally

Might just post it

Razetime

2:08 PM

@rak1507 tried using a polyfill here: Try it online!

@rak1507 works alright locally

rak1507

Yeah :/ I posted it

Currently in economics getting roasted so I'll do the explanation later

RGS

3:09 PM

@rak1507 I thought you were gonna do the exercises of my blog post during economics :P

DISCLAIMER: RGS in no way endorses behaviour that might or might not have a negative impact in academic perfomance and is not to be held responsible for the actions of those who follow his suggestions.

Dyalog APL

<phantomics> Blog post where?

RGS

phantomics: on my blog, about my basic APL interpreter in Python. Read @ mathspp.com/blog/lsbasi-apl-part3

Dyalog APL

<phantomics> Cool

RGS

phantomics, if you do have a look at any of the blog posts there (about APL or not) be sure to let me know your thoughts

Dyalog APL

<phantomics> Certainly, don't write much Python but interested to see your approach

RGS

3:21 PM

sure. I don't have much knowledge on interpreters and related topics, so don't expect great design choices or things like that :P

Dyalog APL

<phantomics> When I took up writing an APL I didn't realize how popular a pursuit it was, it's a great way to learn

<phantomics> And I went for a compiler

rak1507

@RGS lmao

RGS

3:53 PM

@phantomics nice! What did you write it in?

Dyalog APL

<phantomics> Common Lisp: github.com/phantomics/april

<phantomics> @RGS This is my video presentation of it: youtube.com/watch?v=AUEIgfj9koc

<phantomics> Which features a hardware device that it powers

rak1507

Oh wow you're that guy

RGS

Ah so you are "that" guy. I was aware of that video, just did not know you two are the same person

Razetime

Hey mr @DyalogAPL

hey mr ngn

how much of your soul did you sacrifice to solve ascii art challenges

Dyalog APL

<phantomics> That guy? Seems a dubious honor

RGS

4:07 PM

yeah, I am sorry :P I just mean to say I did not establish the connection, even though I was aware of the existence of that video and of April. Also, to be completely honest that doesn't tell me much about you... You could be a "celebrity" in the Common Lisp world that I would have no idea!

All in all I just connected a couple of dots; that's all

ngn

@Razetime about 5.82%

Dyalog APL

4:23 PM

<phantomics> Haha don't worry about it

dzaima

6:54 PM

hacky reply to self userscript (also makes replies to the IRC bridge use usernames because why not)

dzaima

8:24 PM

new body system:
https://dzaima.github.io/paste/#01VXNbtNAEL77KUblhFKQx02K8LFJIw4QRw2cUA/reNOa@ifYTiECpChCXOCQg5F64YI4tFKPufMufgIegdn1JrhJ2pAEKrD2x/PN7Lfzs14DwF3@Gpz2fftloAFA7aAXbLHswwgh@/ge3rDSjy@fL96Vt7SKVFdDv@t6fLFJln7K0sH/0C5XxNdjW6mNKH1DPRt8R@qGQYOuOgoEy0oSGqwoO9wViJAMnFgXVkpwqniQ26bDSZ0uC@OogP9@hOutWkAigof1HunCsEQdqZdzsZDLPBm5PMFwanU1Fwujujn4jV1PJdm4MJ4V8Pk2vgY/K0T@y//RFMWdG0u/eShzb7M1Epp8YoX8Dydquka@5dNFPp1fsboDfR6DCz4Leszz@sAdN@EOJMduvE3oiRsckcChE0Y@SxIhHrPA8Ti4CXSjsMsjWvUq7HkO2Jx0kQPm29uKOZ0p0pLq30o50rkDNXP0lpy1v@vgqh/AAh@161n@tTbeLNo/smo2g@KXnw2@7lWX/PjbocNNDXQddBTlaz5rPQI0ATXAXQn…

RGS

10:06 PM

@dzaima this looks interesting. How does it work?

Dyalog APL

10:31 PM

<phantomics> I'm not sure if anyone from Dyalog reads here but I'm curious about something

<phantomics> Dyalog's arctanh function seems to be different from others

<phantomics> In Dyalog:

<phantomics> ¯7○¯2

<phantomics> 0.5493061443J¯1.570796327

<phantomics> In GNU APL:

<phantomics> ¯7○¯2

<phantomics> ¯0.5493061443J1.570796327

<phantomics> Common Lisp and Wolfram Alpha both return the same result as GNU APL

<phantomics> Why is Dyalog's different?

<moon-child> @Adám

<moon-child> phantomics: on my copy of dyalog, both the real and imaginary parts are negative

<phantomics> The real part is supposed to be negative

<phantomics> Do the high minus signs not show up for you?

<phantomics> Dylog output should be:

<phantomics> _0.5493061443J_1.570796327

<phantomics> The underscores are minus signs like in J

<phantomics> In everything else I've tried, the imaginary part is positive

ngn

10:52 PM

@phantomics does it matter? tanh of both is close to -2

Dyalog APL

<phantomics> Any discrepancy matters when a function that's supposedly a known quantity produces different results in different implementations, what if I'm doing something where I expect a positive imaginary part for this outcome based on the behavior of other languages?

ngn

right, so it's like sqrt: sqrt(4) might as well be -2 but we pick 2 by convention.

code_report

11:12 PM

APL Question: Trying to write an APL solution for "Longest Valid Parentheses". I have gotten to this:
t ← '}{}{{}}}}{}'
+\(~,[0.5]⊢)'{'=t
1 1 2 2 2 3 4 5 6 6 7
0 1 1 2 3 3 3 3 3 4 4
And now I need to reset the "plus scans" any time row 1 is greater than row 2. Is there an easy way to do that?

So the result for the above example would be:
0 0 1 1 1 2 3 0 0 0 1
0 1 1 2 3 3 3 0 0 1 1

ngn

@code_report can you give a link to the challenge?

code_report

@ngn Solution 4 is what I'm implementing leetcode.com/problems/longest-valid-parentheses/solution

Marshall

@code_report Given a mask of where you want to reset, you can get index-of-last-reset with {⌈\⍵×⍳≢⍵}, then subtract the value at the last reset from the current value. Probably need to add a 0 before both sequences to use this.

This is assuming you can compute the reset mask all at once instead of having each reset depend on previous ones. If that's the case (I'm not sure from the way you wrote your question) then it's a lot harder.

So the idea is to get a contiguous sublist that's balanced?

code_report

11:28 PM

at the end of the day, I just need the length of the longest valid parens.

once you apply the reset/plus scans, you can just do the following (assuming that x is the result posted above):
(+⌿(⌈.×)=⌿)x
6

@Marshall for more clarity:
x
0 0 1 1 1 2 3 0 0 0 1
0 1 1 2 3 3 3 0 0 1 1
(+⌿(⌈.×)=⌿)x
6

I just need to know how to get from my first piece of code to x

Marshall

The way I know to do this is to sort or group by bracked depth, so that at a given depth open and closed brackets alternate. Then a bracket is paired unless it's the first closed parenthesis at its depth or the last closed one. And a balanced sublist is the longest sequence of paired brackets.

code_report

"each reset depends on the previous"

i think this is the case

Marshall

It's pretty similar to my markdown link-finding code here (in BQN).

Mine is a little more complicated because non-bracket characters are possible, so I work with indices essentially to filter those out. The only-bracket version shouldn't need indices.

code_report

oh wow, I just realized you are Marshall Lochbaum. I got extremely excited when I saw the "Combinators" section of github.com/mlochbaum/BQN/blob/…

Marshall

With the diagram?

code_report

11:41 PM

and when I saw the S/W/etc characters from the BQN character set - I got extremely excited because I thought those my correspond to the S-/W-/etc combinators from Combinatory Logic

ngn

@code_report does it matter how you solve it? the typical apl way would be to try all substrings

Marshall

@code_report I don't think I've seen the W combinator.

code_report

not sure if that is the case though. I asked in the British APL Webinar if APL-ers are familiar with the fact that trains (forks) and many of the operators correspond to combinators: angelfire.com/tx4/cus/combinator/birds.html

@ngn isn't the complexity of that intractable?

ngn

@code_report it's a finite problem. O(n^3) is tractable :)

there are O(n^2) substrings, each can be tested in O(n) time

code_report

hmmm, i guess thats true. the solution needs to be linear for my case though

ngn

11:58 PM

not linear, but close: ⌈/'x+'⎕s 1⊢'{(x*)}'⎕r'x\1x'⍣≡t

Transcript for