@rak1507 Not just those two. Monadic and dyadic operators are actually entirely different syntactic classes. (In J they have distinct names too.) ALL monadic operators can be assigned directly. NO dyadic operators (except dops) can be assigned directly. Not that this restriction makes sense, but at least ¨ and ⍨ are no different from say & or ⍣2 or {⍺⍺}.

@Jonah Maybe you can use F. and then call Z: in the condition?

@Adám yeah that's an option I forgot about though when I've explored F. / Z: previously the solutions weren't very golfy.

@Bubbler the counter solution might be the best option :(

CMC: Produce 0⍴⊂⍬

I have many at 3

2

Well, it can hardly be shorter than 2.

@xpqz Sorry for making the fun so short-lived.

Well, now that I have a brute-forcer....

My favourite 3 was 0⊂0

My favourite 3 was ⍬⊆⍬

Or maybe ⍴¨⍬

Another blog post?

Still working the first one!

A lot of people struggle with empty arrays and prototypes.

Count me into that set....

I think going through and explaining all these could be a great help to gain an understanding, and maybe even an intuition for the subject.

The 2 makes more intuitive sense than most of the 3s to me.

,⊂⍬

@Adám is there always a way to get low rank empty arrays without ⍴

@Razetime I think you can do anything without ⍴, though it might well be very awkward.

@Razetime Not sure that is correct?

⋄(,⊂⍬)≡0⊂0

@xpqz 0

woohoo, my "let's build a simple APL interpreter" series reached the front page of HN! news.ycombinator.com/item?id=27728071

I'd share a screenshot but I'm on my phone and I can't seem to be able to do it XD

@RGS Well done.

@RGS Interesting that it is part 2 that's on HN. It already has significantly more views than part 1.

@Adám yeah, but iirc someone once shared part 1 on HN as well... I might be wrong, but I seem to have that recollection.

@xpqz yep incorrect

@RGS may be worth linking to part 1 in part 2

@Razetime all parts are linked in the end, and usually the intro of part n+1 directly links to part n. Should I add the series listing to the top of the posts..? That's not something I see commonly

@RGS Same person, three months ago: search.

Hi all, has anyone tried to solve project euler 14 using APL? I tried building a cache where the length of a collatz sequence is stored in the vector vals and the initial starting point is stored in the hashed array (using 1500⌶) but it is still too slow (about 5 minutes on my laptop).

Are you searching by just counting up from 1 and finding the sequence length for each?

Yes. I have a cache though, so if along the way I hit a number I've cached then I just add the current count to get a total and return that + update the cache.

@hyper-neutrino

Ah. That'll take far too long because n/2 means every new value potentially doubles the next term to prepend to your current chain. If you have a number x at the start of your chain and you want to extend it backwards, you essentially want to find some k such that k/2=x or 3k+1=x. Think about how you can do this without just looping up.

(I havent' solved this by the way, but I'll try to do it in C++ with my idea, so do take my advice with a grain of salt)

wait, hold on, i may have misread this. give me a second to think about if my suggestion still works

hm, yeah, my idea may not work, sorry. i'll think more about it later maybe

you don't need a cache

collatz is simple enough to bruteforce via recursion

or via power

but tail recursion is far more optimized

ok, will try but I reckon that I won't get below a 1 minute runtime @Razetime. Do you reckon tail recursion is more optimised than ⍣? I doubt it.

yes, something like that

naive recursion gives 32797ms btw

old conversation about problem 14 (note: spoilers: speed)

interesting

Announcement: Dyalog webinar on Database basics in 10 minutes.

@11Kilobytes ⊃⊃{i a←⍵⋄(m/i)(a/⍨m←(≠a)∧1≠a←m+2÷⍨a×1+5×m←2|a)}⍣{1=≢⊃⍺}⊢2⍴⊂1+⍳1E6 ⍝ thanks dzaima

@11Kilobytes If you put your code into a separate message (it'll be visually merged with you previous one), you can press Ctrl+k to enable monospace font and indenting.

@Adám On APLcart issues, you have a lot to do with adding links to more kinds of content, particularly videos and long-form text. Is there an existing way to implement this, eg with the '?' button, or are you planning to add another button-link type in addition to the current '?' & '▷'?

@AviFS Are you talking about the issues labelled (pub)?

@Adám Good point. Yeah. But the description isn't helpful. It just says concerning /pub

@AviFS Indeed: aplcart.info/pub

Woahhhhh

Now I get it! Thanks.

Did you mean to have a serif font? I find it much harder to read.

@AviFS np. I don't publicise it because it is so far from finished. However, see the internship ad.

@AviFS No, does it render wrong?

Yeah, it's a different font

Which also screws up the logos

Which browser?

Firefox

Let me look up the specifics

@AviFS Then you can go into the dev tools → Inspector → Fonts and have a look what's going on.

I was about to!

In the meantime you might want to check, too, @Adám & others!

I'm on a Mac and it's happening to me across all of Safari/Firefox/Chrome, which is every browser I have

OK, I think I know what the issue is.

@AviFS Should be fixed now. Thanks!

@Adám Yup! Fixed!!

I actually would be very interested in the internship opportunity! I don't think I ever sent in all my info last time. Should I do that?

(I'd have to write up a CV for the first time, haha.)

@AviFS No need for a CV for this specific job.

Phew! What would be the next step in applying then?

11Kilobytes was also interested. Let's find a time when all three of us can meet and I can explain the work in more detail, and if you still are interested after that, I'll hand you over to the appropriate colleagues to discuss payment. How does that sound?

That sounds great! Maybe splitting the work would also be a possibility, then? Or are you thinking it's a one-person job?

I prefer having a couple of interns to do it.

Perfect. I can only commit to being free on any given day between 7pm & 5am UTC, though I know that's not very convenient in your timezone!

But I'm on a lot of the time so we can also play it by ear. Feel free to ping whenever/wherever, if you didn't feel free to do so already!

Between 10:00 and 21:00 UTC for me.

@AviFS @11Kilobytes OK, so right at 19/7pm UTC would suit all three of us. Would Monday Jul 12 work for you?

@Adám Works great for me! See you guys then

Works for me too @Adám!

Great. Added to calendar. Zoom ID 844 0824 9730 with passcode ×/7 73 661

Looking forward to it!

@xpqz Which video is it? I was also wondering about what that meant!

@AviFS This one.