« first day (25 days earlier)      last day (1302 days later) » 

10:23 AM
@Deadcode I started chopping at your abundant regex, -22 so far
 
 
5 hours later…
2:57 PM
There are two spots where you do a multiplication when an addition would suffice. (Z - 1) / (\5 - 1) * \5 is better expressed as (Z - 1) + (Z - 1) / (\5 - 1). Same for \30. Those cost like 20 bytes each.
 
 
2 hours later…
4:38 PM
Down to 406. Here's the full history as a single pastebin: pastebin.com/raw/5jGjARc2
Latest version works in RegexMathEngine but not TIO, so one of those two is buggy :|
Bug report: (?=(x))?\1 matches 0 in ECMA but 1 in RME
Thankfully (?=pattern|) works in both at no cost in bytes
 
 
3 hours later…
7:49 PM
@Grimy Wow, that is absolutely magnificent! Would you like to post it as an answer? I'd like to give you some bounty for this.
@Grimy I'm aware of the bug. But that's the same as ()\1 isn't it? A lookahead will always trigger a no-match due to NEO+, so the (x) inside that lookahead will never be evaluated in any circumstance in ECMA.
If you've found a circumstance in which (?=anything)? will evaluate anything in proper ECMA then I definitely need to know.
 
Thanks!
I think you’re correct about that bug, I didn’t realize it was already known
I’m okay with posting mine as an answer, but it needs some more work first, if only for formatting/commenting
 
Okay awesome! :D
 
I currently have a 387 that only gets 64 wrong from 1-1000
I can’t prove it doesn’t have other false positives on prime powers, though
 
I've verified your 409 byte version up to 8000 so far
 
409 or 406?
 
8:00 PM
Interesting that something gets only 64 wrong in 1-1000...
409
 
The pastebin I’ve posted has the 409 again instead of the 406
But the TIO link has the correct 406
 
Ohh
I took it from the pastebin
Where's the TIO link?
 
Comment on your answer
 
@Grimy :-O did you realize you commented "-98 bytes" exactly on this as well?
 
Yes x)
Though -98 out of 183 is a better ratio
Alright, here’s a 388 that passes 1-1000: https://pastebin.com/raw/hitQV1cZ
I *think* it’s robust but I haven’t fully figured out the math.
To get the 387 I mentioned earlier, just delete the first `x`.
 
8:11 PM
:-O
 
8:22 PM
Do you think amicable numbers might be possible? Domain ^x*,x*$
Or multiperfect numbers
 
Ooh amicable is an interesting challenge
Multi-perfect I think is possible with some virtualized math
 
@Grimy The 388 checks out up to 10000
 
Noice
 

« first day (25 days earlier)      last day (1302 days later) »