Bitshift Variations in C Minor by Robert Miles is a Code Golf music piece wtitten in C (with some additional bash commands). It was originally presented in a 2016 Computerfile video Code Golf & the Bitshift Variations. You can listen to clean renderings of it here (by Lucia Ceionia).
The original 2016 source code presented in the description under the YouTube video was 245 characters long:echo "g(i,x,t,o){return((3&x&(i*((3&i>>16?\"BY}6YB6%\":\"Qj}6jQ6%\")[t%8]+51)>>o))<<4);};main(i,n,s){for(i=0;;i++)putchar(g(i,1,n=i>>14,12)+g(i,s=i>>17,n^i>>13,10)+g(i,s/3,n+((i>>11)%3),10)+g(i,s/5,8+n-((i>>10)%3),9));}"|gcc -xc -&&./a.out|aplay
A tweet from 2017 (or some other Rob Miles tweet, I'm not sure) stands 216 characters:gcc -xc -oa -<<<'i;n;g(x,t,o){return(3&x&(i*((3&i>>16?"BY}6YB6%":"Qj}6jQ6%")[t%8]+51)>>o))<<4;}main(s){for(;;)putchar(g(1,n=++i>>14,12)+g(s=i>>17,n^i>>13,10)+g(s/3,n+(i>>11)%3,10)+g(s/5,8+n-(i>>10)%3,9));}';./a|aplay
With a month or two of work, I managed to shrink it down to 185 characters:gcc -xc -oa -<<<'i;n;g(m,t,o){return("GTj?TG?5"[7&t]+!(n&12|t&2)*9)*i>>o&m&24;}main(s){for(;;)putchar(g(8,n=++i>>14,8)+g(n,n^(s=i>>10)/8,6)+g(n/3,n+s/2%3,6)+g(n/5,n-s%3,5));}';./a|aplay
In my previous market, a 184 characters version was found, by replacing gcc with cc:cc -xc -oa -<<<'i;n;g(m,t,o){return("GTj?TG?5"[7&t]+!(n&12|t&2)*9)*i>>o&m&24;}main(s){for(;;)putchar(g(8,n=++i>>14,8)+g(n,n^(s=i>>10)/8,6)+g(n/3,n+s/2%3,6)+g(n/5,n-s%3,5));}';./a|aplay
Will someone be able to shrink it down below 184 characters, or the limit has been reached? This market resolves to YES as soon as someone posts a shorter version in bash+C, and I verify that it works. It resolves to NO on Jan 1st 2024 if no such version is found.
Conditions:
- The bash command should compile the source code and run + play it using aplay.
- You ARE allowed to slightly change the played notes, but their relative frequencies should not differ more than 1.4% from the original, and the entire piece should not be shifted more than one semitone (6%) from the original.
- You are allowed to use all sorts of tricks and hack in C language, as long as it compiles and works as intended on an average linux machine.
- Reading from pre-written files or from the internet is not allowed.
- I reserve the right to disqualify methods that I judge to be against the spirit of code golf, but I will hold a discussion about it in the comments section beforehand.
Useful links:
Deep analysis of Bitshift Variations in C Minor code
BitShift-Variations-unrolled
Bitshift Variations Humanized + Rust version
1,000🏅 Top traders
| # | Name | Total profit |
|---|---|---|
| 1 | Ṁ232 | |
| 2 | Ṁ110 | |
| 3 | Ṁ63 | |
| 4 | Ṁ16 | |
| 5 | Ṁ0 |
@MayMeta I had tried for loop stuffing for a while before finding the gcc->cc bit, but originally didn't realize there was such a good candidate for the body of the loop until later. i wouldn't be surprised if there's one or two more little transformations like this remaining before reaching a local optima. after which I assume expanding it back out and using more arcane approaches would be necessary. perhaps a job for AlphaDev if a successor market's stakes are high enough :) https://www.deepmind.com/blog/alphadev-discovers-faster-sorting-algorithms
@firstuserhere alphadev outputs assembly, which could theoretically be a smaller c program by naively executing a string representing assembly, but i'd posit the question of which representation is smaller is undecideable via rice's theorem
@brubsby using alphadev assembly to inform a hand crafted c solution to earn monopoly money would certainly be a new frontier in man machine cooperation
@brubsby btw as a side-effect of your solution we got another repeated trigram: ")))", joining the "<<<" that threw people off in the png-decoding market. I'm so happy to see the code getting more and more cursed
@brubsby thanks for linking the article for alphadev, sounds really good. I'll check out their abseil library, cuz a 30% increase in hashing perf has got to be seen