All this ascii art and elves makes me want to play nethack...(coffeebug on nethack.alt.org)
You consume deep fried Elvenking. Eating deep fried food made your
fingers very slippery. Your weapon slips from your hands. You feel wide awake!
Language: R. Been a while.
Started a day late. Let's see if I can keep it up.
Struggling a bit remembering how to be an R-soul, very different to being Pythonic.
The data manipulation is stupidly easy once you construct the dataframe (data.table is an amazing package, far better
than Pandas). But where in Python you'd just write for-loops and they would be easy to read, the R way is to do
everything vectorised. The nested *apply are quite difficult to read.
Hard so far: day 12, day 16. The common theme is recursive functions/efficiency in/managing R's "node stack" limit (whatever that is). And instead of trying to be efficient in the recursion, write the dumbest version first (e.g. move one character at a time rather than fast-forwarding).
| Day | Comment |
|---|---|
| 25 | Was gonna do it properly but visualising my data just so happened to point out the solution, so ... ¯\(ツ)/¯ |
| 24 | Very angry at this day, don't think it was a coding problem so much as a "use a library to do it for you" |
| 23 | Frustrating - turns out I had the correct approach very early on but ran into the slowness of R - tried Rcpp hours later and it ran in seconds. |
| 22 | A nice breather (thank goodness) |
| 21 | I'm a dirty dirty statistician and I love it. |
| 20 | Had a lot of difficulty reading this one, part 1 took me AGES. Part 2 was done by inspection and by hand. |
| 19 | Picked the wrong direction to scale in for part 1 and had to abandon it entirely for part 2. Think like a programmer not a statistician :( |
| 18 | Should have got that way sooner. Gauss' shoelace formula is AMAZING |
| 17 | Wasted time going down the recursive function train of thought (because of last few days) before realising. |
| 16 | Frustrating - had to re-write the nature of the recursion because it got too many frames deep for R? (but only 10,282 function calls?) |
| 15 | The difficulty was in reading the instructions. |
| 14 | Cheesed it with visual insepction and a calculator. "Identified patterns" ;) because that implementation ain't fast |
| 13 | Took way too long due to staying up til midnight trying to do day12 part2 lol |
| 12 | AAAAAAAAAAAAAAAAHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH |
| 11 | I thought I was being smart but @Steven's observation was galaxy-brain level |
| 10 | Quite fun. First time both answers were correct on the first try. Writing the intermediate outputs to file was surprisingly helpful. |
| 09 | I see the way they want us to do it and laugh fits order20 polynomials instead what's the point of using R if you can't overkill with it? |
| 08 | Next time I'll try my "naive" part1 solution on part2 while programming up the more general part2 case, in case it turns out to be right :( |
| 07 | That was fun, though I had a brainfart moment and forgot that I should just multisort in "wide" format rather than long |
| 06 | FUCK YEAH MATHS (after the morale-whooping received yesterday) |
| 05 | Just cannot work out the smart way to do it, but I'm so close to conceptualising it. Very frustrating. R is not suited to brute-forcing. |
| 04 | I feel like there is a sweet mathsy way to do part 2 but oh well, bring out the for-loop. Betraying everything R. |
| 03 | "Overlaps" logic does my head in for some reason. Took much longer than it should. Part 2 was pleasingly trivial from the part 1 data.table. |
| 02 | I LOVE data.table. But also, R lets you do "5" < 15 without raising exception?!?! (and returns a bool) |
| 01 | imagine that - the function that said it "extracts the last match" doesn't. ✟ ℝ.𝕀.ℙ. ✟ |
Todo:
- day 17
- optimise the creation of the adjacency-matrix (pre-allocate the sparse matrix OR don't make a matrix at all)
- write Dijkstra myself