Dec 2022 | Dec 2021 | Dec 2020 | Dec 2019 | Dec 2018 | Dec 2017 | Dec 2016 | Dec 2015 | Dec 2014 | Dec 2013 | Dec 2012 | Dec 2011 | Dec 2010
-
Subscribe
-
Reading
-
About
Sunday, 2023-12-31
Friday, 2023-12-01
Advent of Code 2023
Project website: Advent of Code 2023.
I use Perl for all the solutions.
Most assume the input data is in a file called input.txt in the same
directory as the file.
A note on scoring
I award myself one point per star, but only if I manage to solve it myself without help.
TODO
- Day 12
- Day 17
- Day 19, part 2
- Day 20
- Day 21
- Day 22
- Day 23
- Day 24, part 2
- Day 25
Solution comments in reverse order
Running score: 34 / 50
Day 25: Snowverload
TODO
Score: 0
Leaderboard completion time: 14m01s
Day 24: Never Tell Me The Odds
Part 1 for now, simple linear equations
TODO: Part 2
Score: 1
Leaderboard completion time: 1h02m10s
Day 23: A Long Walk
TODO
Score: 0
Leaderboard completion time: 38m20s
Day 22: Sand Slabs
TODO
Score: 0
Leaderboard completion time: 29m38s
Day 21: Step Counter
Hardest problem of the year so far, gonna look at it later maybe. TODO.
Score: 0
Leaderboard completion time: 1h19m03s
Day 20: Pulse Propagation
TODO
Score: 0
Leaderboard completion time: 48m46s
Day 19: Aplenty
Part one only for now.
TODO: Part 2
Score: 1
Leaderboard completion time: 29m12s
Day 18: Lavaduct Lagoon
This was a fun one. Part 1 was basically interior point finding like in Day 10, but I struggled to get it to work with the “shapes” involved. Finally got an answer… and realized part 2 was gonna be a whole different ballgame.
But I’d head about the Shoelace formula when studying Day 10, and when I saw people referencing it I used that, along with parts of Pick’s theorem to get the correct values.
Score: 2
Leaderboard completion time: 20m55s
Day 17: Clumsy Crucible
I have to give this a pass for now because of prior commitments.
TODO
Score: 0
Leaderboard completion time: 20m00s
Day 16: The Floor Will Be Lava
My first solution to part one used a counter to detect loops, which gave each entry point a runtime of about 30s. I quickly realized it would take 2hrs to bruteforce this, but I had a Christmas tree to deal with so I just set it to run. After about 50% I noticed the number of energized cells wasn’t changing, so I just put that in as the part 2 answer and it was accepted!
Obviously this wasn’t general enough so I coded up a better loop detection. Now the entire brute force runs in 30s…
Incidentally for my input part 2 was for the very next row, but from the opposite direction…
Score: 2
Leaderboard completion time: 15m30s
Day 15: Lens Library
Not the most complex of puzzles, once I figured out the key text in part 2: “The result of running the HASH algorithm on the label indicates the correct box for that step”.
I used a hashref internally for each box, for ease of lookup when replacing lenses. This made adding lenses a bit more complex, but not overly so. I had a lot of issues getting the hashref out when adding the focusing, so the code looks gnarly there.
Score: 2
Leaderboard completion time: 11m04s.
Day 14: Parabolic Reflector Dish
Scruffy solution but at least I know what’s going on.
Dealing with matrices yesterday helped a bit with this one.
Score: 2
Leaderboard completion time: 17m15s.
Day 13: Point of Incidence
This wasn’t my favorite puzzle of the year, but I managed a solution with a little help from the functions in 2020D20.
Score: 2
Leaderboard completion time: 13m46s
Day 12: TODO Hot Springs
Punting on this puzzle for now.
Score: 0
Leaderboard completion time: 22m57s
Day 11: Cosmic Expansion
I’m proud that my part 1 solution was easily expanded to handle part 2: instead of “physically” expanding the map, I just added an offset to each galaxy which were stored in a hash. This made changing the offset from double to 1M trivial.
Score: 2
Leaderboard completion time: 9m18s
Day 10: Pipe Maze
Toughest puzzle so far this year.
On Sun, I found an ugly brute force but had to put off solving part 2.
On Mon I read up a bit on the theory of how to determine interior points and implemented that. I also cleaned up my part 1, using a modified flood-fill algo to “paint” the loop.
Regarding counting crossings, many commentators in the subreddit pointed out that moving diagonally simplified the logic of when the loop was crossed.
Score: 2
Leaderboard completion time: 36m31s
Day 9: Mirage Maintenance
Remarkably easy, especially considering it’s a weekend problem. I suspect we’ll see more of this tomorrow.
Score: 2
Leaderboard completion time: 5m36s
Day 8: Haunted Wasteland
Not the most difficult of puzzles. I realized after letting my first naive stab at part 2 run for 10 minutes that There Had To Be A Better Way. The obvious next step was to find the least common multiple of each different “A to Z” path. I was a bit worried that the path lengths would not be the same with each repetition, but either the puzzle input assured that, or there’s some math reason they’re always the same.
The LCD calculation is from the most excellent ntheory module.
Score: 2
Leaderboard completion time: 10m16s
Day 7: Camel Cards
This was a fun problem, which played to Perl’s strengths.
My first attempt was to get a nice sorting pipeline where I could just add the list of hands and it would first sort by type, and then by value. But I couldn’t really get that to work with input from an array of hashes, so I switched over to determining the type first and having that as subkey in the hashref. This made it easier to sort.
Part 2 presented some problems, but bruteforcing all possible
combinations of joker replacements was not hard at all. I had to deal
with the special case JJJJJ separately.
Score: 2
Leaderboard completion time: 16m00s
Day 6: Wait For It
Mood for this year sure is whiplash. This was easy. Even accounting for the brute-force nature of part 2, runtime is around 20s.
I could sit down and try to dick around with solutions to the quadratic equation, but I’m pretty sure fencepost errors would creep in and I’d spend more time troubleshooting those.
Score: 2
Leaderboard completion time: 5m02s
Day 5: If You Give A Seed A Fertilizer
I solved both parts after thinking about it for a while.
I had the right idea from the start for part 2, which was handling ranges of integers instead of arrays. But I got lost in the weeds trying to get my iterations to to work. Studying some other solutions put me on the right track.
This solution supersedes my earlier part 1, which can be dredged from the repo is one wishes to see it.
Score: 2
Leaderboard completion time: 26m37s
Day 4: Scratchcards
The first attempt at part 2 was literally just implement the instructions in code: I had an array of cards, I picked the first, and depending on that card’s wins and its position in the list, I prepended the cloned cards to the front of the array.
This worked in the sense that I got the correct answer, but it took 5s and a huge array.
After thinking for a bit I realized it was more efficient to simply store the number of cloned cards in an array whose indices were the card ID. Much faster.
Previous solution: commit 39daebc
Score: 2
Leaderboard completion time: 7m08s
Day 3: Gear Ratios
Ah, the map, a AoC classic. I usually use a hashref for these, and check for existence of cells. In this case we throw away all cells that contain a period anyway, so using an arrayref would be wasteful. So we go through the input line by line, adding characters to the map if it’s not a period.
The main loop for part 1 is to go through the completed map. If we find a symbol (non-digit), we look in all 8 directions around it to find something else. My code assumes this will be a digit, and that’s the case with my input.
Once we find a digit, we look left and right around it to find other digits. Once we’ve found a run of consecutive digits, they’re concatenated to a numeral.
My first issue was that I’d get duplicates. Say we have input like this
.......
..123..
...+...
456.789
.......
The numeral 123 will be registered three times. At first I tried to disallow that by only allowing one match per symbol and row, but that made me miss the situation with the 456 and 789 being legal connections but not consecutive.
So what I finally settled on was adding a second hashref recording the position of every symbol and all numerals connected to it. Using the numeral as a key ensured there was only going to be one unique numeral in that position. Then for the summary I looped over that hashref and did the sum for part 1.
This construction made part 2 very easy, I just had to check for asterisk symbols with more than one value attached, and sum their products.
Score: 2
Leaderboard completion time: 11m37s
Day 2: Cube Conundrum
Not a very hard problem, but I overthought how to parse the input and store it in a datastructure, which caused me a lot of problems in the beginning.
Score: 2
Leaderboard completion time: 6m15s
Day 1: Trebuchet?!
Well this was a tough one for day 1! I started out ok enough with splitting on a regex, but quickly was stymied by overlaps like twone and oneight. After searching through my input for these I added special cases which found the correct result.
I then peeked in the subreddit for the specific weird-ass syntax required for overlapping matches, and included that as a comment.
Score: 2
Leaderboard completion time: 7m03s
Puzzles by difficulty (leaderboard completion times)
- Day 21 - TODO Step Counter: 1h19m03s
- Day 24 - Never Tell Me The Odds 1: 1h02m10s
- Day 20 - TODO Pulse Propagation: 48m46s
- Day 23 - TODO A Long Walk: 38m20s
- Day 10 - Pipe Maze: 36m31s
- Day 22 - TODO Sand Slabs: 29m38s
- Day 19 - Aplenty 1: 29m12s
- Day 05 - If You Give A Seed A Fertilizer: 26m37s
- Day 12 - TODO Hot Springs: 22m57s
- Day 18 - Lavaduct Lagoon: 20m55s
- Day 17 - TODO Clumsy Crucible: 20m00s
- Day 14 - Parabolic Reflector Dish: 17m15s.
- Day 07 - Camel Cards: 16m00s
- Day 16 - The Floor Will Be Lava: 15m30s
- Day 25 - TODO Snowverload: 14m01s
- Day 13 - Point of Incidence: 13m46s
- Day 03 - Gear Ratios: 11m37s
- Day 15 - Lens Library: 11m04s.
- Day 08 - Haunted Wasteland: 10m16s
- Day 11 - Cosmic Expansion: 9m18s
- Day 04 - Scratchcards: 7m08s
- Day 01 - Trebuchet: 7m03s
- Day 02 - Cube Conundrum: 6m15s
- Day 09 - Mirage Maintenance: 5m36s
- Day 06 - Wait For It: 5m02s
Previous years: 2015, 2016, 2017, 2018. 2019. 2020. 2021. 2022.