This is a repository of personal notes and solutions to this coding tournament / challenge.
For a first participation, I was not at all in the competitive mindset, but aimed at being a "finisher" of the 50 puzzles (I'm still working on it, starting day 22 soon, then I'll go back to the ones I left behind).
I saw that a library to ease the processing of input data and automate entry submission could be useful. For example, the first player in the leaderboard used one for Kotlin.
So I started my own in the aoc_lib.py file, with a blank template to use it in the TEMPLATE.py file.
I'll publish my solutions for days 1 - 9, 11 - 13 once they're retrofitted to this new toolkit and template...
As mentioned on Advent of Code website:
Can I copy/redistribute part of Advent of Code? Please don't. Advent of Code is free to use, not free to copy. If you're posting a code repository somewhere, please don't include parts of Advent of Code like the puzzle text or your inputs.
The competition is now over. The top players were all incredible!
Check the time it took them to get the first and both stars:
| day | **/1st | **/100th | */1st | */100th | Comments and hints |
|---|---|---|---|---|---|
| 25 | 00:02:53 | 00:14:01 | 00:02:43 | 00:13:07 | TODO... |
| 24 | 00:12:42 | 01:02:10 | 00:04:17 | 00:15:56 | TODO... |
| 23 | 00:07:09 | 00:38:20 | 00:02:08 | 00:09:02 | TODO... |
| 22 | 00:12:50 | 00:29:48 | 00:10:04 | 00:21:00 | TODO... |
| 21 | 00:14:35 | 01:19:03 | 00:01:24 | 00:04:44 | HALF WAY. The first part is easy but not that interesting. However it's obvious that the second one won't be brute forced. Perhaps some pattern in the data could help solve it differently? For example, the fact that all the borders of the base pattern are free of rocks? |
| 20 | 00:15:52 | 00:48:46 | 00:10:26 | 00:23:11 | HALF WAY. The first part was again very enjoyable, making use of queues and "complex" data structures. The second part is deceptively simple but this is not the way to proceed. My second part program processes over 1 billion button presses per day, but the result seeked is over 50 billions ! |
| 19 | 00:12:56 | 00:29:12 | 00:02:01 | 00:10:23 | OK. Very pleasant one. Using some regular expressions and recursivity. I provided a file explaining how the probabilities are computed as an hint. |
| 18 | 00:05:31 | 00:20:55 | 00:02:35 | 00:08:21 | HALF WAY. The first part was quite fun and I found a flood filling algorithm that could also be useful in the future. However it appears I'll have to rewrite my code completely to tackle part 2! |
| 17 | 00:05:03 | 00:20:00 | 00:02:25 | 00:14:05 | TODO. Skipping that one until I can review Dijkstra's algorithm. Then I'll need to adapt it |
| 16 | 00:04:33 | 00:15:30 | 00:02:03 | 00:11:36 | OK. That one was also easy, though I needed to take care of never ending beams in first part that blown up the Python recursion limit! |
| 15 | 00:01:26 | 00:11:04 | 00:00:51 | 00:02:10 | OK. That one was surprisingly easy :-) |
| 14 | 00:05:46 | 00:17:15 | 00:01:28 | 00:04:10 | OK. With 1 billion cycles, the second part cannot be brute forced, but after a while a cycle appears that avoids computing all of it |
| 13 | 00:06:06 | 00:13:46 | 00:03:51 | 00:08:58 | TODO. The specification for this one doesn't cover all the cases that can be encountered and my program, although seeming correct, doesn't even give the same result with the example data... I'll get it back at it later on |
| 12 | 00:05:58 | 00:05:58 | 00:02:02 | 00:08:12 | HALF WAY. I need to work on an optimized algorithm for the second part |
| 11 | 00:03:37 | 00:09:18 | 00:02:00 | 00:06:07 | OK |
| 10 | 00:10:04 | 00:36:31 | 00:01:05 | 00:11:33 | OK. I was stuck for a long time (about 3 weeks!) on the second part until I realised that FJ or L7 count as a single wall. I provided a file explaining why with better looking characters. I really needed to experience algorithmic warmup :-) |
| 9 | 00:01:48 | 00:05:36 | 00:01:16 | 00:04:02 | OK |
| 8 | 00:04:24 | 00:10:16 | 00:00:38 | 00:03:30 | HALF WAY. I need to work on an optimized algorithm for the second part |
| 7 | 00:08:45 | 00:16:00 | 00:04:52 | 00:09:57 | HALF WAY. The second part works with the example data, but not the validation data. I need to cross check my results with the solution of someone else to find my mistake(s) |
| 6 | 00:02:05 | 00:05:02 | 00:01:15 | 00:03:11 | OK |
| 5 | 00:08:38 | 00:26:37 | 00:01:35 | 00:08:15 | OK. I got the result for the second part but it took a day to compute, with multiprocessing on a multi-cores machine. There must be a better way to do it! This got me wondering if I was using the right programmation language to do this challenge, but re-reading the rules, it's clearly stated that it's possible to get the results on a 10 years old machine, and seeing the leaderboard there are many people using interpreted languages. So there are better algorithms to be found... |
| 4 | 00:01:22 | 00:07:08 | 00:00:43 | 00:02:51 | OK |
| 3 | 00:05:09 | 00:11:37 | 00:03:11 | 00:07:09 | OK |
| 2 | 00:01:34 | 00:06:15 | 00:00:37 | 00:04:10 | OK |
| 1 | 00:02:24 | 00:07:03 | 00:00:12 | 00:01:39 | OK. Looking at the best competitor: 12 seconds to read the spec, code and submit the result for the first star, how can this even be possible?! |
Half of the top 100 leaders have a GitHub account, and one third of them have published their AoC 2023 code.
Among them, I noticed Jonathan Paulson (#5) Python solutions, and mrphlip (#27) commented Python solutions.
There are some other interesting resources on GitHub, such as a collection of awesome resources related to the yearly Advent of Code challenge, or on the Web, such as an article about Solving Your Puzzles With Python.