Write-Ups.md

Write-Ups

Day 1:

• honestly wasn't all that difficult
• adapting script for day 2 was a bit of effort but honestly took like 2 minutes

Day 2:

• literally wasn't difficult at all
• just the addition of a third variable was all it took to complete part 2

Day 3:

• ok this one was actually difficult
• i had to rewrite the entire program to satisfy part 2
• one of the things i like is utilizing the very little things about python that i almost never use on a daily basis, solely for the purpose of optimizing these

Day 4:

• literally had to write part 1 twice just for it to work
• first attempt was really complex
• second attempt was held together with spit and gum
• once part 1 worked, i had to take what i wrote apart so that i could reverse the whole process
• took literal hours to complete

Day 5:

• was going back and forth on using numpy
• constantly got memoryerrors because there was just too much data
• had to rewrite multiple times
• ultimately reused part 1 solution for part 2 but slightly modified it

Day 6:

• way too easy
• incredibly easier than days 4 and 5
• i think i was just graced with the idea of using a dictionary as early as i did

Day 7:

• lmao what was this
• (i've been informed that weekend problems are made intentionally harder than weekday problems)
• still this one was way too easy
• a lot easier than day 6's
• but i've finally completed a whole week!

Day 8:

• i like to call this one "list comprehension ad nauseum"
• day 1 was really easy
• day 2 was much more difficult but still fun
• works by first attempting to solve for each individual segment, then building each digit from the segments it uses

Day 9:

• i was going to complete it within an hour after it was released
• but i couldn't figure out how to get the top 3 items in a list
• (also the recursive branch function i had beforehand wasn't working)
• so i implemented my own

Day 10:

• 10 DAYS LETS GO
• also really easy
• finished in 11 minutes
• so i finally achieved my goal of completing a day in the first hour it was given

Day 11:

• day 2 was super easy
• just had to add an extra variable and change some stuff
• recurring theme is "it doesn't look pretty but it works"

Day 12:

• pretty easy
• see i hate pathfinding algorithms so i thought today was going to be absolute hell
• but then i gave it a shot and it turned out pretty well
• (runs a bit slowly)
• reached 12 days of completely vanilla solutions

Day 13:

• was somewhat difficult as the phrase The manual says the code is always eight capital letters. was hard to understand
• still managed to finish

Day 14:

• hell
• on earth
• part 1 was fine but part 2 was awful

Day 15:

• well
• i hate to say it
• but i've officially conceded the full vanilla challenge
• there's no easy way to make a priority queue without using queue.PriorityQueue
• Day 15 Part 2 Vanilla.py has an implementation written without queue.PriorityQueue, it just doesn't run in reasonable time
• UPDATE (12/20): thanks to the amazing xelf over at pydis, this can be done in a reasonable time completely vanilla by using dijkstra instead of a*
• it takes an extremely long time to finish (2 hours 44 minutes 18 seconds to be exact), but it can finish in time

Day 16:

• before today, i'd consider some of the solutions i've made this year unnecessarily complicated and overengineered, but not bad
• today i have created the first bad solution of the year
• and it only took 11 hours, 5+ people and a third-party decoder/interpreter
• god i hate opcode

Day 17:

• by some miracle
• brute forcing worked
• and the math looked way too complex for me
• because im not in physics
• so i resorted to brute forcing, because why not

Day 18:

• either i'm a genius
• or i'm completely crazy
• every snailfish is parsed as a string rather than nested lists
• and by some miracle part 2 runs in a minute

Day 19:

• so these last few days i've been abusing the soft time limits
• this one runs in about 3 minutes
• and i somehow managed to complete it without using numpy

Day 20:

• today took way too long
• 5 hours to complete, 15-20 minutes to run each file
• due to a difference between the sample and puzzle inputs, today's files are split up into sample-compatible files and puzzle-compatible files

Day 21:

• giant shoutout to mustafa at pydis for optimizing this
• this one was dumb, confusing, and nearly crashed my system 3 times
• part 2 was significantly harder than part 1

Day 22:

• this is probably the first day that i am completely unsatisfied with my answer
• part 2 was so difficult to understand that i had to look at someone else's answer to understand it
• and only after completing the problem and looking at their answer multiple times over did i truly understand it
• but i'm not a fan of having to use someone else's solution to understand the problem
• it feels like cheating, kinda
• but now i understand the problem, i know what it was asking for, and i know how to implement a solution (or similar solutions in the future)

Day 23:

• oops! no code
• today's parts were done almost entirely by hand, with the help of someone else's solution to give the first few steps needed to solve part 2
• one thing that was made extremely apparent today is that in a lot of instances, i just need a bit of a nudge to start the process to get to the final answer
• and often i'm missing that initial push because i chose to make this year harder for myself than it should be (full vanilla, 24 hour completion time)
• as i'm missing available resources + setting a relatively short time limit, which increase stress
• something that i'm going to do next year is not apply these extreme constraints for myself, because they end up causing me to fail
• in the same vein as day 22, i'm a little unsatisfied that i had to look at someone else's answer to understand their thinking
• but after implementing it myself, i was able to understand the problem better
• now i'm better prepared for future problems
• perhaps the dissatisfaction stems from stubbornness? that i couldn't complete a day 100% on my own

Day 24:

• once again, thank you mustafa for making this 100 times easier to understand
• today i've learned that i can understand logic better if it's written in (roughly) the same style as python code
• i've also learned that these hard problems can be interpreted better by just manually taking an input and going through all the required steps
• now that i think about it, something similar could (or should) have been done with day 20
• but after 24 days, i've finally made it

Day 25:

• after 25 days, i've finally made it to the end
• an entire advent calendar has been completed
• at last, i'm free

BONUS:

• on 12/13 i got a bonus challenge as part of a competition for discord nitro/$100 steam credit
• so i decided to do it
• i included it here because it's somewhat similar
• (i think i was the first person to finish)
• between 12/13 and 12/20 i optimized it so much that i managed to cut down its execution time down to 16.5% of what it was originally