adventofcode2020

Advent Of Code 2020

Day 02: Password policy check

Part 1

No specific difficulty here

Part 2

I made one mistake considering the valdity of a password. I only considered valid a password if the first check was True and the second one was False. This actually was validated by my test suite. I ended up with a wrong answer until I considered that another valid password could be mismatch on the first char and a match on the 2nd char.

Day 03: Toboggan slope trajectory

Part 1

• Wasted time because I failed to strip the input data :(

Part 2

• Learned math.prod is only available in Python >= 3.8.
• Forgot to start parsing the input from line == down_step. data[down_step:0] instead of data[1:].

Day 04: Passport validation

Part 1

• Did not process the last passport because it was triggered by empty lines; first answer submission was short by 1 :(

Part 2

• Tried to validate in one single if statement but couldn't debug.
• Broke down the vailidation tests in functions and unit tested them.
• Careful with range: 5 is not in range(0,5), it is in range(0,6).
• Improved my regexp skills - trying to capture groups but doing one single eval.
• Discovered a Python 3.8 feature: the walrus operator.

Day 05: Finding a seat on the plane

Part 1

• I didn't realize directly the analogy with binary representation.

Part 2

• Lost time trying to optimize with lambda or itertools.
• I still believe there must be an elegant set comparison that can compare "neighbor" seats.

Day 06: Passenger customs form

Part 1

• Need to find a better way to turn a str to a set of chars.

Part 2

• Need to find a better way to initialize the set at first line of the group.

Day 07: Luggage check

Part 1

• I was wondering when recusrsion was going to start ?

Part 2

• I shouldn't have assumed that 'gold' was a sufficient filter, 'shiny gold' was necessary since other types of gold were in the list of colors.

Day 08: Handheld instruction set

Part 1

• Got bitten by empty lines at the end of the input file.

Part 2

• Found an iterative way to get to the solution.
• Can't find a working recursive methodology :(

Part 2b

• I really wanted to improve the solution with a recursive function; I couldn't do it without deep copying the instruction set - which is probably not very efficient.

Day 09: Protocol decryption

Part 1

• Had to isolate pair in a function - wish I could do without the nested loop.

Part 2

• too many off-by-one notations but works OK.

Day 10: Adapter mess

Part 1

• Not elegant but not complex.
• Took a little while to figure out what the problem exactly was.
• Found out it only takes one unknown word (jolt) to make a problem much harder to grasp.

Part 2

• I won't lie, I suffered ... but the satisfaction is proportional
• The testing of the puzzle use case was succesful but the solving of my input took hours to execute ...
• First you think it might take a few mins so my_lazy_and_confident_self is satisfied and just keeps an eye on the laptop So here is what my morning was like: Tested the part2 on the demo input, yeah, it works, run it with the real puzzle input ... hmmm, slower, but my_lazy_and_super_confident_self knows it is a matter of seconds or maybe even a few minutes...
• hmmm, still no output, let's go shower quickly,
• hmmm, still no ouput, let's wake up the wife,
• hmmm, still no output ... but 3% battery, run to find your laptop adapter,
• hmmm, still no output ... prepare breakfast for the kids
• hmmm, still no output ... take the kids to school ... but don't ask me to smile
• really, 2 hours and still no output ... maybe I could start working on a more efficient solution ...
• OK, time's up, it will be for tonight, let's get to work
• Oh, but wait, maybe ... I think ... if I did this ... it should ... let's try ... failed ... back to work for 10 min
• repeat until noon until it finally beautifully executed in 0.02sec

Day 11: Seat allocation

No particular comment, just couldn't find time to finish it in the morning.

Day 12: Seat allocation

Part2

Not fully satisfied with the direction shifting, there probably are more arithmetic ways to store waypoint and position coordinates.

Day 13: Bus schedules

Part1

• No particular difficulty

Part2

• I struggled hard - after multiple hours of not finding, I finally went to subreddit and used a solution for inspiration
• My solution actually worked but way too slow - I kept on bruteforcing with a constant increase (equal to max(buses).
• The main limitation was that I could not find how to increase the step.
• It turned out the right way is to compare 2 buses, then go to the next one by multiplying the increase by the value of the new bus.
• Another hint was to ignore the 'x' values by replacing them with 0 instead of keeping a dict with {offset: bus}.
• Interesting read about the Chinese Remainder Theory (https://en.wikipedia.org/wiki/Chinese_remainder_theorem).

Day 14: Docking data and bitwise operations

Part 1:

• Refreshing reminder about bitwise operations.

Part 2:

• Nothing particular

Day 15: Memory game with the elves

Part 1:

• No particular difficulty

Part2:

• Took ~1.5h to reach acceptable performance (execution time: 57sec).
• Interesting to notice that removing the counter (n 'if' statement) didn't reduce the execution speed.
• Trimmed 2 secs by using IndexErrors instead of testing index validity: ~55sec.
• Trimmed 2 secs by using Python 3.6.8 instead of Python 2.7.16.
• No impact when using a list instead of a dict to hold the numbers.
• ~39sec using a Memory class (day15b.py)

Day 16:

Part 1:

• As usual, pretty straightforward

Part 2:

• Quickly came to the solution but it took a serious amount of troubleshooting to understand a function was returning an int or None if not found; later on, this value was tested with "if value:". The side effect of this is that this became False if the function returned an index of 0 :(((
• On to day 17 now, personal record broken at this point ;)

Day 17:

Part 1:

• This one took some time to understand the best way to model it.

Part 2:

• Once P1 was done, this one was 8 mins.

Day 18:

Part 1:

• Completed on time but needs some refactoring before I share on GitHub

Part 2:

• Completed on time but needs some refactoring before I share on GitHub

Day 19:

Part 1:

• Spent more time on Strava than on AoC ;)
• Needs to catch up with the backlog

Part 2:

• [2020/12/27] Finally wrapped my head around this one. Got a hint from a friend, and gave up the elegant solution for the bruteforce "capped recursivity" - Not 100% satisfied but moving on ;)

Day 20:

Part 1:

• Got one silver star for day 20 on day 20

Part 2:

• Overlooked the instruction to remove the border from the tiles :((

Day 21:

Part 1

• Haven't opened the problem on day 21.
• Hoping to get 19 and 20 finished by tomorrow (22) and get to 21 after tomorrow (23)
• From rank #40 to #rank 60 on Matt Goodrich's leaderboard
• [2020/12/28] Got inspiration from https://github.com/UnicycleBloke/aoc2020/blob/main/day21/day21.py abut the right use of sets.

Part 2:

• Not the cleanest way but good enough for today.
• Read extensively about the usage of argument splitting.

Day 22:

Part 1:

• First time usage of queue.Queue() - working as intended ;)

Part 2:

Day 23:

Part 1:

• No major issue

Part 2:

• Part 1 did not scale with larger numbers
• Changed the array structure to a chain of pointers (nostalgic flashback from CS classes in '96)
• Works sub 25 sec, which is OK

Day 24:

Part 1:

• First time using an hex grid, proud to have found a working solution without external inspiration.
• No major issue

Part 2:

• I failed to understand I needed to make the grid bigger (by 2 units in each direction) at every turn. I still think it was not clear in the problem description.
• Oooh, and we are the 19th of Feb, I missed the end of year deadline, procrastinated for 50 days now but I'm not giving up ;)

Interesting links

Utility from salt-die

https://github.com/norvig/pytudes/blob/master/ipynb/Advent-2020.ipynb

• nice resources around Python optimisation

https://github.com/salt-die/Advent-of-Code/blob/master/2020/aoc_helper

• automated download of input
• automated submission
• automated template-based per-day directory creation