Day 13: Claw Contraption

Megathread guidelines

  • Keep top level comments as only solutions, if you want to say something other than a solution put it in a new post. (replies to comments can be whatever)
  • You can send code in code blocks by using three backticks, the code, and then three backticks or use something such as https://topaz.github.io/paste/ if you prefer sending it through a URL

FAQ

4 points

Haskell

Whee, linear algebra! Converting between numeric types is a bit annoying in Haskell, but I’m reasonably happy with this solution.

import Control.Monad
import Data.Matrix qualified as M
import Data.Maybe
import Data.Ratio
import Data.Vector qualified as V
import Text.Parsec

type C = (Int, Int)

readInput :: String -> [(C, C, C)]
readInput = either (error . show) id . parse (machine `sepBy` newline) ""
  where
    machine = (,,) <$> coords <*> coords <*> coords
    coords =
      (,)
        <$> (manyTill anyChar (string ": X") >> anyChar >> num)
        <*> (string ", Y" >> anyChar >> num)
        <* newline
    num = read <$> many1 digit

presses :: (C, C, C) -> Maybe C
presses ((ax, ay), (bx, by), (px, py)) =
  do
    let m = fromIntegral <$> M.fromLists [[ax, bx], [ay, by]]
    m' <- either (const Nothing) Just $ M.inverse m
    let [a, b] = M.toList $ m' * M.colVector (fromIntegral <$> V.fromList [px, py])
    guard $ denominator a == 1
    guard $ denominator b == 1
    return (numerator a, numerator b)

main = do
  input <- readInput <$> readFile "input13"
  mapM_
    (print . sum . map (\(a, b) -> 3 * a + b) . mapMaybe presses)
    [ input,
      map (\(a, b, (px, py)) -> (a, b, (10000000000000 + px, 10000000000000 + py))) input
    ]
permalink
report
reply
4 points
*

Haskell

Pen and Paper solved these equations for me.

import Control.Arrow

import qualified Data.Char as Char
import qualified Data.List as List
import qualified Data.Maybe as Maybe


window6 :: [Int] -> [[Int]]
window6 [] = []
window6 is = List.splitAt 6 
        >>> second window6
        >>> uncurry (:)
        $ is

parse :: String -> [[Int]]
parse s = window6 . map read . words . List.filter ((Char.isDigit &&& Char.isSpace) >>> uncurry (||)) $ s

solveEquation (ax:ay:bx:by:tx:ty:[]) transformT
        | (aNum `mod` aDenom) /= 0   = Nothing
        | (bNum `mod` bDenom) /= 0   = Nothing
        | otherwise                  = Just (abs $ aNum `div` aDenom, abs $ bNum `div` bDenom)
        where
                tx' = transformT tx
                ty' = transformT ty
                aNum   = (bx*ty')  - (by*tx')
                aDenom = (ax*by)   - (bx*ay)
                bNum   = (ax*ty')  - (ay*tx')
                bDenom = (ax*by)   - (bx*ay)

part1 = map (flip solveEquation id)
        >>> Maybe.catMaybes
        >>> map (first (*3))
        >>> map (uncurry (+))
        >>> sum
part2 = map (flip solveEquation (+ 10000000000000))
        >>> Maybe.catMaybes
        >>> map (first (*3))
        >>> map (uncurry (+))
        >>> sum

main = getContents
        >>= print
        . (part1 &&& part2)
        . parse

(Edit: coding style)

permalink
report
reply
3 points
*

Python

Execution time: ~<1 millisecond (800 microseconds on my machine)

Good old school linear algebra from middle school. we can solve this really really fast. With minimal changes from part 1!

FastCode

[ paste ]

from time import perf_counter_ns
import string

def profiler(method):
    def wrapper_method(*args: any, **kwargs: any) -> any:
        start_time = perf_counter_ns()
        ret = method(*args, **kwargs)
        stop_time = perf_counter_ns() - start_time
        time_len = min(9, ((len(str(stop_time))-1)//3)*3)
        time_conversion = {9: 'seconds', 6: 'milliseconds', 3: 'microseconds', 0: 'nanoseconds'}
        print(f"Method {method.__name__} took : {stop_time / (10**time_len)} {time_conversion[time_len]}")
        return ret

    return wrapper_method

@profiler
def main(input_data):
    part1_total_cost = 0
    part2_total_cost = 0
    for machine in input_data:
        Ax,Ay,Bx,By,Px,Py = [ int(l[2:]) for l in machine.split() if l[-1] in string.digits ]
        y,r = divmod((Ay * Px - Ax * Py), (Ay * Bx - Ax * By))
        if r == 0:
            x,r = divmod(Px - Bx * y, Ax)
            if r == 0:
                part1_total_cost += 3*x + y
        y,r = divmod((Ay * (Px+10000000000000) - Ax * (Py+10000000000000)), (Ay * Bx - Ax * By))
        if r == 0:
            x,r = divmod((Px+10000000000000) - Bx * y, Ax)
            if r == 0:
                part2_total_cost += 3*x + y

    return part1_total_cost,part2_total_cost

if __name__ == "__main__":
    with open('input', 'r') as f:
        input_data = f.read().strip().replace(',', '').split('\n\n')
    part_one, part_two = main(input_data)
    print(f"Part 1: {part_one}\nPart 2: {part_two}")

permalink
report
reply
2 points

This is a really excellent, clean solution! Would you mind breaking down how the piece of linear algebra works (for a shmo like me who doesn’t remember that stuff frum school heh πŸ˜…)

permalink
report
parent
reply
3 points

https://lemmy.world/comment/13950499

take the two equations, solve for y, and make sure y is fully divisible.

permalink
report
parent
reply
2 points

It’s interesting that you’re not checking if the solution to x is a whole number. I guess the data doesn’t contain any counterexamples.

permalink
report
parent
reply
1 point
*

They do, if the remainder returned by divmod(…) wasn’t zero then it wouldn’t be divisble

permalink
report
parent
reply
2 points

you are right, we solve for y, but I am confident that solving for x after y would yield the correct result as long as y is fully divisible.

permalink
report
parent
reply
3 points
*

we are solving for y first. If there is a y then x is found easily.

(Ax)*x + (Bx)*y = Px and (Ay)*x + (By)*y = Py

Because of Ax or Ay and Bx or By, lets pretend that they are not (A*x)*x and (A*y)*y for both. they are just names. could be rewritten as: (Aleft)*x + (Bleft)*y = Pleft and (Aright)*x + (Bright)*y = Pright

but I will keep them short. solving for y turns into this:

y = (Ay*Px - Ax*Py) / (Ay*Bx - Ax*By)

if mod of 1 is equal to 0 then there is a solution. We can be confident that x is also a solution, too. Could there be an edge case? I didn’t proof it, but it works flawlessly for my solution.

Thankfully, divmod does both division and mod of 1 at the same time.

permalink
report
parent
reply
2 points
*

Thank you so much for your explanation! I kind of found some clues looking up perp dot products & other vector math things, but this breaks it down very nicely.

I implemented your solution in rust, and part 2 failed by +447,761,194,259 (this was using signed 64-bit integers, i64). When I changed it to use signed 64 bit floating-point f64 and checked that the solution for x produces a whole number it worked.

permalink
report
parent
reply
2 points
*

Python

I just threw linear algebra and float64 on this question and it stuck. Initially in order to decrease the numbers a bit (to save precision) I tried to find greatest common divisors for the coordinates of the target but in many cases it was 1, so that was that went down the drain. Luckily float64 was able to achieve precisions up to 1e-4 and that was enough to separate wheat from chaff. So in the end I did not have to use exact formulas for the inverse of the matrix though probably would be a more satisfying solution if I did.


import numpy as np
from functools import partial
from pathlib import Path
cwd = Path(__file__).parent

def parse_input(file_path, correction):

  with file_path.open("r") as fp:
    instructions = fp.readlines()

  machine_instructions = []
  for ind in range(0,len(instructions)+1,4):

    mins = instructions[ind:ind+3]
    machine_instructions.append([])
    for i,s in zip(range(3),['+','+','=']):
      machine_instructions[-1].append([int(mins[i].split(',')[0].split(s)[-1]),
                                   int(mins[i].split(',')[1].split(s)[-1])])

    for i in range(2):
      machine_instructions[-1][-1][i] += correction

  return machine_instructions


def solve(threshold, maxn, vectors):

  c = np.array([3, 1])

  M = np.concat([np.array(vectors[0])[:,None],
                 np.array(vectors[1])[:,None]],axis=1).astype(int)

  if np.linalg.det(M)==0:
    return np.nan

  Minv = np.linalg.inv(M)
  nmoves = Minv @ np.array(vectors[2])

  if np.any(np.abs(nmoves - np.round(nmoves))>threshold) or\
    np.any(nmoves>maxn) or np.any(nmoves<0):
      return np.nan

  return np.sum(c * (Minv @ np.array(vectors[2])))


def solve_problem(file_name, correction, maxn, threshold=1e-4):
  # correction 0 or 10000000000000
  # maxn 100 or np.inf

  machine_instructions = parse_input(Path(cwd, file_name), correction)

  _solve = partial(solve, threshold, maxn)

  tokens = list(map(_solve, machine_instructions))

  return int(np.nansum(list(tokens)))
permalink
report
reply
2 points

C#

public partial class Day13 : Solver
{
  private record struct Button(int X, int Y);
  private record struct Machine(int X, int Y, Button A, Button B);
  private List<Machine> machines = [];

  [GeneratedRegex(@"^Button (A|B): X\+(\d+), Y\+(\d+)$")]
  private static partial Regex ButtonSpec();

  [GeneratedRegex(@"^Prize: X=(\d+), Y=(\d+)$")]
  private static partial Regex PrizeSpec();

  public void Presolve(string input) {
    var machine_specs = input.Trim().Split("\n\n").ToList();
    foreach (var spec in machine_specs) {
      var lines = spec.Split("\n").ToList();
      if (ButtonSpec().Match(lines[0]) is not { Success: true } button_a_match
        || ButtonSpec().Match(lines[1]) is not { Success: true } button_b_match
        || PrizeSpec().Match(lines[2]) is not { Success:true} prize_match) {
        throw new InvalidDataException($"parse error: ${lines}");
      }
      machines.Add(new Machine(
        int.Parse(prize_match.Groups[1].Value),
        int.Parse(prize_match.Groups[2].Value),
        new Button(int.Parse(button_a_match.Groups[2].Value), int.Parse(button_a_match.Groups[3].Value)),
        new Button(int.Parse(button_b_match.Groups[2].Value), int.Parse(button_b_match.Groups[3].Value))
        ));
    }
  }

  private string Solve(bool unit_conversion) {
    BigInteger total_cost = 0;
    foreach (var machine in machines) {
      long prize_x = machine.X + (unit_conversion ? 10000000000000 : 0);
      long prize_y = machine.Y + (unit_conversion ? 10000000000000 : 0);
      BigInteger det = machine.A.X * machine.B.Y - machine.B.X * machine.A.Y;
      if (det == 0) continue;
      BigInteger det_a = prize_x * machine.B.Y - machine.B.X * prize_y;
      BigInteger det_b = prize_y * machine.A.X - machine.A.Y * prize_x;
      var (a, a_rem) = BigInteger.DivRem(det_a, det);
      var (b, b_rem) = BigInteger.DivRem(det_b, det);
      if (a_rem != 0 || b_rem != 0) continue;
      total_cost += a * 3 + b;
    }
    return total_cost.ToString();
  }

  public string SolveFirst() => Solve(false);
  public string SolveSecond() => Solve(true);
}
permalink
report
reply

Advent Of Code

!advent_of_code@programming.dev

Create post

An unofficial home for the advent of code community on programming.dev!

Advent of Code is an annual Advent calendar of small programming puzzles for a variety of skill sets and skill levels that can be solved in any programming language you like.

AoC 2024

Solution Threads

M T W T F S S
1
2 3 4 5 6 7 8
9 10 11 12 13 14 15
16 17 18 19 20 21 22
23 24 25

Rules/Guidelines

  • Follow the programming.dev instance rules
  • Keep all content related to advent of code in some way
  • If what youre posting relates to a day, put in brackets the year and then day number in front of the post title (e.g. [2024 Day 10])
  • When an event is running, keep solutions in the solution megathread to avoid the community getting spammed with posts

Relevant Communities

Relevant Links

Credits

Icon base by Lorc under CC BY 3.0 with modifications to add a gradient

console.log('Hello World')

Community stats

  • 477

    Monthly active users

  • 110

    Posts

  • 1.1K

    Comments