Today was easy enough that I felt confident enough to hammer out a solution in Uiua, so read and enjoy (or try it out live):

{"Time:      7  15   30"
 "Distance:  9  40  200"}
StoInt ← /(+ ×10) ▽×⊃(≥0)(≤9). -@0
Count ← (
  ⊙⊢√-×4:×.⍘⊟.           # Determinant, and time
  +1-⊃(+1↥0⌊÷2-)(-1⌈÷2+) # Diff of sanitised roots
)
≡(↘1⊐⊜∘≠@\s.)
⊃(/×≡Count⍉∵StoInt)(Count⍉≡(StoInt⊐/⊂))

Rust

I went with solving the quadratic equation, so part 2 was just a trivial change in parsing. It was a bit janky to find the integer that is strictly larger than a floating point number, but it all worked out.

Nim

Hey, waitaminute, this isn’t a programming puzzle. This is algebra homework!

Part 2 only required a trivial change to the parsing, the rest of the code still worked. I kept the data as singleton arrays to keep it compatible.

• Day 6, part 1
• Day 6, part 2

Dart Solution

I decided to use the quadratic formula to solve part 1 which slowed me down while I struggled to remember how it went, but meant that part 2 was a one line change.

This year really is a roller coaster…

int countGoodDistances(int time, int targetDistance) {
  var det = sqrt(time * time - 4 * targetDistance);
  return (((time + det) / 2).ceil() - 1) -
      (max(((time - det) / 2).floor(), 0) + 1) +
      1;
}

solve(List> data, [param]) {
  var distances = data.first
      .indices()
      .map((ix) => countGoodDistances(data[0][ix], data[1][ix]));
  return distances.reduce((s, t) => s * t);
}

getNums(l) => l.split(RegExp(r'\s+')).skip(1);

part1(List lines) =>
    solve([for (var l in lines) getNums(l).map(int.parse).toList()]);

part2(List lines) => solve([
      for (var l in lines) [int.parse(getNums(l).join(''))]
    ]);

Rust

Feedback welcome! Feel like I’m getting the hand of Rust more and more.

use regex::Regex;
pub fn part_1(input: &str) {
    let lines: Vec<&str> = input.lines().collect();
    let time_data = number_string_to_vec(lines[0]);
    let distance_data = number_string_to_vec(lines[1]);

    // Zip time and distance into a single iterator
    let data_iterator = time_data.iter().zip(distance_data.iter());

    let mut total_possible_wins = 1;
    for (time, dist_req) in data_iterator {
        total_possible_wins *= calc_possible_wins(*time, *dist_req)
    }
    println!("part possible wins: {:?}", total_possible_wins);
}

pub fn part_2(input: &str) {
    let lines: Vec<&str> = input.lines().collect();
    let time_data = number_string_to_vec(&lines[0].replace(" ", ""));
    let distance_data = number_string_to_vec(&lines[1].replace(" ", ""));

    let total_possible_wins = calc_possible_wins(time_data[0], distance_data[0]);
    println!("part 2 possible wins: {:?}", total_possible_wins);
}

pub fn calc_possible_wins(time: u64, dist_req: u64) -> u64 {
    let mut ways_to_win: u64 = 0;

    // Second half is a mirror of the first half, so only calculate first part
    for push_time in 1..=time / 2 {
        // If a push_time crosses threshold the following ones will too so break loop
        if push_time * (time - push_time) > dist_req {
            // There are (time+1) options (including 0).
            // Subtract twice the minimum required push time, also removing the longest push times
            ways_to_win += time + 1 - 2 * push_time;
            break;
        }
    }
    ways_to_win
}

fn number_string_to_vec(input: &str) -> Vec {
    let regex_number = Regex::new(r"\d+").unwrap();
    let numbers: Vec = regex_number
        .find_iter(input)
        .filter_map(|m| m.as_str().parse().ok())
        .collect();
    numbers
}

Haskell

This problem has a nice closed form solution, but brute force also works.

(My keyboard broke during part two. Yet another day off the bottom of the leaderboard…)

import Control.Monad
import Data.Bifunctor
import Data.List

readInput :: String -> [(Int, Int)]
readInput = map (\[t, d] -> (read t, read d)) . tail . transpose . map words . lines

-- Quadratic formula
wins :: (Int, Int) -> Int
wins (t, d) =
  let c = fromIntegral t / 2 :: Double
      h = sqrt (fromIntegral $ t * t - 4 * d) / 2
   in ceiling (c + h) - floor (c - h) - 1

main = do
  input <- readInput <$> readFile "input06"
  print $ product . map wins $ input
  print $ wins . join bimap (read . concatMap show) . unzip $ input

A nice simple one today. And only a half second delay for part two instead of half an hour. What a treat. I could probably have nicer input parsing, but that seems to be the theme this year, so that will become a big focus of my next round through these I’m guessing. The algorithm here to get the winning possibilities could also probably be improved upon by figuring out what the number of seconds for the current record is, and only looping from there until hitting a number that doesn’t win, as opposed to brute-forcing the whole loop.

https://github.com/capitalpb/advent_of_code_2023/blob/main/src/solvers/day06.rs

#[derive(Debug)]
struct Race {
    time: u64,
    distance: u64,
}

impl Race {
    fn possible_ways_to_win(&self) -> usize {
        (0..=self.time)
            .filter(|time| time * (self.time - time) > self.distance)
            .count()
    }
}

pub struct Day06;

impl Solver for Day06 {
    fn star_one(&self, input: &str) -> String {
        let mut race_data = input
            .lines()
            .map(|line| {
                line.split_once(':')
                    .unwrap()
                    .1
                    .split_ascii_whitespace()
                    .filter_map(|number| number.parse::().ok())
                    .collect::>()
            })
            .collect::>();

        let times = race_data.pop().unwrap();
        let distances = race_data.pop().unwrap();

        let races = distances
            .into_iter()
            .zip(times)
            .map(|(time, distance)| Race { time, distance })
            .collect::>();

        races
            .iter()
            .map(|race| race.possible_ways_to_win())
            .fold(1, |acc, count| acc * count)
            .to_string()
    }

    fn star_two(&self, input: &str) -> String {
        let race_data = input
            .lines()
            .map(|line| {
                line.split_once(':')
                    .unwrap()
                    .1
                    .replace(" ", "")
                    .parse::()
                    .unwrap()
            })
            .collect::>();

        let race = Race {
            time: race_data[0],
            distance: race_data[1],
        };

        race.possible_ways_to_win().to_string()
    }
}

A nice change of pace from the previous puzzles, more maths and less parsing

import math
import re


def create_table(filename: str) -> list[tuple[int, int]]:
    with open('day6.txt', 'r', encoding='utf-8') as file:
        times: list[str] = re.findall(r'\d+', file.readline())
        distances: list[str] = re.findall(r'\d+', file.readline())
    table: list[tuple[int, int]] = []
    for t, d in zip(times, distances):
        table.append((int(t), int(d)))
    return table


def get_possible_times_num(table_entry: tuple[int, int]) -> int:
    t, d = table_entry
    l_border: int = math.ceil(0.5 * (t - math.sqrt(t**2 -4 * d)) + 0.0000000000001)  # Add small num to ensure you round up on whole numbers
    r_border: int = math.floor(0.5*(math.sqrt(t**2 - 4 * d) + t) - 0.0000000000001)  # Subtract small num to ensure you round down on whole numbers
    return r_border - l_border + 1


def puzzle1() -> int:
    table: list[tuple[int, int]] = create_table('day6.txt')
    possibilities: int = 1
    for e in table:
        possibilities *= get_possible_times_num(e)
    return possibilities


def create_table_2(filename: str) -> tuple[int, int]:
    with open('day6.txt', 'r', encoding='utf-8') as file:
        t: str = re.search(r'\d+', file.readline().replace(' ', '')).group(0)
        d: str = re.search(r'\d+', file.readline().replace(' ', '')).group(0)
    return int(t), int(d)


def puzzle2() -> int:
    t, d = create_table_2('day6.txt')
    return get_possible_times_num((t, d))


if __name__ == '__main__':
    print(puzzle1())
    print(puzzle2())

Scala3

// math.floor(i) == i if i.isWhole, but we want i-1
def hardFloor(d: Double): Long = (math.floor(math.nextAfter(d, Double.NegativeInfinity))).toLong
def hardCeil(d: Double): Long = (math.ceil(math.nextAfter(d, Double.PositiveInfinity))).toLong

def wins(t: Long, d: Long): Long =
    val det = math.sqrt(t*t/4.0 - d)
    val high = hardFloor(t/2.0 + det)
    val low = hardCeil(t/2.0 - det)
    (low to high).size

def task1(a: List[String]): Long = 
    def readLongs(s: String) = s.split(raw"\s+").drop(1).map(_.toLong)
    a match
        case List(s"Time: $time", s"Distance: $dist") => readLongs(time).zip(readLongs(dist)).map(wins).product
        case _ => 0L

def task2(a: List[String]): Long =
    def readLong(s: String) = s.replaceAll(raw"\s+", "").toLong
    a match
        case List(s"Time: $time", s"Distance: $dist") => wins(readLong(time), readLong(dist))
        case _ => 0L

C

Brute forced it, runs in 60 ms or so. Only shortcut is quitting the loop when the distance drops below the record. I didn’t bother with the closed form solution here because a) it ran so fast and b) I was concerned about floats, rounding and off-by-one errors. Will probably implement it later!

GitHub link

Edit: implemented the closed form solution. Feels dirty copying a formula without really understanding it…

GitHub link (closed form)

Raku

I spent a lot more time than necessary optimizing the count-ways-to-beat function, but I’m happy with the result. This is my first time using the | operator to flatten a list into function arguments.

edit: unfortunately, the lemmy web page is unable to properly display the source code in a code block. It doesn’t display text enclosed in pointy brackets <>, perhaps it looks too much like HTML. View code on github.

use v6;

sub MAIN($input) {
    my $file = open $input;

    grammar Records {
        token TOP {  "\n"  "\n"* }
        token times { "Time:" \s* +%\s+ }
        token distances { "Distance:" \s* +%\s+ }
        token num { \d+ }
    }

    my $records = Records.parse($file.slurp);

    my $part-one-solution = 1;
    for $records».Int Z $records».Int -> $record {
        $part-one-solution *= count-ways-to-beat(|$record);
    }
    say "part 1: $part-one-solution";

    my $kerned-time = $records.join.Int;
    my $kerned-distance = $records.join.Int;
    my $part-two-solution = count-ways-to-beat($kerned-time, $kerned-distance);
    say "part 2: $part-two-solution";
}

sub count-ways-to-beat($time, $record-distance) {
    # time = button + go
    # distance = go * button
    # 0 = go^2 - time * go + distance
    # go = (time +/- sqrt(time**2 - 4*distance))/2

    # don't think too hard:
    # if odd t then t/2 = x.5,
    #   so sqrt(t**2-4*d)/2 = 2.3 => result = 4
    #   and sqrt(t**2-4*d)/2 = 2.5 => result = 6
    #   therefore result = 2 * (sqrt(t**2-4*d)/2 + 1/2).floor
    # even t then t/2 = x.0
    #   so sqrt(t^2-4*d)/2 = 2.x => result = 4 + 1(shared) = 5
    #   therefore result = 2 * (sqrt(t^2-4*d)/2).floor + 1
    # therefore result = 2 * ((sqrt(t**2-4*d)+t%2)/2).floor + 1 - t%2
    # Note: sqrt produces a Num, so perhaps the result could be off by 1 or 2,
    #       but it solved my AoC inputs correctly 😃.

    my $required-distance = $record-distance + 1;
    return 2 * ((sqrt($time**2 - 4*$required-distance) + $time%2)/2).floor + 1 - $time%2;
}

Factor on github (with comments and imports):

I didn’t use any math smarts.

: input>data ( -- races )
  "vocab:aoc-2023/day06/input.txt" utf8 file-lines     
  [ ": " split harvest rest [ string>number ] map ] map
  first2 zip                                           
;

: go ( press-ms total-time -- distance )
  over - *
;

: beats-record? ( press-ms race -- ? )
  [ first go ] [ last ] bi >
;

: ways-to-beat ( race -- n )
  dup first [1..b)          
  [                         
    over beats-record?      
  ] map [ ] count nip       
;

: part1 ( -- )
  input>data [ ways-to-beat ] map-product .
;

: input>big-race ( -- race )
  "vocab:aoc-2023/day06/input.txt" utf8 file-lines             
  [ ":" split1 nip " " without string>number ] map
;

: part2 ( -- )
  input>big-race ways-to-beat .
;

Not sure if it’s the most optimal, but I figured it’s probably quicker to calculate the first point when you start winning, and then reverse it to get the last point when you’ll last win. Subtracting the two to get the total number of ways to win.

Takes about 3 seconds to run on the real input

class Race:
    def __init__(self, time, distance):
        self.time = time
        self.distance = distance

    def get_win(self, start, stop, step):
        for i in range(start, stop, step):
            if (self.time - i) * i > self.distance:
                return i

    def get_winners(self):
        return (
            self.get_win(0, self.time, 1),
            self.get_win(self.time, 0, -1),
        )

race = Race(71530, 940200)
winners = race.get_winners()
print(winners[1] - winners[0] + 1)

[Language: Lean4]

This one was straightforward, especially since Lean’s Floats are 64bits. There is one interesting piece in the solution though, and that’s the function that combines two integers, which I wrote because I want to use the same parse function for both parts. This combineNumbers function is interesting, because it needs a proof of termination to make the Lean4 compiler happy. Or, in other words, the compiler needs to be told that if n is larger than 0, n/10 is a strictly smaller integer than n. That proof actually exists in Lean’s standard library, but the compiler doesn’t find it by itself. Supplying it is as easy as invoking the simp tactic with that proof, and a proof that n is larger than 0.

As with the previous days, I won’t post the full source here, just the relevant parts. The full solution is on github, including the main function of the program, that loads the input file and runs the solution.

structure Race where
  timeLimit : Nat
  recordDistance : Nat
  deriving Repr

private def parseLine (header : String) (input : String) : Except String (List Nat) := do
  if not $ input.startsWith header then
    throw s!"Unexpected line header: {header}, {input}"
  let input := input.drop header.length |> String.trim
  let numbers := input.split Char.isWhitespace
    |> List.map String.trim
    |> List.filter (not ∘ String.isEmpty)
  numbers.mapM $ Option.toExcept s!"Failed to parse input line: Not a number {input}" ∘  String.toNat?

def parse (input : String) : Except String (List Race) := do
  let lines := input.splitOn "\n"
    |> List.map String.trim
    |> List.filter (not ∘ String.isEmpty)
  let (times, distances) ← match lines with
    | [times, distances] =>
      let times ← parseLine "Time:" times
      let distances ← parseLine "Distance:" distances
      pure (times, distances)
    | _ => throw "Failed to parse: there should be exactly 2 lines of input"
  if times.length != distances.length then
    throw "Input lines need to have the same number of, well, numbers."
  let pairs := times.zip distances
  if pairs = [] then
    throw "Input does not have at least one race."
  return pairs.map $ uncurry Race.mk

-- okay, part 1 is a quadratic equation. Simple as can be
-- s = v * tMoving
-- s = tPressed * (tLimit - tPressed)
-- (tPressed - tLimit) * tPressed + s = 0
-- tPressed² - tPressed * tLimit + s = 0
-- tPressed := tLimit / 2 ± √(tLimit² / 4 - s)
-- beware: We need to _beat_ the record, so s here is the record + 1

-- Inclusive! This is the smallest number that can win, and the largest number that can win
private def Race.timeRangeToWin (input : Race) : (Nat × Nat) :=
  let tLimit  := input.timeLimit.toFloat
  let sRecord := input.recordDistance.toFloat
  let tlimitHalf := 0.5 * tLimit
  let theRoot := (tlimitHalf^2 - sRecord - 1.0).sqrt
  let lowerBound := tlimitHalf - theRoot
  let upperBound := tlimitHalf + theRoot
  let lowerBound := lowerBound.ceil.toUInt64.toNat
  let upperBound := upperBound.floor.toUInt64.toNat
  (lowerBound,upperBound)

def part1 (input : List Race) : Nat :=
  let limits := input.map Race.timeRangeToWin
  let counts := limits.map $ λ p ↦ p.snd - p.fst + 1 -- inclusive range
  counts.foldl (· * ·) 1

-- part2 is the same thing, but here we need to be careful.
-- namely, careful about the precision of Float. Which luckily is enough, as confirmed by pen&paper
-- but _barely_ enough.
-- If Lean's Float were an actual C float and not a C double, this would not work.

-- we need to concatenate the numbers again (because I don't want to make a separate parse for part2)
private def combineNumbers (left : Nat) (right : Nat) : Nat :=
  let rec countDigits := λ (s : Nat) (n : Nat) ↦
    if p : n > 0 then
      have : n > n / 10 := by simp[p, Nat.div_lt_self]
      countDigits (s+1) (n/10)
    else
      s
  let d := if right = 0 then 1 else countDigits 0 right
  left * (10^d) + right

def part2 (input : List Race) : Nat :=
  let timeLimits := input.map Race.timeLimit
  let timeLimit := timeLimits.foldl combineNumbers 0
  let records := input.map Race.recordDistance
  let record := records.foldl combineNumbers 0
  let limits := Race.timeRangeToWin $ {timeLimit := timeLimit, recordDistance := record}
  limits.snd - limits.fst + 1 -- inclusive range

open DayPart
instance : Parse ⟨6, by simp⟩ (ι := List Race) where
  parse := parse

instance : Part ⟨6, _⟩ Parts.One (ι := List Race) (ρ := Nat) where
  run := some ∘ part1

instance : Part ⟨6, _⟩ Parts.Two (ι := List Race) (ρ := Nat) where
  run := some ∘ part2

Nim

Today’s puzzle was too easy. I solved it with bruteforce in 20 minutes, but that’s boring. So here’s the optimized solution with quadratic formula.

Total runtime: 0.008 ms
Puzzle rating: Too Easy 5/10
Code: day_06/solution.nim