bjorng
Most Liked
igorb
Solution using dynamic programming (prefix sums): https://github.com/ibarakaiev/advent-of-code-2023/blob/main/lib/advent_of_code/day_11.ex. There are some possible optimizations like recording coordinates at the same time as the vertical prefix sum, but it would take away from readability 
lud
To expand I sort the empty Y coords descending, and for each one I bump each XY in the galaxy list. Same for X.
I have an idea to do it all at once but I’m not chasing milliseconds today 
midouest
This one seemed like a great opportunity to learn a bit of Nx. I took the naive approach for part 1 and duplicated all of the rows and columns. Obviously that didn’t work out in part 2 
Part 1
sky =
for line <- String.split(input, "\n", trim: true) do
for char <- String.graphemes(line) do
if char == "#", do: 1, else: 0
end
end
sky_tensor = Nx.tensor(sky, type: :u8, names: [:y, :x])
empty_ys = Nx.any(sky_tensor, axes: [:x]) |> Nx.logical_not() |> Nx.to_list()
empty_xs = Nx.any(sky_tensor, axes: [:y]) |> Nx.logical_not() |> Nx.to_list()
pad_ys =
empty_ys
|> Enum.with_index()
|> Enum.flat_map(fn {empty, index} -> List.duplicate(index, empty + 1) end)
|> Nx.tensor(type: :u8)
pad_xs =
empty_xs
|> Enum.with_index()
|> Enum.flat_map(fn {empty, index} -> List.duplicate(index, empty + 1) end)
|> Nx.tensor(type: :u8)
big_sky =
sky_tensor
|> Nx.take(pad_ys, axis: :y)
|> Nx.take(pad_xs, axis: :x)
galaxies =
for {line, y} <- big_sky |> Nx.to_list() |> Enum.with_index(),
{galaxy, x} <- Enum.with_index(line),
reduce: [] do
acc -> if galaxy == 1, do: [{y, x} | acc], else: acc
end
pairs =
for {{y1, x1}, i} <- Enum.with_index(galaxies),
{y2, x2} <- Enum.drop(galaxies, i + 1) do
[[y1, x1], [y2, x2]]
end
|> Nx.tensor(type: :s64)
Nx.abs(Nx.subtract(pairs[[.., 0, 0]], pairs[[.., 1, 0]]))
|> Nx.add(Nx.abs(Nx.subtract(pairs[[.., 0, 1]], pairs[[.., 1, 1]])))
|> Nx.sum()
Part 2
galaxies =
for {line, y} <- sky |> Enum.with_index(),
{galaxy, x} <- Enum.with_index(line),
reduce: [] do
acc -> if galaxy == 1, do: [{y, x} | acc], else: acc
end
y_indexes =
empty_ys
|> Enum.with_index()
|> Enum.flat_map(fn
{1, i} -> [i]
_ -> []
end)
x_indexes =
empty_xs
|> Enum.with_index()
|> Enum.flat_map(fn
{1, i} -> [i]
_ -> []
end)
galaxies =
for {y, x} <- galaxies do
sy = Enum.count(y_indexes, &(&1 < y))
sx = Enum.count(x_indexes, &(&1 < x))
{y + 999_999 * sy, x + 999_999 * sx}
end
pairs =
for {{y1, x1}, i} <- Enum.with_index(galaxies),
{y2, x2} <- Enum.drop(galaxies, i + 1) do
[[y1, x1], [y2, x2]]
end
|> Nx.tensor(type: :s64)
Nx.abs(Nx.subtract(pairs[[.., 0, 0]], pairs[[.., 1, 0]]))
|> Nx.add(Nx.abs(Nx.subtract(pairs[[.., 0, 1]], pairs[[.., 1, 1]])))
|> Nx.sum()
woojiahao
Added quite a few new helper functions to speed up writing AOC solutions, notably:
load_day_as_gridto automatically parse input as a map of coordinates to pointsdistinct_pairsto generate all distinct pairs of elements within a listlist_set_differenceto calculate the set difference between 2 lists (reducing the need forMapSet.newcalls)
Aetherus
I only expanded the universe once for each part of the puzzle. Dividing by two is because for each pair of the galaxies a and b, I calculated the distance between a and b and the distance between b and a. I know I could calculate the distances only once. I just didn’t bother to do that optimization.
Popular in Challenges
Other popular topics
Latest on Elixir Forum
Sponsor Spotlight
Build Elixir applications with speed and confidence.
Categories:
Sub Categories:
Forums
Our Sponsors
Build Elixir applications with speed and confidence.
Supporting innovation across the BEAM ecosystem.
We build reliable cloud platforms for business-critical systems.
Catch errors, track performance, monitor hosts and more.
Practical resources that improve the lives of professional developers.
Develop your skills with books, videos, and courses.
Producing high quality Elixir screencasts since 2017.
Courses that'll move you from confusion to "Aha, now I get it!"