If I’m reading your code correctly, I took essentially the same approach but used a plain Map instead of the balanced tree structure you’ve used. I was also tracking the “seen” points in a list rather than a MapSet. My solution could not solve part 1 for the input without melting my CPU, though it works for the smaller example. I’m going to try stealing your use of :gb_sets and MapSet and see if that fixes my issue. Thanks for sharing this.

I’ve just implemented the whole dijkstra part on my own.
Especiallty for part two not fast at all (took about 20min for the result). But I am happy to have finally solved it.

Mine is very similar to others. I’ve never implemented Dijkstra’s before, so that was kind of fun.
My original solution was just based on Wikipedia’s psuedocode, so part two was horribly slow/naive. I was finding the minimum value from a map of point -> risk. So shoutout to @bjorng 's solution which introduced me to :gb_sets. Refactored my solution leading to a substantial time improvement (O(min) → <4sec on apple m1)

Here is my solution:

Runtime for both parts and the examples is less than 4 seconds on my computer.

ah, good old pathfinding.

I did in anticipation of this implement a naive priority queue a few days ago, so it came in very handy today.

both parts + tests run in under half a second