--- Day 18: RAM Run ---
Finally decided to do one more. Day 18 is also running in a grid, but I figured out a different approach. No arrays, no indexing, no testing for out of bounds indexes!
I implemented the RAM matrix as a hashtable. Of-course, initial setup requires some indexing to set-up the hashtable, but there's no indexing after that.
Hashtable entries are for example '(0 . 0) - (that is a pair, or improper list in Scheme) as a key, and '((1 . 0) (0 . 1)) as a value - (neighbors of the '(0 . 0) key).
This of-course, slows me down (compared to arrays), but I have absolutely no indexing after that!
The path search is a very simple graph traversal (probably would qualify as DFS)The slowdown here is member function that is filtering out neighbors that have been visited in the current generation of recursion. Member is O(n) function.I'm cheating here ![:D]( "Very Happy")
. This is on banana, but a second run, when the memoization hashtable has already been initialized. The real timing is 24.9s for the first run.
My i5 (which is a 5 year old computer) does the first run in 1.9s, so I guess it is a good enough solution.
Finally decided to do one more. Day 18 is also running in a grid, but I figured out a different approach. No arrays, no indexing, no testing for out of bounds indexes!
I implemented the RAM matrix as a hashtable. Of-course, initial setup requires some indexing to set-up the hashtable, but there's no indexing after that.
Hashtable entries are for example '(0 . 0) - (that is a pair, or improper list in Scheme) as a key, and '((1 . 0) (0 . 1)) as a value - (neighbors of the '(0 . 0) key).
This of-course, slows me down (compared to arrays), but I have absolutely no indexing after that!
The path search is a very simple graph traversal (probably would qualify as DFS)
Code:
(define (run seen from to) ;; DFS - I think??? (cond ((equal? from to) 0) (else (let* ((nbrs (hashtable-ref ram from #f)) (not-seen (filter (lambda (n) (not (member n seen))) nbrs)) (new-seen (append not-seen seen))) (cond ((null? not-seen) #f) (else(flatten (map (lambda (nbr) (inc1 (run new-seen nbr to))) not-seen))))))))Code:
hrvoje@BPI-F3:~/Projects/advent-of-code/2024/day-18/Chez$ schemeChez Scheme Version 10.1.0Copyright 1984-2024 Cisco Systems, Inc.> (load "day18-part2.ss")> (time (lambda () (day18 "input.txt")))360#<time-duration 0.053240108>
. This is on banana, but a second run, when the memoization hashtable has already been initialized. The real timing is 24.9s for the first run.My i5 (which is a 5 year old computer) does the first run in 1.9s, so I guess it is a good enough solution.
Statistics: Posted by hrvoje064 — Fri Feb 07, 2025 4:47 pm