Day 23

I solved this initially in a way that took around 15 seconds to run (check every possible combination of computers to see which ones were all connected), then did some research and found the "Bron-Kerbosch algorithm" for solving this problem, and wouldn't ya know, a named algorithm solves the problem way faster than brute force. Funny that.

src/23.cs

@@ -0,0 +1,93 @@
+namespace aoc2024;
+
+internal class Day23 : Day
+{
+    private readonly Dictionary<string, HashSet<string>> connections = [];
+
+    internal override void Parse()
+    {
+        var lines = Util.Parsing.ReadAllLines($"{GetDay()}");
+        var pairs = lines.Select(line => line.Split('-'));
+        foreach (var pair in pairs)
+        {
+            if (!connections.TryAdd(pair.First(), [pair.Last()]))
+            {
+                connections[pair.First()].Add(pair.Last());
+            }
+
+            if (!connections.TryAdd(pair.Last(), [pair.First()]))
+            {
+                connections[pair.Last()].Add(pair.First());
+            }
+        }
+    }
+
+    internal override string Part1()
+    {
+        HashSet<string> seen = [];
+        var numGroups = 0;
+
+        foreach (var d1 in connections)
+        {
+            foreach (var d2Name in d1.Value)
+            {
+                foreach (var d3Name in connections[d2Name])
+                {
+                    if (!connections[d3Name].Contains(d1.Key) ||
+                        (d1.Key[0] != 't' && d2Name[0] != 't' && d3Name[0] != 't'))
+                    {
+                        continue;
+                    }
+
+                    List<string> set = [d1.Key, d2Name, d3Name];
+                    set.Sort();
+                    var setStr = string.Join(',', set);
+                    if (!seen.Add(setStr))
+                    {
+                        continue;
+                    }
+
+                    numGroups++;
+                }
+            }
+        }
+
+        return $"# cliques containing a computer starting with 't': <+white>{numGroups}";
+    }
+
+    // Bron–Kerbosch algorithm, see https://en.wikipedia.org/wiki/Bron%E2%80%93Kerbosch_algorithm
+    private void BronKerbosch(HashSet<string> r, HashSet<string> p, HashSet<string> x, List<HashSet<string>> cliques)
+    {
+        if (p.Count == 0 && x.Count == 0)
+        {
+            cliques.Add(r);
+            return;
+        }
+
+        foreach (var v in p.Except(connections[FindNonNeighbor(p, x)]))
+        {
+            BronKerbosch([..r, v], [..p.Intersect(connections[v])], [..x.Intersect(connections[v])], cliques);
+
+            p.Remove(v);
+            x.Add(v);
+        }
+    }
+
+    private string FindNonNeighbor(HashSet<string> p, HashSet<string> x)
+        => p.Union(x).OrderByDescending(v => connections[v].Count).First();
+
+    private List<HashSet<string>> GetAllCliques()
+    {
+        var cliques = new List<HashSet<string>>();
+        BronKerbosch([], [..connections.Keys], [], cliques);
+
+        return cliques;
+    }
+
+    internal override string Part2()
+    {
+        var cliques = GetAllCliques();
+        var password = string.Join(",", cliques.MaxBy(c => c.Count)!.Order());
+        return $"LAN party password: <+white>{password}";
+    }
+}