From 99abec12d15d62c0f16ed5c602153cceff1d2559 Mon Sep 17 00:00:00 2001
From: Felix Martin <mail@felixm.de>
Date: Sat, 6 Feb 2021 13:12:15 -0500
Subject: [PATCH] Finish k-clustering implementation

---
 src/k_clustering.rs | 52 +++++++++++++++++++--------------------------
 src/util.rs         |  1 -
 2 files changed, 22 insertions(+), 31 deletions(-)

diff --git a/src/k_clustering.rs b/src/k_clustering.rs
index 9bbf7c7..2444efd 100644
--- a/src/k_clustering.rs
+++ b/src/k_clustering.rs
@@ -1,63 +1,55 @@
-use std::collections::HashSet;
-
 #[derive(Debug, Clone, Copy)]
 pub struct Edge {
-    pub weight: i32,
+    pub weight: usize,
     pub source_id: usize,
     pub target_id: usize,
 }
 
 pub struct WeightedGraph {
     pub n_nodes: usize,
-    pub cluster: Vec<usize>,
     pub edges: Vec<Edge>,
 }
 
 pub fn k_clustering(graph: &mut WeightedGraph) -> usize {
     let k: usize = 4;
 
+    let mut node_id_to_cluster_id: Vec<usize> = (0..graph.n_nodes).collect();
     let mut cluster_count = graph.n_nodes;
     let mut clusters: Vec<Vec<usize>> = (0..graph.n_nodes).map(|x| vec![x]).collect();
-    println!("{:?}", clusters);
 
-    // Sort edges by decreasing weight
+    // Sort edges by increasing weight
     graph
         .edges
-        .sort_by(|a, b| b.weight.partial_cmp(&a.weight).unwrap());
+        .sort_by(|a, b| a.weight.partial_cmp(&b.weight).unwrap());
 
     for edge in &graph.edges {
-        let ca_id = graph.cluster[edge.source_id];
-        let cb_id = graph.cluster[edge.target_id];
+        let cluster_id_a = node_id_to_cluster_id[edge.source_id];
+        let cluster_id_b = node_id_to_cluster_id[edge.target_id];
 
-        if ca_id == cb_id {
-            continue; // Nodes are already in the same cluster
+        if cluster_id_a != cluster_id_b {
+	        let mut cluster_b = std::mem::take(&mut clusters[cluster_id_b]);
+	        for node_id in &cluster_b {
+	        	node_id_to_cluster_id[*node_id] = cluster_id_a;
+	        }
+	        clusters[cluster_id_a].append(&mut cluster_b);
+	        cluster_count -= 1;
         }
 
-        let ca: &mut Vec<usize> = &mut clusters[ca_id];
-        let cb = clusters[cb_id].clone();
-
-        for node in cb {
-            ca.push(node);
-        }
-
-        // clusters[ca].append(&mut clusters[cb]);
-
-        // clusters[]
-        // clusters[cluster_target].clear();
-        // graph.cluster[edge.target_id]= cluster_source;
-
-        cluster_count -= 1;
-
         if cluster_count <= k {
             break;
         }
     }
 
-    for cluster in &clusters {
-        if cluster.len() > 0 {
-            // println!("{:?}", cluster);
+    let mut min_spacing = usize::MAX;
+    for edge in &graph.edges {
+        let cluster_id_a = node_id_to_cluster_id[edge.source_id];
+        let cluster_id_b = node_id_to_cluster_id[edge.target_id];
+        if cluster_id_a != cluster_id_b {
+        	if edge.weight < min_spacing {
+        		min_spacing = edge.weight;
+        	}
         }
     }
 
-    graph.n_nodes
+    min_spacing
 }
diff --git a/src/util.rs b/src/util.rs
index b03f667..394e2c3 100644
--- a/src/util.rs
+++ b/src/util.rs
@@ -202,7 +202,6 @@ pub fn read_weighted_graph_clustering(
 
     let mut g = k_clustering::WeightedGraph {
         n_nodes: n_nodes,
-        cluster: (0..n_nodes).collect(),
         edges: Vec::new(),
     };