From a503ba80eda54f82afd27e706e9defe2e94e3562 Mon Sep 17 00:00:00 2001
From: Niko Abeler <niko@rerere.org>
Date: Tue, 22 Nov 2022 22:26:09 +0100
Subject: [PATCH] networkx spring_layout copy

---
 src/networkx_model.rs | 132 ++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 132 insertions(+)
 create mode 100644 src/networkx_model.rs

diff --git a/src/networkx_model.rs b/src/networkx_model.rs
new file mode 100644
index 0000000..ff5ad58
--- /dev/null
+++ b/src/networkx_model.rs
@@ -0,0 +1,132 @@
+use crate::graph::{new_edge_matrix, EdgeMatrix, Node, NodeVector};
+use crate::model::ForceModel;
+
+/**
+ * Translation of the NetworkX spring_layout function.
+ */
+pub struct NetworkXModel {
+    edges: EdgeMatrix,
+    size: usize,
+    k: f32,  // optimal distance
+    t: f32,  // temperature
+    dt: f32, // temperature decrease
+}
+
+impl NetworkXModel {
+    pub fn new() -> Self {
+        NetworkXModel {
+            edges: new_edge_matrix(0),
+            size: 0,
+            k: 1.0,
+            t: 1.0,
+            dt: 0.0,
+        }
+    }
+}
+
+impl ForceModel for NetworkXModel {
+    fn init(&mut self, edges: EdgeMatrix, size: usize, iterations: usize) {
+        self.k = (1.0 / size as f32).sqrt();
+        self.t = 0.1;
+        self.dt = self.t / ((iterations + 1) as f32);
+        self.t += self.dt; // prepare is called before the first step
+        self.edges = edges;
+        self.size = size;
+    }
+
+    fn prepare(&mut self, _nodes: &NodeVector) {
+        self.t -= self.dt; // decrease temperature
+    }
+
+    fn step(&self, nodes: &NodeVector, i_node: usize) -> Node {
+        let node = nodes[i_node].read().unwrap();
+        let edges = self.edges.read().unwrap();
+
+        let node_x = node.x;
+        let node_y = node.y;
+
+        let mut displacement_x = 0.0;
+        let mut displacement_y = 0.0;
+
+        for o in 0..self.size {
+            if o == i_node {
+                continue;
+            }
+
+            let edge = edges[i_node][o].weight;
+
+            let o_x: f32;
+            let o_y: f32;
+            {
+                let other = nodes[o].read().unwrap();
+                o_x = other.x;
+                o_y = other.y;
+            }
+            // difference between node and other
+            let delta_x = node_x - o_x;
+            let delta_y = node_y - o_y;
+            // distance between node and other
+            let dist = (delta_x * delta_x + delta_y * delta_y).sqrt();
+            // enforce minimum distance of 0.01
+            let dist = if dist < 0.01 { 0.01 } else { dist };
+
+            // displacement "force"
+            displacement_x += delta_x * (self.k.powi(2) / dist.powi(2) - edge * dist / self.k);
+            displacement_y += delta_y * (self.k.powi(2) / dist.powi(2) - edge * dist / self.k);
+        }
+        // update positions
+        let length = (displacement_x * displacement_x + displacement_y * displacement_y).sqrt();
+        let length = if length < 0.01 { 0.01 } else { length };
+        let delta_pos_x = displacement_x * self.t / length;
+        let delta_pos_y = displacement_y * self.t / length;
+
+        Node {
+            x: node_x + delta_pos_x,
+            y: node_y + delta_pos_y,
+        }
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use super::*;
+    use crate::graph::{new_edge_matrix, new_node_vector};
+
+    #[test]
+    fn test_simple_spring_model_attraction() {
+        let mut model = NetworkXModel::new();
+        let edges = new_edge_matrix(2);
+        edges.write().unwrap()[0][1].weight = 1.0;
+        edges.write().unwrap()[1][0].weight = 1.0;
+        model.init(edges, 2, 1);
+
+        let nodes = new_node_vector(2);
+        nodes[0].write().unwrap().x = 0.0;
+        nodes[0].write().unwrap().y = 0.0;
+        nodes[1].write().unwrap().x = 1.0;
+        nodes[1].write().unwrap().y = 1.0;
+
+        model.prepare(&nodes);
+        let node = model.step(&nodes, 0);
+        assert!(node.x > 0.0);
+        assert!(node.y > 0.0);
+    }
+
+    #[test]
+    fn test_simple_spring_model_repulsion() {
+        let mut model = NetworkXModel::new();
+        let edges = new_edge_matrix(2);
+        model.init(edges, 2, 1);
+
+        let nodes = new_node_vector(2);
+        nodes[0].write().unwrap().x = 0.0;
+        nodes[0].write().unwrap().y = 0.0;
+        nodes[1].write().unwrap().x = 1.0;
+        nodes[1].write().unwrap().y = 1.0;
+
+        model.prepare(&nodes);
+        let node = model.step(&nodes, 0);
+        assert!(node.x < 0.0);
+        assert!(node.y < 0.0);
+    }
+}