You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
Niko Abeler 456e027824 version bump 2 months ago
.github/workflows adjust test action. 4 2 months ago
.vscode formatting 3 months ago
src initial position 2 months ago
tests initial position 2 months ago
.gitignore more tests + allowing generators as input 3 months ago
Cargo.lock version bump 2 months ago
Cargo.toml version bump 2 months ago install and dev 3 months ago readme update 2 months ago
pyproject.toml adjust test action. 2 2 months ago
requirements.txt more tests + allowing generators as input 3 months ago

Graph Force

A python/rust library for embedding graphs in 2D space, using force-directed layouts.


pip install graph_force


The first parameter defines the number of nodes in graph. The second parameter is an iterable of edges, where each edge is a tuple of two integers representing the nodes it connects. Node ids start at 0.

import graph_force

edges = [(0, 1), (1, 2), (2, 3), (3, 0)]
pos = graph_force.layout_from_edge_list(4, edges)

Example with networkx

This library does not have a function to consume a networkx graph directly, but it is easy to convert it to an edge list.

import networkx as nx
import graph_force

G = nx.grid_2d_graph(10, 10)
# we have to map the names to integers
# as graph_force only supports integers as node ids at the moment
edges = []
mapping = {n: i for i, n in enumerate(G.nodes)}
i = 0
for edge in G.edges:
    edges.append((mapping[edge[0]], mapping[edge[1]]))

pos = graph_force.layout_from_edge_list(len(G.nodes), edges, iter=1000)
nx.draw(G, {n: pos[i] for n, i in mapping.items()}, node_size=2, width=0.1)

Example with edge file

This methods can be used with large graphs, where the edge list does not fit into memory.

Format of the file:

  • Little endian
  • 4 bytes: number of nodes(int)
  • 12 bytes: nodeA(int), nodeB(int), weight(float)
import graph_force
import struct

with open("edges.bin", "rb") as f:
    n = 10
    f.write(struct.pack("i", n))
    for x in range(n-1):
        f.write(struct.pack("iif", x, x+1, 1))

pos = graph_force.layout_from_edge_file("edges.bin", iter=50)


iter, threads and model, initial_pos are optional parameters, supported by layout_from_edge_list and layout_from_edge_file.

pos = graph_force.layout_from_edge_list(
    iter=500,  # number of iterations, default 500
    threads=0,  # number of threads, default 0 (all available)
    model="spring_model", # model to use, default "spring_model", other option is "networkx_model"
    initial_pos=[(0.4, 0.7), (0.7, 0.2), ...], # initial positions, default None (random)

Available models