You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Why does it show that Dijkstra is faster than Barrier after I run the demo.py file? I also come in to test like the Quick Start section, randomly generated a map with 10,000 nodes and 40,000 edges, and the test results showed that Dijkstra is still faster. The test.py file contains my testing process, where did I go wrong?
The algorithm is designed for very large graphs (100k+ nodes). On your 10,000-node test, Dijkstra is faster because it has lower overhead.
I just pushed fixes that disables the advanced features on smaller graphs and matches Dijkstra's performance. The theoretical O(m log^(2/3) n) advantage only shows up on massive graphs where the asymptotic improvement outweighs the implementation overhead.
Activity
Fix performance issues and add missing LICENSE file
arman0z commentedon Nov 24, 2025
This is actually the expected behavior.
The algorithm is designed for very large graphs (100k+ nodes). On your 10,000-node test, Dijkstra is faster because it has lower overhead.
I just pushed fixes that disables the advanced features on smaller graphs and matches Dijkstra's performance. The theoretical O(m log^(2/3) n) advantage only shows up on massive graphs where the asymptotic improvement outweighs the implementation overhead.