Open disperse opened 6 years ago
However, according to their homepage: it may be less scalable to very large graphs.
Since we can track when changes are being made to the graph it might be in our best interest to control the simulation manually using simulation.stop. This way we can reduce CPU usage when we know nothing is changing in the graph.
If we limit graph access to custom methods we could optimize this further by only running the simulation when a real change happens to the graph. We could only run the simulation if it's nodes or edges being added or removed and not run the simulation if the colors or other properties are changing.
Furthermore, we could move the graph simulation to a web worker and avoid blocking the UI while the CPU intensive calculations are happening.
Here's a list I found of graph visualization libraries.
Here's a notable example of a graphing library that has very high performance
http://marvl.infotech.monash.edu/webcola/