Decrease temporal shortest path memory footprint

pathpy / pathpyG

GPU-accelerated Next-Generation Network Analytics and Graph Learning for Time Series Data on Complex Networks.

https://www.pathpy.net

GNU Affero General Public License v3.0

33 stars 3 forks source link

Decrease temporal shortest path memory footprint #156

Closed M-Lampert closed 6 months ago

M-Lampert commented 7 months ago

By saving the node sequences for each higher-order node, we get a lot of duplicate paths. To decrease the memory footprint, we remove all duplicates and only save all unique paths and use an additional index to be able to build the node sequences for each nodes when it is necessary.

IngoScholtes commented 6 months ago

I guess this change is not needed anymore given the new memory-efficient functions to calculate time-respecting shortest paths and temporal betweenness centrality. Closing this for now.

M-Lampert commented 6 months ago

Yes, the specific changes here won't be necessary anymore but the general idea behind it would also work for the general lift order in MultiOrderModel. It is not yet implemented there but we should keep this potential improvement in mind.