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
Currently, there are multiple for loops required in order to identify neighbours of any given agent. This is because the network ids and the agent unique_ids do not correlate with each other and thus requires an extensive search through the whole list to identify the correct agent associated with a neighbour node.
This could be fixed using (we need to try!):
Networkx: set_node_attributes function which might present a 1-1 relation between agent and network nodes
Use dgl to redefine the network so that this is no longer necessary.
Currently we saw a 10% speedup by adding a break statement in the for loops to remove unnecessary iterations. We foresee another 10% speedup by changing the above.*
The change was noted for 500 agents simulated for 125 time-steps. Time went from 615s down to 540s
The text was updated successfully, but these errors were encountered:
Currently, there are multiple for loops required in order to identify neighbours of any given agent. This is because the network ids and the agent unique_ids do not correlate with each other and thus requires an extensive search through the whole list to identify the correct agent associated with a neighbour node.
This could be fixed using (we need to try!):
Currently we saw a 10% speedup by adding a break statement in the for loops to remove unnecessary iterations. We foresee another 10% speedup by changing the above.*
The text was updated successfully, but these errors were encountered: