Performance Improvements

[erikpohl53]

Feeders with a large number of buildings, nodes, and/or transformers take a long time to build the primary. The following areas could be improved to reduce run time. The times listed below correspond to building a feeder with approximately 130 transformers, 5000 buildings, and 1800 nodes.

1. Get substation node (45 seconds): This calls get_nearest_points_in_the_network twice, which iterates through all nodes in the sliced network. This would be improved by using a bounding box around the substation location to first filter a database of network nodes. Further improvements could be made to get_nearest_points_in_the_network by using a^2+b^2 as a relative distance instead of geopy's geodesic distance in get_distance function.
2. Mapping for transformer nodes (160 seconds): iterating through dfs_tree.edges() takes the majority of this time and returns a warning: "<class 'networkx.utils.decorators.argmap'> compilation 28:4: FutureWarning: laplacian_matrix will return a scipy.sparse array instead of a matrix in Networkx 3.0." See below for profiling of for loop "for edge in dfs_tree.edges():". Indexing lists takes a lot of time.

158708457 function calls (158474888 primitive calls) in 197.829 seconds

Ordered by: internal time

ncalls tottime percall cumtime percall filename:lineno(function)
6426494 81.465 0.000 81.465 0.000 {method 'index' of 'list' objects}
1274 13.199 0.010 19.832 0.016 c:\Users\epohl.conda\envs_shift\lib\site-packages\networkx\classes\graph.py:530(add_nodes_from)
1274 8.278 0.006 44.120 0.035 c:\Users\epohl.conda\envs_shift\lib\site-packages\networkx\classes\graph.py:894(add_edges_from)
14638188 7.641 0.000 19.012 0.000 c:\Users\epohl.conda\envs_shift\lib\site-packages\networkx\classes\coreviews.py:282()
9748650 7.524 0.000 33.316 0.000 c:\Users\epohl.conda\envs_shift\lib_collections_abc.py:904(iter)
9807252 7.327 0.000 13.787 0.000 c:\Users\epohl.conda\envs_shift\lib\site-packages\networkx\classes\coreviews.py:336(new_node_ok)
3255071 5.618 0.000 5.621 0.000 c:\Users\epohl.conda\envs_shift\lib\site-packages\networkx\classes\coreviews.py:275(iter)
24069757 5.187 0.000 5.187 0.000 c:\Users\epohl.conda\envs_shift\lib\site-packages\networkx\classes\filters.py:54(call)
3455498 5.019 0.000 7.001 0.000 c:\Users\epohl.conda\envs_shift\lib\site-packages\networkx\classes\coreviews.py:333(getitem)
3247426 4.363 0.000 33.753 0.000 c:\Users\epohl.conda\envs_shift\lib\site-packages\networkx\classes\graph.py:1541()
9741004 4.236 0.000 4.236 0.000 c:\Users\epohl.conda\envs_shift\lib\site-packages\networkx\classes\filters.py:65()