Rustworkx acceleration#381

Draft

InnovativeInventor wants to merge 26 commits intomainfrom

retworkx

Commits on Apr 25, 2023

Add __slots__ and replace PopulatedGraph.degree with direct dictionary lookup

InnovativeInventor
authored and
pjrule
committed
Switch to using more performant/preferred lookup call

InnovativeInventor
authored and
pjrule
committed
Add retworkx converter

InnovativeInventor
authored and
pjrule
committed
Use graph.node_indicies in PopulatedGraph

InnovativeInventor
authored and
pjrule
committed
Add keep_attributes flag (will be released in retworkx)

InnovativeInventor
authored and
pjrule
committed
Naive bipartition_tree retworkx impl

InnovativeInventor
authored and
pjrule
committed
Improve subgraph instantiation times

InnovativeInventor
authored and
pjrule
committed
More perf improvements

InnovativeInventor
authored and
pjrule
committed
Stop using recursive_tree_part

InnovativeInventor
authored and
pjrule
committed
Revert "Stop using recursive_tree_part"

InnovativeInventor
authored and
pjrule
committed
Resolve memory leak issue with caching

InnovativeInventor
authored and
pjrule
committed
Prevent unnecessary subgraphing

InnovativeInventor
authored and
pjrule
committed
Maybe faster record perf (todo: benchmark carefully)

InnovativeInventor
authored and
pjrule
committed
Re-enable lookup cache

InnovativeInventor
authored and
pjrule
committed
Use dict.update operator to support older versions of Python

InnovativeInventor
authored and
pjrule
committed
Fix misspelled node_indicies [sic]

InnovativeInventor
authored and
pjrule
committed
Fix linter issues

InnovativeInventor
authored and
pjrule
committed
Add retworkx as setup.py dep

InnovativeInventor
authored and
pjrule
committed
Reduce overhead of to_series call

InnovativeInventor
authored and
pjrule
committed
Switch to __getattr__ call to reduce overhead of attr fetches

InnovativeInventor
authored and
pjrule
committed
Force self.mapping to maintain sorted order invariant for faster serialization

InnovativeInventor
authored and
pjrule
committed
Update retworkx call to use renamed function

InnovativeInventor
authored and
pjrule
committed
Rename bipartition_graph to bipartition_graph_mst

InnovativeInventor
authored and
pjrule
committed
Make rustworkx (née retworkx) an optional dependency
pjrule
committed
Revert changes to Partition class
pjrule
committed

Commits on Apr 29, 2023

WIP: Use gerrychain.rs for acceleration
pjrule
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Comments

Rustworkx acceleration#381

Rustworkx acceleration#381
InnovativeInventor wants to merge 26 commits intomainfrom
retworkx

Commits on Apr 25, 2023

Add slots and replace PopulatedGraph.degree with direct dictionary lookup

Switch to using more performant/preferred lookup call

Add retworkx converter

Use graph.node_indicies in PopulatedGraph

Add keep_attributes flag (will be released in retworkx)

Naive bipartition_tree retworkx impl

Improve subgraph instantiation times

More perf improvements

Stop using recursive_tree_part

Revert "Stop using recursive_tree_part"

Resolve memory leak issue with caching

Prevent unnecessary subgraphing

Maybe faster record perf (todo: benchmark carefully)

Re-enable lookup cache

Use dict.update operator to support older versions of Python

Fix misspelled node_indicies [sic]

Fix linter issues

Add retworkx as setup.py dep

Reduce overhead of to_series call

Switch to getattr call to reduce overhead of attr fetches

Force self.mapping to maintain sorted order invariant for faster serialization

Update retworkx call to use renamed function

Rename bipartition_graph to bipartition_graph_mst

Make rustworkx (née retworkx) an optional dependency

Revert changes to Partition class

Commits on Apr 29, 2023

WIP: Use gerrychain.rs for acceleration