Open francois-drielsma opened 1 year ago
I would recommend testing the centroid method to estimating the track direction. Here's the lartpc_mrleco3d implementation:
https://github.com/francois-drielsma/lartpc_mlreco3d/blob/d720174b9fdd23e616d3f15c6772a1a720b2b4e8/mlreco/utils/gnn/cluster.py#L564-L628
Here is the local np.mean implementation:
np.mean
https://github.com/francois-drielsma/lartpc_mlreco3d/blob/d720174b9fdd23e616d3f15c6772a1a720b2b4e8/mlreco/utils/numba_local.py#L69-L95
and the local scipy.spatial.distance.cdist implementation:
scipy.spatial.distance.cdist
https://github.com/francois-drielsma/lartpc_mlreco3d/blob/d720174b9fdd23e616d3f15c6772a1a720b2b4e8/mlreco/utils/numba_local.py#L318-L340
Note to self: update README with this information when/if implemented
I would recommend testing the centroid method to estimating the track direction. Here's the lartpc_mrleco3d implementation:
https://github.com/francois-drielsma/lartpc_mlreco3d/blob/d720174b9fdd23e616d3f15c6772a1a720b2b4e8/mlreco/utils/gnn/cluster.py#L564-L628