mlreco.utils.groups module

mlreco.utils.groups.get_group_types(particle_v, meta, point_type='3d')[source]

Gets particle classes for voxel groups

mlreco.utils.groups.filter_duplicate_voxels(data, usebatch=True)[source]

return array that will filter out duplicate voxels Only first instance of voxel will appear Assume data[:4] = [x,y,z,batchid] Assumes data is lexicographically sorted in x,y,z,batch order

mlreco.utils.groups.filter_duplicate_voxels_ref(data, reference, meta, usebatch=True, precedence=[1, 2, 0, 3, 4])[source]

return array that will filter out duplicate voxels Sort with respect to a reference and following the specified precedence order Assume data[:4] = [x,y,z,batchid] Assumes data is lexicographically sorted in x,y,z,batch order

mlreco.utils.groups.filter_nonimg_voxels(data_grp, data_img, usebatch=True)[source]

return array that will filter out voxels in data_grp that are not in data_img ASSUME: data_grp and data_img are lexicographically sorted in x,y,z,batch order ASSUME: all points in data_img are also in data_grp ASSUME: all voxels in data are unique

mlreco.utils.groups.filter_group_data(data_grp, data_img)[source]

return return array that will permute and filter out voxels so that data_grp and data_img have same voxel locations 1) lexicographically sort group data (images are lexicographically sorted) 2) remove voxels from group data that are not in image 3) choose only one group per voxel (by lexicographic order) WARNING: (3) is certainly not a canonical choice

mlreco.utils.groups.process_group_data(data_grp, data_img)[source]

return processed group data 1) lexicographically sort group data (images are lexicographically sorted) 2) remove voxels from group data that are not in image 3) choose only one group per voxel (by lexicographic order) WARNING: (3) is certainly not a canonical choice

mlreco.utils.groups.get_interaction_id(particle_v, num_ancestor_loop=1)[source]

A function to sort out interaction ids. Note that this assumes cluster_id==particle_id. Inputs:

  • particle_v (array) : larcv::EventParticle.as_vector()

  • num_ancestor_loop (int): number of ancestor loops (default 1)

Outputs:

  • interaction_ids: a numpy array with the shape (n,)

mlreco.utils.groups.get_nu_id(cluster_event, particle_v, interaction_ids, particle_mpv=None)[source]

A function to sorts interactions into nu or not nu (0 for cosmic, 1 for nu). CAVEAT: Dirty way to sort out nu_ids

Assuming only one nu interaction is generated and first group/cluster belongs to such interaction

Inputs:

  • cluster_event (larcv::EventClusterVoxel3D): (N) Array of cluster tensors

  • particle_v vector: larcv::EventParticle.as_vector()

  • interaction_id: a numpy array with shape (n, 1) where 1 is interaction id

  • (optional) particle_mpv: vector of particles from mpv generator, used to work around

the lack of proper interaction id for the time being.

Outputs:

  • nu_id: a numpy array with the shape (n,1)

mlreco.utils.groups.get_particle_id(particles_v, nu_ids, include_mpr=False, include_secondary=False)[source]

Function that gives one of five labels to particles of particle species predictions. This function ensures: - Particles that do not originate from an MPV are labeled -1,

unless the include_mpr flag is set to true

  • Secondary particles (includes Michel/delta and neutron activity) are labeled -1, unless the include_secondary flag is true

  • All shower daughters are labeled the same as their primary. This makes sense as otherwise an electron primary gets overruled by its many photon daughters (voxel-wise majority vote). This can lead to problems as, if an electron daughter is not clustered with the primary, it is labeled electron, which is counter-intuitive. This is handled downstream with the high_purity flag.

  • Particles that are not in the list target are labeled -1

Inputs:

  • particles_v (array of larcv::Particle) : (N) LArCV Particle objects

  • nu_ids: a numpy array with shape (n, 1) where 1 is neutrino id (0 if not an MPV)

  • include_mpr: include MPR (cosmic-like) particles to PID target

  • include_secondary: include secondary particles into the PID target

Outputs:

  • array: (N) list of group ids

mlreco.utils.groups.get_shower_primary_id(cluster_event, particles_v)[source]

Function that assigns valid primary tags to shower fragments. This could be handled somewhere else (e.g. SUPERA)

Inputs:

  • cluster_event (larcv::EventClusterVoxel3D): (N) Array of cluster tensors

  • particles_v (array of larcv::Particle) : (N) LArCV Particle objects

Outputs:

  • array: (N) list of group ids

mlreco.utils.groups.get_group_primary_id(particles_v, nu_ids=None, include_mpr=True)[source]

Function that assigns valid primary tags to particle groups. This could be handled somewhere else (e.g. SUPERA)

Inputs:

  • particles_v (array of larcv::Particle) : (N) LArCV Particle objects

  • nu_ids: a numpy array with shape (n, 1) where 1 is neutrino id (0 if not an MPV)

  • include_mpr: include MPR (cosmic-like) particles to primary target

Outputs:

  • array: (N) list of group ids

mlreco.utils.deghosting module mlreco.utils.metrics module