Moved code to this repo from OrcaNet

Readme.md

+## Generating DL images based on KM3NeT-ORCA neutrino simulation data 
+
+This directory contains code that produces 2D/3D/4D histograms ('images') for CNNs based on raw MC h5 files.
+
+The main code for generating the images is located in 'h5_data_to_h5_input.py'. <br>
+If the simulated .h5 files are not calibrated yet, you need to specify the directory of a .detx file in 'h5_data_to_h5_input.py'.
+
+Currently, a bin size of 11x13x18x50 (x/y/z/t) is used for the final ORCA detector layout. 

Results/4dTo2d/h5/xt/.gitkeep

+ 

Results/4dTo2d/h5/xy/.gitkeep

+ 

Results/4dTo2d/h5/xz/.gitkeep

+ 

Results/4dTo2d/h5/yt/.gitkeep

+ 

Results/4dTo2d/h5/yz/.gitkeep

+ 

Results/4dTo2d/h5/zt/.gitkeep

+ 

Results/4dTo3d/h5/rzt/.gitkeep

+ 

Results/4dTo3d/h5/xyt/.gitkeep

+ 

Results/4dTo3d/h5/xyz/.gitkeep

+ 

Results/4dTo3d/h5/xzt/.gitkeep

+ 

Results/4dTo3d/h5/yzt/.gitkeep

+ 

Results/4dTo4d/h5/xyzt/.gitkeep

+ 

file_to_hits.py

+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+"""This utility code contains functions that read the raw MC .h5 files"""
+
+import os
+import pandas as pd
+import numpy as np
+import km3pipe as kp
+#from memory_profiler import profile
+#import line_profiler # call with kernprof -l -v file.py args
+
+
+def get_primary_track_index(event_blob):
+    """
+    Gets the index of the primary (neutrino) track.
+    Uses bjorkeny in order to get the primary track, since bjorkeny!=0 for the initial interacting neutrino.
+    :param kp.io.HDF5Pump.blob event_blob: HDF5Pump event blob.
+    :return: int primary index: Index of the primary track (=neutrino) in the 'McTracks' branch.
+    """
+    bjorken_y_array = event_blob['McTracks'].bjorkeny
+    primary_index = np.where(bjorken_y_array != 0.0)[0][0]
+    return primary_index
+
+
+def get_time_residual_nu_interaction_mean_triggered_hits(time_interaction, hits_time, triggered):
+    """
+
+    :param time_interaction:
+    :param hits_time:
+    :param triggered:
+    :return:
+    """
+    hits_time_triggered = hits_time[triggered == 1]
+    t_mean_triggered = np.mean(hits_time_triggered, dtype=np.float64)
+    time_residual_vertex = t_mean_triggered - time_interaction
+
+    print t_mean_triggered
+    print time_interaction
+
+    return time_residual_vertex
+
+
+def get_event_data(event_blob, geo, do_mc_hits, use_calibrated_file, data_cuts, do4d):
+    """
+    Reads a km3pipe blob which contains the information for one event.
+    Returns a hit array and a track array that contains all relevant information of the event.
+    :param kp.io.HDF5Pump.blob event_blob: Event blob of the HDF5Pump which contains all information for one event.
+    :param kp.Geometry geo: km3pipe Geometry instance that contains the geometry information of the detector.
+                            Only used if the event_blob is from a non-calibrated file!
+    :param bool do_mc_hits: tells the function of the hits (mc_hits + BG) or the mc_hits only should be parsed.
+                            In the case of mc_hits, the dom_id needs to be calculated thanks to the jpp output.
+    :param bool use_calibrated_file: specifies if a calibrated file is used as an input for the event_blob.
+                                     If False, the hits of the event_blob are calibrated based on the geo parameter.
+    :param dict data_cuts: specifies if cuts should be applied. Contains the keys 'triggered' and 'energy_lower_limit'.
+    :param (bool, str) do4d: Tuple that declares if 4D histograms should be created [0] and if yes, what should be used as the 4th dim after xyz.
+                             In the case of 'channel_id', this information needs to be included in the event_hits as well.
+    :return: ndarray(ndim=2) event_hits: 2D array containing the hit information of the event [pos_xyz time (channel_id)].
+    :return: ndarray(ndim=1) event_track: 1D array containing important MC information of the event.
+                                          [event_id, particle_type, energy, isCC, bjorkeny, dir_x/y/z, time]
+    """
+    p = get_primary_track_index(event_blob)
+
+    # parse tracks [event_id, particle_type, energy, isCC, bjorkeny, dir_x/y/z, time]
+    event_id = event_blob['EventInfo'].event_id[0]
+    run_id = event_blob['EventInfo'].run_id[0]
+    time_interaction = event_blob['EventInfo'].mc_t[0] # measured in JTE time
+    particle_type = event_blob['McTracks'][p].type
+    energy = event_blob['McTracks'][p].energy
+    is_cc = event_blob['McTracks'][p].is_cc
+    bjorkeny = event_blob['McTracks'][p].bjorkeny
+    dir_x, dir_y, dir_z = event_blob['McTracks'][p].dir_x, event_blob['McTracks'][p].dir_y, event_blob['McTracks'][p].dir_z
+    time_track = event_blob['McTracks'][p].time # actually always 0 for primary neutrino, measured in MC time
+    vertex_pos_x, vertex_pos_y, vertex_pos_z = event_blob['McTracks'][p].pos_x, event_blob['McTracks'][p].pos_y, event_blob['McTracks'][p].pos_z
+
+    # parse hits [x,y,z,time]
+    if do_mc_hits is True:
+        hits = event_blob["McHits"]
+    else:
+        hits = event_blob["Hits"]
+
+    if use_calibrated_file is False:
+        hits = geo.apply(hits)
+
+    if data_cuts['triggered'] is True:
+        hits = hits.__array__[hits.triggered.astype(bool)]
+        #hits = hits.triggered_hits # alternative, though it only works for the triggered condition!
+
+    pos_x, pos_y, pos_z = hits.pos_x.astype('float32'), hits.pos_y.astype('float32'), hits.pos_z.astype('float32')
+    hits_time = hits.time.astype('float32')
+    triggered = hits.triggered.astype('float32')
+
+    # save collected information to arrays event_track and event_hits
+    event_track = np.array([event_id, particle_type, energy, is_cc, bjorkeny, dir_x, dir_y, dir_z, time_track,
+                            run_id, vertex_pos_x, vertex_pos_y, vertex_pos_z], dtype=np.float32)
+
+    # time_residual_vertex = get_time_residual_nu_interaction_mean_triggered_hits(time_interaction, hits_time, triggered)
+    #
+    # # save collected information to arrays event_track and event_hits
+    # event_track = np.array([event_id, particle_type, energy, is_cc, bjorkeny, dir_x, dir_y, dir_z, time_track,
+    #                         run_id, vertex_pos_x, vertex_pos_y, vertex_pos_z, time_residual_vertex], dtype=np.float32)
+
+    ax = np.newaxis
+    event_hits = np.concatenate([pos_x[:, ax], pos_y[:, ax], pos_z[:, ax], hits_time[:, ax], triggered[:, ax]], axis=1)
+
+    if do4d[0] is True and do4d[1] == 'channel_id' or do4d[1] == 'xzt-c':
+        channel_id = hits.channel_id.astype('float32')
+        event_hits = np.concatenate([event_hits, channel_id[:, ax]], axis=1)
+
+    return event_hits, event_track
\ No newline at end of file