dwong, a package for DarkQuest data analysis.

dwong is a comprehensive Python package, created by student Dowling Wong, tailored for data analysis and neural network-based particle identification in the DarkQuest experiment. The aim of this project is to streamline DarkQuest's data analysis process by providing exemplary data-processing functions.

The package mainly contains four modules: dwong, dplot, dcsv and dkeras.

Contents and useful functions.

• dwong
  • emcal_bytuple(dq_events)
  • multi_clusters(dq_events)
  • h4_bytuple(dq_events)
  • prepare_data_bytuple(dq_events)
• dplot
  • emcal_evt(x, y, eng)
  • emcal_pdf(ntuple_name, fname, absolute_path)
• dkeras
  • train_model(x, y)
  • save_model(model, fname)
  • load_model(mname)
  • plot_confusion_matrix(cm, names, title='Confusion matrix', cmap=plt.cm.Blues)
  • plot_roc(pred,y)
• dcsv
  • gen_csv(filename)

Smaple of use.

dwong, main module for data analysis.

 import dwong
 
 dq_events = dwong.getData(filename, "Events") #data acquisition from n-tuple.
 (x, y, eng, labels, labels_decrease, seeds, seed_labels) = dwong.multi_clusters(dq_events)#here performed clustering
 (h4x, h4y) = h4_bytuple(dq_events)
 dq_st23 = dq_events["st23"]
 dq_track = dq_events["track"]
 gpz = dq_events["gen"]["pz"]
 trkls_coord = np.stack((dq_st23["x"], dq_st23["y"], dq_st23["z"], dq_st23["px"], dq_st23["py"], dq_st23["pz"]), axis=1)
 trkls_cal = np.stack((dq_st23["Cal_x"], dq_st23["Cal_y"]), axis=1)
 track_st3 = np.stack((dq_track["x"], dq_track["y"], dq_track["pz"]), axis=1)

 folded_list=dwong.prepare_data_bytuple(filename)#return a list of events, each event may contain multiple particles.   
 flat_list = [particle for event in folded_list for particle in event]
 labels = [0] * len(flat_list)
 labeled_flat_list = [[label, *particle] for label, particle in zip(labels, flat_list)]#list of particles, in a flat list.

dplot, plot module.

 import dwong
 from dwong import dplot

 dq_events = dwong.getData(filename, "Events") #data acquisition from n-tuple.
 (x, y, eng) = emcal_bytuple(dq_events)
 fig = emcal_evt(x, y, eng)
 #The emcal_evt will plot the emcal page in jupyter notebook, and if you have further consideration, it returns fig

 #save a pdf booklet for all the emcal plots for events in a root file
 if ntuple_name.endswith(".root")& (ntuple_name not in train):
        emcal_pdf(ntuple_name, fname, absolute_path)

 #or you can plot all n-tuple under a directory
 target_dir = os.listdir("/Users/dwong/Desktop/n-tuples/5_80_training/")
 for file_name in target_dir:
     os.chdir(taregt_dir)
     if file_name.endswith(".root")& (file_name not in train):
         emcal_pdf(ntuple_name, fname, absolute_path)
 

External Link

*DarkQuest Snowmass paper. *DarkQuest Collaboration code collection. *The source for this project is available here. *Dowling's code collection for data analysis, model training, particle ID and samples. *Analysis package dwong's Pypi page. *Dowling's personal website.

---