Pulse: snr plot: indicate masking

simulations/11_pulsed_timing.py

@@ -5,7 +5,7 @@ from lib import util
 from scipy import signal, interpolate, stats
 import matplotlib.pyplot as plt
 import numpy as np
-from itertools import zip_longest
+from itertools import zip_longest, pairwise
 import h5py
 from copy import deepcopy
 
@@ -232,6 +232,7 @@ if __name__ == "__main__":
     h5_cache_fname = f'11_pulsed_timing.hdf5'
 
     time_accuracies = np.zeros(len(snr_factors))
+    mask_counts = np.zeros(len(snr_factors))
     for k, snr_sigma_factor in tqdm(enumerate(snr_factors)):
         # Read in cached time residuals
         if True:
@@ -448,15 +449,37 @@ if __name__ == "__main__":
 
         # Make a plot of the time residuals
         if N_residuals > 1:
-            time_accuracies[k] = np.std(time_residuals[:N_residuals])
+            time_residuals = time_residuals[:N_residuals]
+
+            for i in range(1 + cut_wrong_peak_matches):
+                mask_count = 0
+
+                if i==1: # if cut_wrong_peak_matches:
+                    wrong_peak_condition = lambda t_res: abs(t_res) > antenna_dt*4
+
+                    mask = wrong_peak_condition(time_residuals)
+
+                    mask_count = np.count_nonzero(mask)
+
+                    print("Masking {} residuals".format(mask_count))
+                    time_residuals = time_residuals[~mask]
+
+                    if not mask_count:
+                        print("Continuing")
+                        continue
+
+                time_accuracies[k] = np.std(time_residuals)
+                mask_counts[k] = mask_count
 
                 hist_kwargs = dict(bins='sqrt', density=False, alpha=0.8, histtype='step')
                 fig, ax = plt.subplots()
                 ax.set_title(
                         "Template Correlation Lag finding"
-                    + f"\n template dt: {template_dt*1e3: .1e}ps"
+                        + f"\n template dt: {template_dt: .1e}ns"
                         + f"; antenna dt: {antenna_dt: .1e}ns"
-                    + f"; noise_factor: {noise_sigma_factor: .1e}"
+                        + ";" if not mask_count else "\n"
+                        + f"snr_factor: {snr_sigma_factor: .1e}"
+                        + "" if not mask_count else f"; N_masked: {mask_count}"
                         )
                 ax.set_xlabel("Time Residual [ns]")
                 ax.set_ylabel("#")
@@ -503,7 +526,10 @@ if __name__ == "__main__":
 
                     ax.text( *(0.02, 0.95), text_str, fontsize=12, ha='left', va='top', transform=ax.transAxes)
 
-            fig.savefig(f"figures/11_time_residual_hist_tdt{template_dt:0.1e}_n{noise_sigma_factor: .1e}.pdf")
+                if mask_count:
+                    fig.savefig(f"figures/11_time_residual_hist_tdt{template_dt:0.1e}_n{snr_sigma_factor:.1e}_masked.pdf")
+                else:
+                    fig.savefig(f"figures/11_time_residual_hist_tdt{template_dt:0.1e}_n{snr_sigma_factor:.1e}.pdf")
 
                 if True:
                     plt.close(fig)
@@ -526,7 +552,17 @@ if __name__ == "__main__":
             ax.set_yscale('log')
 
         # plot the values
-        ax.plot(np.asarray(snr_factors), time_accuracies, ls='none', marker='o')
+        l = None
+        for j, mask_threshold in enumerate(pairwise([np.inf, 250, 50, 1, 0])):
+            kwargs = dict(
+                    ls='none',
+                    marker=['^', 'v','8', 'o',][j],
+                    color=None if l is None else l[0].get_color(),
+                    )
+            mask = mask_counts >= mask_threshold[1]
+            mask &= mask_counts < mask_threshold[0]
+
+            l = ax.plot(snr_factors[mask], time_accuracies[mask], **kwargs)
 
         if True: # limit y-axis to 1e0
             ax.set_ylim([None, 1e1])