Save min and max values

slr/get_slr_distributions.py

@@ -57,6 +57,8 @@ for i, row in dff.iterrows():
 
     dff.loc[i, 'loc'] = loc
     dff.loc[i, 'scale'] = scale
+    dff.loc[i, 'min'] = x_min
+    dff.loc[i, 'max'] = x_max
 
     if not PLOT:
         continue
@@ -87,7 +89,8 @@ for i, row in dff.iterrows():
     plt.show()
 
 # Save distribution parameters
-dff[['loc', 'scale']].to_csv('cauchy-values.csv', float_format='%g')
+dff[['loc', 'scale', 'min', 'max']].to_csv('cauchy-values.csv',
+                                           float_format='%g')
 
 if PLOT:
     # Plot all distributions
@@ -100,9 +103,7 @@ if PLOT:
     for i, row in dff.iterrows():
         j += 1
 
-        x_min = row[5] + (row[5] - row[50])
-        x_max = row[95] + (row[95] - row[50])
-        x = np.linspace(x_min, x_max, num=1000)
+        x = np.linspace(row['min'], row['max'], num=1000)
 
         y = stats.cauchy(loc=row['loc'], scale=row['scale']).pdf(x)
         ax.plot(x, y * row['scale'], c=c[j])
@@ -131,26 +132,24 @@ if PLOT:
         r = stats.cauchy(loc=row['loc'], scale=row['scale']).rvs(1000000)
 
         # Clip to our range
-        x_min = row[5] + (row[5] - row[50])
-        x_max = row[95] + (row[95] - row[50])
-        rc = np.clip(r, a_min=x_min, a_max=x_max)
+        rc = np.clip(r, a_min=row['min'], a_max=row['max'])
         f = (r != rc).sum() / len(r)  # Fraction outside range
 
         ax.hist(rc, 100)
 
         ym = ax.get_ylim()[1]  # Maximum y value
 
-        ax.axvline(x=x_min, c='C3')
+        ax.axvline(x=row['min'], c='C3')
         ax.axvline(x=row[5], c='C3')
         ax.axvline(x=row[50], c='C3')
         ax.axvline(x=row[95], c='C3')
-        ax.axvline(x=x_max, c='C3')
+        ax.axvline(x=row['max'], c='C3')
 
-        ax.annotate(' P0', (x_min, ym))
+        ax.annotate(' P_min', (row['min'], ym))
         ax.annotate(' P5', (row[5], ym))
         ax.annotate(' P50', (row[50], ym))
         ax.annotate(' P95', (row[95], ym))
-        ax.annotate(' P100', (x_max, ym))
+        ax.annotate(' P_max', (row['max'], ym))
 
         ax.annotate(f' Samples clipped = {100 * f:0.1f}%', (x_max, ym / 2))