Add nominal debugging option

Author: LukeSeifert

mosden/countrate.py

@@ -159,6 +159,8 @@ def sample_parameter(val: ufloat, dist: str) -> float:
                 return np.random.normal(val.n, val.s)
             elif dist == 'uniform':
                 return np.random.uniform(val.n - val.s, val.n + val.s)
+            elif dist == 'nominal':
+                return val.n
             else:
                 raise NotImplementedError(f'{dist} sampling not implemented')
 

mosden/groupfit.py

@@ -715,7 +715,11 @@ def _nonlinear_least_squares(self,
             self.post_data['concMC'] = list()
         for post_data_vals in post_data_save:
             for key in self.post_data.keys():
-                self.post_data[key].append(post_data_vals[key])
+                try:
+                    self.post_data[key].append(post_data_vals[key])
+                except KeyError:
+                    # Running with pre-existing post data
+                    pass
         self.save_postproc()
 
         yields = np.zeros((self.num_groups, self.MC_samples))

mosden/postprocessing.py

@@ -252,6 +252,12 @@ def _chart_form(self, name: str, data: dict, cbar_label: str, vmin: float=1e-1,
                 continue
         vmin_use = 10 ** np.floor(np.log10(vmin))
         vmax_use = 10 ** np.ceil(np.log10(vmax))
+        if vmin_use == vmax_use:
+            if vmin_use == 0.0:
+                vmin_use = 0.1
+                vmax_use = 1.0
+            vmin_use = 0.1 * vmin_use
+            vmax_use = 10 * vmax_use
         norm = LogNorm(vmin=vmin_use, vmax=vmax_use)
         plt.scatter(N, Z, c=C, norm=norm, marker="s", s=60)
         plt.set_cmap('viridis')

mosden/templates/input_schema.json

@@ -335,7 +335,7 @@
                 "sample_func": {
                     "type": "string",
                     "description": "The function type to use for randomly sampling data",
-                    "enum": ["normal", "uniform"]
+                    "enum": ["normal", "uniform", "nominal"]
                 },
                 "initial_params": {
                     "type": "object",