Save dtype of valid_fit as bool (#123)

* Save dtype of `valid_fit` as bool

* Fix typo

alea/runner.py

@@ -257,8 +257,10 @@ def _get_parameter_list(self):
         """Get parameter list and result list from statistical model."""
         parameter_list = sorted(self.model.get_parameter_list())
         # add likelihood, lower limit, upper limit, and the migrad valid fit bool
-        result_names = parameter_list + ["ll", "dl", "ul", "valid_fit"]
+        result_names = parameter_list + ["ll", "dl", "ul"]
         result_dtype = [(n, float) for n in result_names]
+        result_names += ["valid_fit"]
+        result_dtype += [("valid_fit", bool)]
         return result_names, result_dtype
 
     def _get_hypotheses(self):

alea/submitters/local.py

@@ -187,12 +187,16 @@ def submit(
             lltrue = results[true_name]["ll"]
             llrs = 2.0 * (llfree - lltrue)
             if llrs.min() < 0.0:
+                mean_valid = (
+                    results[free_name]["valid_fit"] & results[true_name]["valid_fit"]
+                ).mean()
                 self.logging.warning(
-                    f"The lowest log likelihood ratio is negative {llrs.min():.2e}, "
+                    f"The lowest log likelihood ratio is negative {llrs.min():.02e}, "
                     f"total fraction of negative log likelihood ratio is "
-                    f"{(llrs < 0.0).sum() / len(llrs):.2f}, "
-                    "the median if negative log likelihood ratios "
-                    f"is {np.median(llrs[llrs < 0.0]):.2e}, "
+                    f"{(llrs < 0.0).sum() / len(llrs):.02f}, "
+                    f"total fraction of invalid fit is {1 - mean_valid:.02f}, "
+                    f"the median of negative log likelihood ratios "
+                    f"is {np.median(llrs[llrs < 0.0]):.02e}, "
                     f"there might be a problem in your fitting."
                 )
             if len(llrs) < 1000: