Merge pull request #12 from jeremyleung521/westpa-2.0-restruct-thresh…

…olds2

WESTPA2-threshold2 fix

src/westpa/core/sim_manager.py

@@ -197,6 +197,11 @@ def report_bin_statistics(self, bins, save_summary=False):
         self.rc.pstatus('norm = {:g}, error in norm = {:g} ({:.2g}*epsilon)'.format(norm, (norm - 1), (norm - 1) / EPS))
         self.rc.pflush()
 
+        if min_seg_prob < 1e-100:
+            log.warning(
+                '\n Minimum segment weight is < 1e-100 and might not be physically relevant. Please reconsider your progress coordinate or binning scheme.'
+            )
+
         if save_summary:
             iter_summary = self.data_manager.get_iter_summary()
             iter_summary['n_particles'] = len(segments)

src/westpa/core/we_driver.py

@@ -534,10 +534,10 @@ def _split_by_weight(self, bin, target_count, ideal_weight, number_of_groups):
             #   print(i.weight)
             # print(self.smallest_allowed_weight)
             if all(new_segment.weight < self.smallest_allowed_weight for new_segment in new_segments_list):
-                #    print("instance of threshold break (bw)")
+                # print("instance of threshold break (bw)")
                 bin.add(segment)
             else:
-                #    print("no threshold break (bw)")
+                # print("no threshold break (bw)")
                 bin.update(new_segments_list)
 
     # KFW CHECK BACK            for new_segment in new_segments_list:
@@ -586,6 +586,7 @@ def _adjust_count(self, bin, groups, target_count):
 
         # split
         while len(bin) < threshold_target_count:
+            last_bin = len(bin)
             for i in sorted_groups:
                 log.debug('adjusting counts by splitting')
                 # always split the highest probability walker into two
@@ -594,15 +595,15 @@ def _adjust_count(self, bin, groups, target_count):
                 bin.remove(segments[-1])
                 i.remove(segments[-1])
                 new_segments_list = self._split_walker(segments[-1], 2, bin)
+                # print(new_segments_list)
                 # for j in new_segments_list:
-                #   print(j.weight)
-                # print(self.smallest_allowed_weight)
+                #    print(j.weight)
                 if all(new_segment.weight < self.smallest_allowed_weight for new_segment in new_segments_list):
-                    #    print("instance of threshold break (ac)")
+                    # print("instance of threshold break (ac)")
                     bin.add(segments[-1])
                     i.add(segments[-1])
                 else:
-                    #    print("no threshold break (ac)")
+                    # print("no threshold break (ac)")
                     i.update(new_segments_list)
                     bin.update(new_segments_list)
 
@@ -614,13 +615,16 @@ def _adjust_count(self, bin, groups, target_count):
 
                 if len(bin) == target_count:
                     break
-                elif i == sorted_groups[-1]:
+                elif i == sorted_groups[-1] and last_bin == len(
+                    bin
+                ):  # If the last "for" iteration didn't change anything, soft-break.
                     threshold_target_count = len(bin)
 
         threshold_target_count = target_count
 
         # merge
         while len(bin) > threshold_target_count:
+            last_bin = len(bin)
             sorted_groups.reverse()
             # Adjust to go from lowest weight group to highest to merge
             for i in sorted_groups:
@@ -649,7 +653,9 @@ def _adjust_count(self, bin, groups, target_count):
                     # in bash, "find . -name \*.py | xargs fgrep -n '_merge_walkers'"
                     if len(bin) == target_count:
                         break
-                    elif i == sorted_groups[-1]:
+                    elif i == sorted_groups[-1] and last_bin == len(
+                        bin
+                    ):  # If the last "for" iteration didn't change anything, soft-break.
                         threshold_target_count = len(bin)
 
     def _check_pre(self):