Merge pull request #10 from s-kuberski/feature/irregularMC

Feature/irregular mc

pyerrors/linalg.py

@@ -42,7 +42,7 @@ def derived_array(func, data, **kwargs):
     n_obs = len(raveled_data)
     new_names = sorted(set([y for x in [o.names for o in raveled_data] for y in x]))
 
-    is_merged = len(list(filter(lambda o: o.is_merged is True, raveled_data))) > 0
+    is_merged = {name: (len(list(filter(lambda o: o.is_merged.get(name, False) is True, raveled_data))) > 0) for name in new_names}
     reweighted = len(list(filter(lambda o: o.reweighted is True, raveled_data))) > 0
     new_idl_d = {}
     for name in new_names:
@@ -52,8 +52,8 @@ def derived_array(func, data, **kwargs):
             if tmp is not None:
                 idl.append(tmp)
         new_idl_d[name] = _merge_idx(idl)
-        if not is_merged:
+        if not is_merged[name]:
-            is_merged = (1 != len(set([len(idx) for idx in [*idl, new_idl_d[name]]])))
+            is_merged[name] = (1 != len(set([len(idx) for idx in [*idl, new_idl_d[name]]])))
 
     if data.ndim == 1:
         values = np.array([o.value for o in data])
@@ -100,17 +100,19 @@ def derived_array(func, data, **kwargs):
         new_samples = []
         new_means = []
         new_idl = []
-        if is_merged:
+        filtered_deltas = {}
-            filtered_names, filtered_deltas, filtered_idl_d = _filter_zeroes(new_names, new_deltas, new_idl_d)
+        filtered_idl_d = {}
-        else:
+        for name in new_names:
-            filtered_names = new_names
+            if is_merged[name]:
-            filtered_deltas = new_deltas
+                filtered_deltas[name], filtered_idl_d[name] = _filter_zeroes(new_deltas[name], new_idl_d[name])
-            filtered_idl_d = new_idl_d
+            else:
-        for name in filtered_names:
+                filtered_deltas[name] = new_deltas[name]
+                filtered_idl_d[name] = new_idl_d[name]
+        for name in new_names:
             new_samples.append(filtered_deltas[name])
             new_means.append(new_r_values[name][i_val])
             new_idl.append(filtered_idl_d[name])
-        final_result[i_val] = Obs(new_samples, filtered_names, means=new_means, idl=new_idl)
+        final_result[i_val] = Obs(new_samples, new_names, means=new_means, idl=new_idl)
         final_result[i_val]._value = new_val
         final_result[i_val].is_merged = is_merged
         final_result[i_val].reweighted = reweighted

pyerrors/obs.py

@@ -118,7 +118,7 @@ class Obs:
                     raise Exception('Incompatible samples and idx for %s: %d vs. %d' % (name, len(sample), self.shape[name]))
                 self.r_values[name] = np.mean(sample)
                 self.deltas[name] = sample - self.r_values[name]
-        self.is_merged = False
+        self.is_merged = {}
         self.N = sum(list(self.shape.values()))
 
         self._value = 0
@@ -947,44 +947,33 @@ def _expand_deltas_for_merge(deltas, idx, shape, new_idx):
     return np.array([ret[new_idx[i] - new_idx[0]] for i in range(len(new_idx))])
 
 
-def _filter_zeroes(names, deltas, idl, eps=Obs.filter_eps):
+def _filter_zeroes(deltas, idx, eps=Obs.filter_eps):
     """Filter out all configurations with vanishing fluctuation such that they do not
-       contribute to the error estimate anymore. Returns the new names, deltas and
+       contribute to the error estimate anymore. Returns the new deltas and
-       idl according to the filtering.
+       idx according to the filtering.
        A fluctuation is considered to be vanishing, if it is smaller than eps times
        the mean of the absolute values of all deltas in one list.
 
     Parameters
     ----------
-    names : list
+    deltas : list
-        List of names
+        List of fluctuations
-    deltas : dict
+    idx : list
-        Dict lists of fluctuations
+        List or ranges of configs on which the deltas are defined.
-    idx : dict
-        Dict of lists or ranges of configs on which the deltas are defined.
-        Has to be a subset of new_idx.
     eps : float
         Prefactor that enters the filter criterion.
     """
-    new_names = []
+    new_deltas = []
-    new_deltas = {}
+    new_idx = []
-    new_idl = {}
+    maxd = np.mean(np.fabs(deltas))
-    for name in names:
+    for i in range(len(deltas)):
-        nd = []
+        if abs(deltas[i]) > eps * maxd:
-        ni = []
+            new_deltas.append(deltas[i])
-        maxd = np.mean(np.fabs(deltas[name]))
+            new_idx.append(idx[i])
-        for i in range(len(deltas[name])):
+    if new_idx:
-            if not np.isclose(0.0, deltas[name][i], atol=eps * maxd):
+        return np.array(new_deltas), new_idx
-                nd.append(deltas[name][i])
-                ni.append(idl[name][i])
-        if nd:
-            new_names.append(name)
-            new_deltas[name] = np.array(nd)
-            new_idl[name] = ni
-    if new_names:
-        return (new_names, new_deltas, new_idl)
     else:
-        return (names, deltas, idl)
+        return deltas, idx
 
 
 def derived_observable(func, data, **kwargs):
@@ -1034,7 +1023,7 @@ def derived_observable(func, data, **kwargs):
     n_obs = len(raveled_data)
     new_names = sorted(set([y for x in [o.names for o in raveled_data] for y in x]))
 
-    is_merged = len(list(filter(lambda o: o.is_merged is True, raveled_data))) > 0
+    is_merged = {name: (len(list(filter(lambda o: o.is_merged.get(name, False) is True, raveled_data))) > 0) for name in new_names}
     reweighted = len(list(filter(lambda o: o.reweighted is True, raveled_data))) > 0
     new_idl_d = {}
     for name in new_names:
@@ -1044,8 +1033,8 @@ def derived_observable(func, data, **kwargs):
             if tmp is not None:
                 idl.append(tmp)
         new_idl_d[name] = _merge_idx(idl)
-        if not is_merged:
+        if not is_merged[name]:
-            is_merged = (1 != len(set([len(idx) for idx in [*idl, new_idl_d[name]]])))
+            is_merged[name] = (1 != len(set([len(idx) for idx in [*idl, new_idl_d[name]]])))
 
     if data.ndim == 1:
         values = np.array([o.value for o in data])
@@ -1110,17 +1099,17 @@ def derived_observable(func, data, **kwargs):
         new_samples = []
         new_means = []
         new_idl = []
-        if is_merged:
+        for name in new_names:
-            filtered_names, filtered_deltas, filtered_idl_d = _filter_zeroes(new_names, new_deltas, new_idl_d)
+            if is_merged[name]:
-        else:
+                filtered_deltas, filtered_idl_d = _filter_zeroes(new_deltas[name], new_idl_d[name])
-            filtered_names = new_names
+            else:
-            filtered_deltas = new_deltas
+                filtered_deltas = new_deltas[name]
-            filtered_idl_d = new_idl_d
+                filtered_idl_d = new_idl_d[name]
-        for name in filtered_names:
+
-            new_samples.append(filtered_deltas[name])
+            new_samples.append(filtered_deltas)
+            new_idl.append(filtered_idl_d)
             new_means.append(new_r_values[name][i_val])
-            new_idl.append(filtered_idl_d[name])
+        final_result[i_val] = Obs(new_samples, new_names, means=new_means, idl=new_idl)
-        final_result[i_val] = Obs(new_samples, filtered_names, means=new_means, idl=new_idl)
         final_result[i_val]._value = new_val
         final_result[i_val].is_merged = is_merged
         final_result[i_val].reweighted = reweighted
@@ -1242,7 +1231,7 @@ def correlate(obs_a, obs_b):
         new_idl.append(obs_a.idl[name])
 
     o = Obs(new_samples, sorted(obs_a.names), idl=new_idl)
-    o.is_merged = obs_a.is_merged or obs_b.is_merged
+    o.is_merged = {name: (obs_a.is_merged.get(name, False) or obs_b.is_merged.get(name, False)) for name in o.names}
     o.reweighted = obs_a.reweighted or obs_b.reweighted
     return o
 
@@ -1603,6 +1592,6 @@ def merge_obs(list_of_obs):
 
     names = sorted(new_dict.keys())
     o = Obs([new_dict[name] for name in names], names, idl=[idl_dict[name] for name in names])
-    o.is_merged = np.any([oi.is_merged for oi in list_of_obs])
+    o.is_merged = {name: np.any([oi.is_merged.get(name, False) for oi in list_of_obs]) for name in o.names}
     o.reweighted = np.max([oi.reweighted for oi in list_of_obs])
     return o