fix vector observables

python/loadleveller/mcextract.py

@@ -10,9 +10,8 @@ class Observable:
         self.rebinning_bin_count = np.zeros(num_tasks)
         self.autocorrelation_time = np.zeros(num_tasks)+np.nan
 
-        # have to be set later because dimension is unknown
-        self.mean = None
-        self.error = None        
+        self.mean = [None for i in range(num_tasks)]
+        self.error = [None for i in range(num_tasks)]
 
 class MCArchive:
     def __init__(self, filename):
@@ -36,19 +35,8 @@ class MCArchive:
                 o.rebinning_bin_count[i] = value['rebinning_bin_count']
                 o.autocorrelation_time[i] = value['autocorrelation_time']
 
-                # fill in dimensions on first occurence of the observable
-                if np.all(o.mean == None):
-                    o.mean = np.zeros([self.num_tasks, len(value['mean'])]) + np.nan
-                    o.error = np.zeros([self.num_tasks, len(value['error'])]) + np.nan
-
-                    if o.mean.shape != o.error.shape:
-                        raise Exception('observable "{}": dimension mismatch between mean and error'.format(obs))
-                else:
-                    if len(value['mean']) != o.mean.shape[1] or len(value['error']) != o.error.shape[1]:
-                        raise Exception('observable "{}": dimension mismatch between different tasks'.format(obs))
-
-                o.mean[i,:] = value['mean']
-                o.error[i,:] = value['error']
+                o.mean[i] = value['mean']
+                o.error[i] = value['error']
 
     def filter_mask(self, filter):
         if not filter:
@@ -71,11 +59,15 @@ class MCArchive:
         selection.rebinning_bin_count = orig.rebinning_bin_count[mask]
         selection.rebinning_bin_length = orig.rebinning_bin_length[mask]
         selection.autocorrelation_time = orig.autocorrelation_time[mask]
-        selection.mean = orig.mean[mask,:]
-        selection.error = orig.error[mask,:]
+        
+        selection.mean = [m for i, m in enumerate(orig.mean) if mask[i]]
+        selection.error = [m for i, m in enumerate(orig.error) if mask[i]]
+
+        if all(len(m) == len(selection.mean[0]) for m in selection.mean):
+            selection.mean = np.array(selection.mean)
+            selection.error = np.array(selection.error)
 
-        # if it is not a vector observable we can make it simpler for the user
-        if selection.mean.shape[-1] == 1:
+        if selection.mean.shape[1] == 1:
             selection.mean = selection.mean.flatten()
             selection.error = selection.error.flatten()
 

src/merger.cpp

@@ -124,18 +124,19 @@ results merge(const std::vector<std::string> &filenames, const std::vector<evala
 				size_t vector_idx = i % vector_length;
 
 				obs.mean[vector_idx] += samples[i];
-				metadata[obs_name].sample_counter++;
+				if(vector_idx == vector_length-1) {
+					metadata[obs_name].sample_counter++;
+				}
 			}
 		}
 	}
 
-	for(auto &entry : res.observables) {
-		auto &obs = entry.second;
-		assert(metadata[entry.first].sample_counter ==
+	for(auto &[obs_name, obs] : res.observables) {
+		assert(metadata[obs_name].sample_counter ==
 		       obs.rebinning_bin_count * obs.rebinning_bin_length);
-
+		
 		for(auto &mean : obs.mean) {
-			mean /= metadata[entry.first].sample_counter;
+			mean /= metadata[obs_name].sample_counter;			
 		}
 	}
 
@@ -181,9 +182,8 @@ results merge(const std::vector<std::string> &filenames, const std::vector<evala
 		}
 	}
 
-	for(auto &entry : res.observables) {
-		auto &obs = entry.second;
-		assert(metadata[entry.first].current_rebin == obs.rebinning_bin_count);
+	for(auto &[obs_name, obs] : res.observables) {
+		assert(metadata[obs_name].current_rebin == obs.rebinning_bin_count);
 		for(size_t i = 0; i < obs.error.size(); i++) {
 			int used_samples = obs.rebinning_bin_count * obs.rebinning_bin_length;
 			double no_rebinning_error =