merge Reimar's changes

nifty5/minimization/descent_minimizers.py

@@ -166,7 +166,8 @@ class NewtonCG(DescentMinimizer):
     """
 
     def __init__(self, controller, napprox=0, line_searcher=None, name=None,
-                 nreset=20, file_name=None):
+                 nreset=20, max_cg_iterations=200, energy_reduction_factor=0.1,
+                 file_name=None):
         if line_searcher is None:
             line_searcher = LineSearch(preferred_initial_step_size=1.)
         super(NewtonCG, self).__init__(controller=controller,
@@ -175,15 +176,16 @@ class NewtonCG(DescentMinimizer):
         self._name = name
         self._nreset = nreset
         self._file_name = file_name
+        self._max_cg_iterations = max_cg_iterations
+        self._alpha = energy_reduction_factor
 
     def get_descent_direction(self, energy, old_value=None):
         if old_value is None:
             ic = GradientNormController(iteration_limit=5)
         else:
-            alpha = 0.1
-            ediff = alpha*(old_value-energy.value)
+            ediff = self._alpha*(old_value-energy.value)
             ic = AbsDeltaEnergyController(
-                ediff, iteration_limit=200, name=self._name,
+                ediff, iteration_limit=self._max_cg_iterations, name=self._name,
                 file_name=self._file_name)
         e = QuadraticEnergy(0*energy.position, energy.metric, energy.gradient)
         p = None

nifty5/probing.py

@@ -138,7 +138,6 @@ def probe_diagonal(op, nprobes, random_type="pm1"):
 
 
 def approximation2endo(op, nsamples):
-    print('Calculate preconditioner')
     sc = StatCalculator()
     for _ in range(nsamples):
         sc.add(op.draw_sample())