Merge pull request #274 from python-adaptive/feature/min_npoints

AverageLearner: implement min_npoints

Author: basnijholt

adaptive/learner/average_learner.py

@@ -19,6 +19,8 @@ class AverageLearner(BaseLearner):
         Desired absolute tolerance.
     rtol : float
         Desired relative tolerance.
+    min_npoints : int
+        Minimum number of points to sample.
 
     Attributes
     ----------
@@ -30,7 +32,7 @@ class AverageLearner(BaseLearner):
         Number of evaluated points.
     """
 
-    def __init__(self, function, atol=None, rtol=None):
+    def __init__(self, function, atol=None, rtol=None, min_npoints=2):
         if atol is None and rtol is None:
             raise Exception("At least one of `atol` and `rtol` should be set.")
         if atol is None:
@@ -44,6 +46,8 @@ def __init__(self, function, atol=None, rtol=None):
         self.atol = atol
         self.rtol = rtol
         self.npoints = 0
+        # Cannot estimate standard deviation with fewer than 2 points.
+        self.min_npoints = max(min_npoints, 2)
         self.sum_f = 0
         self.sum_f_sq = 0
 
@@ -92,7 +96,7 @@ def std(self):
         """The corrected sample standard deviation of the values
         in `data`."""
         n = self.npoints
-        if n < 2:
+        if n < self.min_npoints:
             return np.inf
         numerator = self.sum_f_sq - n * self.mean ** 2
         if numerator < 0:
@@ -106,7 +110,7 @@ def loss(self, real=True, *, n=None):
             n = self.npoints if real else self.n_requested
         else:
             n = n
-        if n < 2:
+        if n < self.min_npoints:
             return np.inf
         standard_error = self.std / sqrt(n)
         return max(
@@ -150,10 +154,11 @@ def __getstate__(self):
             self.function,
             self.atol,
             self.rtol,
+            self.min_npoints,
             self._get_data(),
         )
 
     def __setstate__(self, state):
-        function, atol, rtol, data = state
-        self.__init__(function, atol, rtol)
+        function, atol, rtol, min_npoints, data = state
+        self.__init__(function, atol, rtol, min_npoints)
         self._set_data(data)

adaptive/tests/test_average_learner.py

@@ -4,6 +4,7 @@
 import numpy as np
 
 from adaptive.learner import AverageLearner
+from adaptive.runner import simple
 
 
 def test_only_returns_new_points():
@@ -46,3 +47,15 @@ def test_avg_std_and_npoints():
             assert learner.npoints == len(learner.data)
             assert abs(learner.sum_f - values.sum()) < 1e-13
             assert abs(learner.std - std) < 1e-13
+
+
+def test_min_npoints():
+    def constant_function(seed):
+        return 0.1
+
+    for min_npoints in [1, 2, 3]:
+        learner = AverageLearner(
+            constant_function, atol=0.01, rtol=0.01, min_npoints=min_npoints
+        )
+        simple(learner, lambda l: l.loss() < 1)
+        assert learner.npoints >= max(2, min_npoints)