Fixed updating in on travis pip env, minor fixes in stagedoptimizer to avoid warnings with BayesianOptimizer, and solved instability of constrained_bo notebook

Author: javdrher

.travis.yml

@@ -9,10 +9,10 @@ cache: pip
 install:
   - pip install -U pip wheel
   - pip install tensorflow==1.0.1
-  - pip install --process-dependency-links .
-  - pip install .[test]
-  - pip install codecov
+  - pip install -U --process-dependency-links .
+  - pip install -U .[test]
+  - pip install -U codecov
 script:
-  - nosetests --nocapture testing
+  - nosetests testing
 after_success:
   - codecov

GPflowOpt/optim.py

@@ -142,6 +142,8 @@ def __init__(self, domain, nsamples):
         """
         super(MCOptimizer, self).__init__(domain, exclude_gradient=True)
         self._nsamples = nsamples
+        # Clear the initial data points
+        self.set_initial(np.empty((0, self.domain.size)))
 
     @Optimizer.domain.setter
     def domain(self, dom):
@@ -163,9 +165,11 @@ def _optimize(self, objective):
 
     def set_initial(self, initial):
         initial = np.atleast_2d(initial)
-        super(MCOptimizer, self).set_initial(initial)
         if initial.size > 0:
             warnings.warn("Initial points set in {0} are ignored.".format(self.__class__.__name__), UserWarning)
+            return
+
+        super(MCOptimizer, self).set_initial(initial)
 
 
 class CandidateOptimizer(MCOptimizer):
@@ -183,8 +187,6 @@ def __init__(self, domain, candidates):
         super(CandidateOptimizer, self).__init__(domain, candidates.shape[0])
         assert (candidates in domain)
         self.candidates = candidates
-        # Clear the initial data points
-        self.set_initial(np.empty((0, self.domain.size)))
 
     def _get_eval_points(self):
         return self.candidates
@@ -235,11 +237,11 @@ def __init__(self, optimizers):
         no_gradient = any(map(lambda opt: not opt.gradient_enabled(), optimizers))
         super(StagedOptimizer, self).__init__(optimizers[0].domain, exclude_gradient=no_gradient)
         self.optimizers = optimizers
-        self.set_initial(np.empty((0, self.domain.size)))
+        del self._initial
 
     @Optimizer.domain.setter
     def domain(self, domain):
-        super(StagedOptimizer, self.__class__).domain.fset(self, domain)
+        self._domain = domain
         for optimizer in self.optimizers:
             optimizer.domain = domain
 
@@ -254,7 +256,6 @@ def optimize(self, objectivefx):
         is returned.
         """
 
-        self.optimizers[0].set_initial(self.get_initial())
         results = []
         for current, following in zip(self.optimizers[:-1], self.optimizers[1:]):
             result = current.optimize(objectivefx)
@@ -273,3 +274,9 @@ def optimize(self, objectivefx):
         if any(r.success for r in results):
             result.x, result.fun = self._best_x(results)
         return result
+
+    def get_initial(self):
+        return self.optimizers[0].get_initial()
+
+    def set_initial(self, initial):
+        self.optimizers[0].set_initial(initial)

doc/source/notebooks/constrained_bo.ipynb

@@ -9,7 +9,6 @@
 def parabola2d(X):
     return np.atleast_2d(np.sum(X ** 2, axis=1)).T
 
-
 def plane(X):
     return X[:, [0]] - 0.5
 
@@ -125,7 +124,6 @@ def test_enable_scaling(self):
         self.assertFalse(
             any(m.wrapped.X.value in GPflowOpt.domain.UnitCube(self.domain.size) for m in self.acquisition.models))
         self.acquisition.enable_scaling(self.domain)
-        print(self.acquisition.models[0].wrapped.X.value)
         self.assertTrue(
             all(m.wrapped.X.value in GPflowOpt.domain.UnitCube(self.domain.size) for m in self.acquisition.models))
 

testing/test_acquisition.py

@@ -103,6 +103,4 @@ def test_predict_scaling(self):
         Yt = parabola2d(Xt) #+ np.random.rand(20, 1) * 0.05
         fr = m.predict_density(Xt, Yt)
         fs = n.predict_density(Xt, Yt)
-        print(fr)
-        print(fs)
         np.testing.assert_allclose(fr, fs, rtol=1e-3)

testing/test_datascaler.py

@@ -81,7 +81,7 @@ def test_set_initial(self):
         # When run separately this test works, however when calling nose to run all tests on python 2.7 this records
         # no warnings
         with warnings.catch_warnings(record=True) as w:
-            super(TestCandidateOptimizer, self).test_set_initial()
+            self.optimizer.set_initial([1, 1])
             assert len(w) == 1
             assert issubclass(w[-1].category, UserWarning)
 
@@ -150,7 +150,14 @@ def setUp(self):
                                                           GPflowOpt.optim.SciPyOptimizer(self.domain, maxiter=10)])
 
     def test_default_initial(self):
-        self.assertTupleEqual(self.optimizer._initial.shape, (0,2))
+        self.assertTupleEqual(self.optimizer.optimizers[0]._initial.shape, (0,2))
+
+    def test_set_initial(self):
+        self.optimizer.set_initial([1, 1])
+        self.assertTupleEqual(self.optimizer.optimizers[0]._initial.shape, (0, 2))
+        self.assertTupleEqual(self.optimizer.optimizers[1]._initial.shape, (0, 2))
+        self.assertTupleEqual(self.optimizer.optimizers[2]._initial.shape, (1, 2))
+        self.assertTupleEqual(self.optimizer.get_initial().shape, (0, 2))
 
     def test_object_integrity(self):
         self.assertEqual(len(self.optimizer.optimizers), 3, msg="Two optimizers expected in optimizerlist")
@@ -181,7 +188,6 @@ def test_optimizer_interrupt(self):
             self.assertEqual(result.nfev, 5)
 
             result = self.optimizer.optimize(KeyboardRaiser(12, parabola2d))
-            print(result)
             self.assertFalse(result.success, msg="non-succesfull result expected.")
             self.assertEqual(result.nfev, 12)
             self.assertFalse(np.allclose(result.x[0, :], 0.0), msg="The optimum should not be found yet")