Merge pull request #1057 from dstl/akkf

Fix failing test for AKKF proposal multivariate sampling to compare mean of particle state

Author: sdhiscocks

stonesoup/predictor/kernel.py

@@ -12,9 +12,9 @@
 
 
 class AdaptiveKernelKalmanPredictor(KalmanPredictor):
-    r"""An implementation of the adaptive kernel Kalman filter (AKKF) predictor. Here, the AKKF draws
-    inspiration from the concepts of kernel mean embeddings (KME) and Kalman Filter to address
-    tracking problems in nonlinear systems.
+    r"""An implementation of the adaptive kernel Kalman filter (AKKF) predictor. Here, the AKKF
+    draws inspiration from the concepts of kernel mean embeddings (KME) and Kalman Filter to
+    address tracking problems in nonlinear systems.
 
     In the state space, at time :math:`k`, the prior state
     particles are generated by passing the proposal particles at time :math:`k-1`, i.e.,

stonesoup/predictor/tests/test_kernel.py

@@ -13,7 +13,6 @@
 from ...types.update import KernelParticleStateUpdate
 
 number_particles = 4
-np.random.seed(50)
 timestamp = datetime.datetime.now()
 samples = multivariate_normal.rvs([0, 0, 0, 0],
                                   np.diag([0.01, 0.005, 0.1, 0.5])**2,

stonesoup/types/tests/test_state.py

@@ -863,7 +863,6 @@ def test_asd_weighted_gaussian_state():
 def test_kernel_particle_state():
     number_particles = 5
     weights = np.array([1 / number_particles] * number_particles)
-    np.random.seed(50)
 
     samples = multivariate_normal.rvs([0, 0, 0, 0],
                                       np.diag([0.01, 0.005, 0.1, 0.5]) ** 2,

stonesoup/updater/tests/test_kernel.py

@@ -19,7 +19,6 @@
 new_timestamp = timestamp + time_diff
 
 number_particles = 5
-np.random.seed(50)
 samples = multivariate_normal.rvs([0, 0, 0, 0],
                                   np.diag([0.01, 0.005, 0.1, 0.5])**2,
                                   size=number_particles)
@@ -107,7 +106,10 @@ def test_kernel_updater(kernel, measurement_model, c, ialpha):
     assert measurement.timestamp == gt_state.timestamp
     assert update.hypothesis.measurement.timestamp == gt_state.timestamp
     assert np.allclose(update.state_vector, prediction.state_vector)
-    assert np.allclose(update.proposal, StateVectors(new_state_vector.T), atol=1e0)
+    assert np.allclose(
+        np.mean(update.proposal, axis=1),
+        np.mean(StateVectors(new_state_vector.T), axis=1),
+        atol=2)
     assert np.allclose(update.weight, updated_weights.ravel())
     assert np.allclose(update.kernel_covar, updated_covariance)