FIX inconsistent hyperparameter and bounds issue in GP kernels (sciki…

…t-learn#17946)

examples/gaussian_process/plot_gpr_co2.py

@@ -129,7 +129,7 @@ def load_mauna_loa_atmospheric_co2():
 k3 = 0.5**2 * RationalQuadratic(length_scale=1.0, alpha=1.0)
 k4 = 0.1**2 * RBF(length_scale=0.1) \
  + WhiteKernel(noise_level=0.1**2,
- noise_level_bounds=(1e-3, np.inf)) # noise terms
+ noise_level_bounds=(1e-5, np.inf)) # noise terms
 kernel = k1 + k2 + k3 + k4
 
 gp = GaussianProcessRegressor(kernel=kernel, alpha=0,

sklearn/gaussian_process/kernels.py

@@ -395,6 +395,8 @@ def _check_bounds_params(self):
  np.atleast_2d(self.theta).T)
  idx = 0
  for hyp in self.hyperparameters:
+ if hyp.fixed:
+ continue
  for dim in range(hyp.n_elements):
  if list_close[idx, 0]:
  warnings.warn("The optimal value found for "

sklearn/gaussian_process/tests/test_gpr.py

@@ -14,7 +14,7 @@
 from sklearn.gaussian_process import GaussianProcessRegressor
 from sklearn.gaussian_process.kernels \
  import RBF, ConstantKernel as C, WhiteKernel
-from sklearn.gaussian_process.kernels import DotProduct
+from sklearn.gaussian_process.kernels import DotProduct, ExpSineSquared
 from sklearn.gaussian_process.tests._mini_sequence_kernel import MiniSeqKernel
 from sklearn.exceptions import ConvergenceWarning
 
@@ -525,3 +525,14 @@ def test_warning_bounds():
  "specified lower bound 10.0. "
  "Decreasing the bound and calling "
  "fit again may find a better value.")
+
+
+def test_bound_check_fixed_hyperparameter():
+ # Regression test for issue #17943
+ # Check that having a hyperparameter with fixed bounds doesn't cause an
+ # error
+ k1 = 50.0**2 * RBF(length_scale=50.0) # long term smooth rising trend
+ k2 = ExpSineSquared(length_scale=1.0, periodicity=1.0,
+ periodicity_bounds="fixed") # seasonal component
+ kernel = k1 + k2
+ GaussianProcessRegressor(kernel=kernel).fit(X, y)