Bugfix/172 prediction with no inputs bug

.github/workflows/test-package.yml

@@ -10,7 +10,7 @@ jobs:
  strategy:
  fail-fast: false
  matrix:
- python-version: ['3.8', '3.9', '3.10', '3.11']
+ python-version: ['3.8', '3.9', '3.10', '3.11', '3.12']
  steps:
  - uses: actions/checkout@v2
  - name: Set up Python ${{matrix.python-version}}

pykoop/centers.py

@@ -11,6 +11,7 @@
 from scipy import stats
 
 log = logging.getLogger(__name__)
+log.addHandler(logging.NullHandler())
 
 
 class Centers(sklearn.base.BaseEstimator, metaclass=abc.ABCMeta):

pykoop/kernel_approximation.py

@@ -426,7 +426,7 @@ def fit(
  X_hashed = self._hash_samples(X_scaled)
  encoder_args = {
  'categories': 'auto',
- 'sparse': False,
+ 'sparse_output': False,
  'handle_unknown': 'ignore',
  }
  if self.encoder_kw is not None:

pykoop/lmi_regressors.py

@@ -191,7 +191,7 @@ def __init__(
  inv_method: str = 'svd',
  tsvd: Optional[tsvd.Tsvd] = None,
  square_norm: bool = False,
- picos_eps: float = 0,
+ picos_eps: float = 1e-6,
  solver_params: Optional[Dict[str, Any]] = None,
  ) -> None:
  """Instantiate :class:`LmiEdmd`.
@@ -494,7 +494,7 @@ def __init__(
  tsvd_shifted: Optional[tsvd.Tsvd] = None,
  reg_method: str = 'tikhonov',
  square_norm: bool = False,
- picos_eps: float = 0,
+ picos_eps: float = 1e-6,
  solver_params: Optional[Dict[str, Any]] = None,
  ) -> None:
  """Instantiate :class:`LmiDmdc`.
@@ -672,7 +672,7 @@ def _create_base_problem(
  - big_constant.T * U_hat.T, U_hat * Q_bar_Sigma_tld
  ],
  [Q_bar_Sigma_tld.T * U_hat.T, m1],
- ]) << picos_eps)
+ ]) << -1 * picos_eps)
  problem.set_objective('min', picos.trace(W_hat))
  return problem
 
@@ -736,7 +736,7 @@ def __init__(
  alpha: float = 0,
  inv_method: str = 'svd',
  tsvd: Optional[tsvd.Tsvd] = None,
- picos_eps: float = 0,
+ picos_eps: float = 1e-6,
  solver_params: Optional[Dict[str, Any]] = None,
  ) -> None:
  """Instantiate :class:`LmiEdmdSpectralRadiusConstr`.
@@ -988,7 +988,7 @@ def __init__(
  alpha: float = 0,
  tsvd_unshifted: Optional[tsvd.Tsvd] = None,
  tsvd_shifted: Optional[tsvd.Tsvd] = None,
- picos_eps: float = 0,
+ picos_eps: float = 1e-6,
  solver_params: Optional[Dict[str, Any]] = None,
  ) -> None:
  """Instantiate :class:`LmiDmdcSpectralRadiusConstr`.
@@ -1264,7 +1264,7 @@ def __init__(
  inv_method: str = 'svd',
  tsvd: Optional[tsvd.Tsvd] = None,
  square_norm: bool = False,
- picos_eps: float = 0,
+ picos_eps: float = 1e-6,
  solver_params: Optional[Dict[str, Any]] = None,
  ) -> None:
  """Instantiate :class:`LmiEdmdHinfReg`.
@@ -1603,7 +1603,7 @@ def __init__(
  tsvd_unshifted: Optional[tsvd.Tsvd] = None,
  tsvd_shifted: Optional[tsvd.Tsvd] = None,
  square_norm: bool = False,
- picos_eps: float = 0,
+ picos_eps: float = 1e-6,
  solver_params: Optional[Dict[str, Any]] = None,
  ) -> None:
  """Instantiate :class:`LmiDmdcHinfReg`.
@@ -1924,7 +1924,7 @@ def __init__(
  iter_rtol: float = 0,
  inv_method: str = 'svd',
  tsvd: Optional[tsvd.Tsvd] = None,
- picos_eps: float = 0,
+ picos_eps: float = 1e-6,
  solver_params: Optional[Dict[str, Any]] = None,
  ) -> None:
  """Instantiate :class:`LmiEdmdDissipativityConstr`.

requirements.txt

@@ -1,7 +1,7 @@
 # Requirements to install ``pykoop`` and run examples
 numpy>=1.21.0
-scipy>=1.7.0
-scikit-learn>=1.0.0
+scipy>=1.8.0
+scikit-learn>=1.2.0
 PICOS>=2.4.0
 optht>=0.2.0
 Deprecated>=1.2.13

tests/test_koopman_pipeline.py

@@ -1097,6 +1097,95 @@ def test_strip_initial_conditons(self):
  np.testing.assert_allclose(X1s, X2s)
 
 
+@pytest.mark.filterwarnings(
+ 'ignore:Call to deprecated method predict_multistep')
+@pytest.mark.parametrize(
+ 'kp',
+ [
+ pykoop.KoopmanPipeline(
+ lifting_functions=None,
+ regressor=pykoop.Edmd(),
+ ),
+ pykoop.KoopmanPipeline(
+ lifting_functions=[
+ ('pl', pykoop.PolynomialLiftingFn(order=2)),
+ ],
+ regressor=pykoop.Edmd(),
+ ),
+ pykoop.KoopmanPipeline(
+ lifting_functions=[
+ ('dl',
+ pykoop.DelayLiftingFn(
+ n_delays_state=2,
+ n_delays_input=2,
+ )),
+ ],
+ regressor=pykoop.Edmd(),
+ ),
+ pykoop.KoopmanPipeline(
+ lifting_functions=[
+ ('sp',
+ pykoop.SplitPipeline(
+ lifting_functions_state=[
+ ('pl', pykoop.PolynomialLiftingFn(order=2)),
+ ],
+ lifting_functions_input=None,
+ ))
+ ],
+ regressor=pykoop.Edmd(),
+ ),
+ ],
+)
+class TestPredictionNoInput:
+ """Test fit Koopman pipeline prediction without input."""
+
+ def test_predict_trajectory_no_input(
+ self,
+ kp,
+ mass_spring_damper_no_input,
+ ):
+ """Test :func:`predict_trajectory` with no input."""
+ msg = 'Test only works when there is no episode feature.'
+ assert (not mass_spring_damper_no_input['episode_feature']), msg
+ # Fit estimator
+ kp.fit(
+ mass_spring_damper_no_input['X_train'],
+ n_inputs=mass_spring_damper_no_input['n_inputs'],
+ episode_feature=False,
+ )
+ # Extract initial conditions
+ x0 = pykoop.extract_initial_conditions(
+ mass_spring_damper_no_input['X_train'],
+ kp.min_samples_,
+ n_inputs=mass_spring_damper_no_input['n_inputs'],
+ episode_feature=False,
+ )
+ # Extract input
+ u = pykoop.extract_input(
+ mass_spring_damper_no_input['X_train'],
+ n_inputs=mass_spring_damper_no_input['n_inputs'],
+ episode_feature=False,
+ )
+ # Predict new states
+ X_sim = kp.predict_trajectory(
+ x0,
+ u,
+ episode_feature=False,
+ relift_state=True,
+ )
+ # Predict manually
+ X_sim_exp = np.zeros(X_sim.shape)
+ X_sim_exp[:kp.min_samples_, :] = x0
+ for k in range(kp.min_samples_, u.shape[0]):
+ X = np.hstack((
+ X_sim_exp[(k - kp.min_samples_):k, :],
+ u[(k - kp.min_samples_):k, :],
+ ))
+ Xp = kp.predict(X)
+ X_sim_exp[[k], :] = Xp[[-1], :]
+ np.testing.assert_allclose(X_sim, X_sim_exp)
+
+
 @pytest.mark.filterwarnings(
  'ignore:Call to deprecated method predict_multistep')
 @pytest.mark.parametrize(

tests/test_lmi_regressors.py

@@ -499,18 +499,38 @@ class TestSkLearn:
  """Test ``scikit-learn`` compatibility."""
 
  @sklearn.utils.estimator_checks.parametrize_with_checks([
- pykoop.lmi_regressors.LmiEdmd(alpha=1e-3, ),
- pykoop.lmi_regressors.LmiEdmdSpectralRadiusConstr(max_iter=1, ),
- pykoop.lmi_regressors.LmiEdmdHinfReg(alpha=1, ratio=1, max_iter=1),
- pykoop.lmi_regressors.LmiEdmdDissipativityConstr(max_iter=1, ),
- pykoop.lmi_regressors.LmiDmdc(alpha=1e-3, ),
- pykoop.lmi_regressors.LmiDmdcSpectralRadiusConstr(max_iter=1, ),
- pykoop.lmi_regressors.LmiDmdcHinfReg(alpha=1, ratio=1, max_iter=1),
+ pykoop.lmi_regressors.LmiEdmd(alpha=1e-3, picos_eps=0),
+ pykoop.lmi_regressors.LmiEdmdSpectralRadiusConstr(
+ max_iter=1,
+ picos_eps=0,
+ ),
+ pykoop.lmi_regressors.LmiEdmdHinfReg(
+ alpha=1,
+ ratio=1,
+ max_iter=1,
+ picos_eps=0,
+ ),
+ pykoop.lmi_regressors.LmiEdmdDissipativityConstr(
+ max_iter=1,
+ picos_eps=0,
+ ),
+ pykoop.lmi_regressors.LmiDmdc(alpha=1e-3, picos_eps=0),
+ pykoop.lmi_regressors.LmiDmdcSpectralRadiusConstr(
+ max_iter=1,
+ picos_eps=0,
+ ),
+ pykoop.lmi_regressors.LmiDmdcHinfReg(
+ alpha=1,
+ ratio=1,
+ max_iter=1,
+ picos_eps=0,
+ ),
  pykoop.lmi_regressors.LmiHinfZpkMeta(
  pykoop.lmi_regressors.LmiEdmdHinfReg(
  alpha=1,
  ratio=1,
  max_iter=1,
+ picos_eps=0,
  )),
  ])
  def test_compatible_estimator(self, estimator, check, mosek_solver_params):