Tranquilo Refactoring (#445)

src/estimagic/optimization/tranquilo/aggregate_models.py

@@ -110,12 +110,13 @@ def aggregator_identity(vector_model):
     2. model_type: quadratic
 
     """
+    n_params = vector_model.linear_terms.size
     intercept = float(vector_model.intercepts)
-    linear_terms = np.squeeze(vector_model.linear_terms)
+    linear_terms = vector_model.linear_terms.flatten()
     if vector_model.square_terms is None:
-        square_terms = np.zeros((len(linear_terms), len(linear_terms)))
+        square_terms = np.zeros((n_params, n_params))
     else:
-        square_terms = np.squeeze(vector_model.square_terms)
+        square_terms = vector_model.square_terms.reshape(n_params, n_params)
     return intercept, linear_terms, square_terms
 
 

src/estimagic/optimization/tranquilo/geometry.py

@@ -1,85 +1,5 @@
-from functools import partial
-
 import numpy as np
 
-from estimagic.optimization.tranquilo.region import Region
-from estimagic.optimization.tranquilo.sample_points import get_sampler
-
-
-def get_geometry_checker_pair(
-    checker, reference_sampler, n_params, n_simulations=200, bounds=None
-):
-    """Get a geometry checker.
-
-    Args:
-        checker (str or Dict[callable]): Name of a geometry checker method or a
-            dictionary with entries 'quality_calculator' and 'cutoff_simulator'.
-            - 'quality_calculator': A callable that takes as argument a sample and
-            returns a measure on the quality of the geometry of the sample.
-            - 'cutoff_simulator': A callable that takes as argument 'n_samples',
-            'n_params', 'reference_sampler' and 'rng'.
-        reference_sampler (str): Either "box" or "ball", corresponding to comparison
-            samples drawn inside a box or a ball, respectively.
-        n_params (int): Number of parameters.
-        n_simulations (int): Number of simulations for the mean calculation.
-        bounds (Bounds): The parameter bounds. See module bounds.py.
-
-    Returns:
-        callable: The sample quality calculator.
-        callable: The quality cutoff simulator.
-
-    """
-    if reference_sampler not in {"box", "ball"}:
-        raise ValueError("reference_sampler need to be either 'box' or 'ball'.")
-
-    built_in_checker = {
-        "d_optimality": {
-            "quality_calculator": log_d_quality_calculator,
-            "cutoff_simulator": log_d_cutoff_simulator,
-        },
-    }
-
-    _checker = built_in_checker[checker]
-
-    quality_calculator = _checker["quality_calculator"]
-    cutoff_simulator = partial(
-        _checker["cutoff_simulator"],
-        reference_sampler=reference_sampler,
-        bounds=bounds,
-        n_params=n_params,
-        n_simulations=n_simulations,
-    )
-    return quality_calculator, cutoff_simulator
-
-
-def log_d_cutoff_simulator(
-    n_samples, rng, reference_sampler, bounds, n_params, n_simulations
-):
-    """Simulate the mean logarithm of the d-optimality criterion.
-
-    Args:
-        n_samples (int): Size of the sample.
-        rng (np.random.Generator): The random number generator.
-        reference_sampler (str): Either "box" or "ball", corresponding to comparison
-            samples drawn inside a box or a ball, respectively.
-        bounds (Bounds): The parameter bounds. See module bounds.py.
-        n_params (int): Dimensionality of the sample.
-        n_simulations (int): Number of simulations for the mean calculation.
-
-    Returns:
-        float: The simulated mean logarithm of the d-optimality criterion.
-
-    """
-    _sampler = get_sampler(reference_sampler)
-    trustregion = Region(center=np.zeros(n_params), radius=1.0, bounds=bounds)
-    sampler = partial(_sampler, trustregion=trustregion)
-    raw = []
-    for _ in range(n_simulations):
-        x = sampler(n_points=n_samples, rng=rng)
-        raw.append(log_d_quality_calculator(x, trustregion))
-    out = np.nanmean(raw)
-    return out
-
 
 def log_d_quality_calculator(sample, trustregion):
     """Logarithm of the d-optimality criterion.

src/estimagic/optimization/tranquilo/region.py

@@ -55,7 +55,10 @@ def map_from_unit(self, x: np.ndarray) -> np.ndarray:
         if self.shape == "sphere":
             out = _map_from_unit_sphere(x, center=self.center, radius=self.radius)
         else:
-            out = _map_from_unit_cube(x, cube_bounds=self.cube_bounds)
+            cube_bounds = self.cube_bounds
+            out = _map_from_unit_cube(x, cube_bounds=cube_bounds)
+            # Bounds may not be satisfied exactly due to numerical inaccuracies.
+            out = np.clip(out, cube_bounds.lower, cube_bounds.upper)
         return out
 
     # make it behave like a NamedTuple