verbosity for mmphi, unique

Author: bartzbeielstein

pyproject.toml

@@ -7,7 +7,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "spotpython"
-version = "0.29.19"
+version = "0.29.20"
 authors = [
   { name="T. Bartz-Beielstein", email="tbb@bartzundbartz.de" }
 ]

src/spotpython/light/regression/nn_funnel_regressor.py

@@ -110,23 +110,22 @@ def __init__(
         if self.hparams.l1 < 8:
             raise ValueError("l1 must be at least 8")
 
-
         layers = []
         in_features = self._L_in
         hidden_size = self.hparams.l1
         output_dim = self._L_out
 
         for i in range(self.hparams.num_layers):
             out_features = max(hidden_size // 2, 8)  # Enforce minimum of 8 units
-            
+
             layers.append(nn.Linear(in_features, hidden_size))
-            
+
             if self.hparams.batch_norm:
                 layers.append(nn.BatchNorm1d(hidden_size))  # Add BatchNorm if enabled
-            
+
             layers.append(self.hparams.act_fn)
             layers.append(nn.Dropout(self.hparams.dropout_prob))
-            
+
             in_features = hidden_size
             hidden_size = out_features
 
@@ -264,7 +263,7 @@ def configure_optimizers(self) -> torch.optim.Optimizer:
         """
         # optimizer = torch.optim.Adam(self.parameters(), lr=self.learning_rate)
         optimizer = optimizer_handler(optimizer_name=self.hparams.optimizer, params=self.parameters(), lr_mult=self.hparams.lr_mult)
-        
+
         # If the lr_sched hyperparameter is set to True, we will use a learning rate scheduler.
         if self.hparams.lr_sched:
             num_milestones = 3  # Number of milestones to divide the epochs

src/spotpython/utils/sampling.py

@@ -311,7 +311,7 @@ def mm(X1: np.ndarray, X2: np.ndarray, p: Optional[float] = 1.0) -> int:
         return c[idx]
 
 
-def mmphi(X: np.ndarray, q: Optional[float] = 2.0, p: Optional[float] = 1.0) -> float:
+def mmphi(X: np.ndarray, q: Optional[float] = 2.0, p: Optional[float] = 1.0, verbosity=0) -> float:
     """
     Calculates the Morris-Mitchell sampling plan quality criterion.
 
@@ -324,6 +324,9 @@ def mmphi(X: np.ndarray, q: Optional[float] = 2.0, p: Optional[float] = 1.0) ->
         p (float, optional):
             The distance norm to use. For example, p=1 is Manhattan (L1),
             p=2 is Euclidean (L2). Defaults to 1.0.
+        verbosity (int, optional):
+            If set to 1, prints additional information about the computation.
+            Defaults to 0 (no additional output).
 
     Returns:
         float:
@@ -358,8 +361,15 @@ def mmphi(X: np.ndarray, q: Optional[float] = 2.0, p: Optional[float] = 1.0) ->
         >>> print(quality)
         # This value indicates how well points are spread out, with smaller being better.
     """
+    # check that X has unique rows
+    if X.shape[0] != len(np.unique(X, axis=0)):
+        # issue a warning if there are duplicate rows
+        print("Warning: X contains duplicate rows. This may affect the space-fillingness metric.")
+        # make X unique
+        X = np.unique(X, axis=0)
     # Compute the distance multiplicities: J, and unique distances: d
     J, d = jd(X, p)
+    print(f"J: {J}, d: {d}") if verbosity > 0 else None
 
     # Summation of J[i] * d[i]^(-q), then raised to 1/q
     # This follows the Morris-Mitchell definition.