Add where to flow (#126)

src/stream_ml/pytorch/builtin/compat/flow.py

@@ -10,7 +10,7 @@
 
 import torch as xp
 
-from stream_ml.core.params.scaler import scale_params
+from stream_ml.core.builtin._utils import WhereRequiredError
 
 from stream_ml.pytorch._base import ModelBase
 from stream_ml.pytorch.utils import names_intersect
@@ -30,7 +30,13 @@ class FlowModel(ModelBase):
     with_grad: bool = True
 
     def ln_likelihood(
-        self, mpars: Params[Array], /, data: Data[Array], **kwargs: Array
+        self,
+        mpars: Params[Array],
+        /,
+        data: Data[Array],
+        *,
+        where: Data[Array] | None = None,
+        **kwargs: Array,
     ) -> Array:
         """Log-likelihood of the array.
 
@@ -43,26 +49,47 @@ def ln_likelihood(
         data : Data[Array]
             Data (phi1, phi2).
 
+        where : Data[Array[(N,), bool]] | None, optional keyword-only
+            Where to evaluate the log-likelihood. If not provided, then the
+            log-likelihood is evaluated at all data points. ``where`` must
+            contain the fields in ``coord_names``. Each field must be a boolean
+            array of the same length as `data`. `True` indicates that the data
+            point is available, and `False` indicates that the data point is not
+            available.
+
         **kwargs : Array
             Additional arguments.
 
         Returns
         -------
         Array
         """
-        # TODO: support `where` argument.
+        # 'where' is used to indicate which data points are available. If
+        # 'where' is not provided, then all data points are assumed to be
+        # available.
+        where_: Array  # (N, F)
+        if where is not None:
+            where_ = where[self.coord_names].array
+        elif self.require_where:
+            raise WhereRequiredError
+        else:
+            where_ = self.xp.ones((len(data), self.ndim), dtype=bool)
+        idx = where_.all(axis=1)
+        # TODO: allow for missing data in only some of the dimensions
+
         data = self.data_scaler.transform(
             data, names=names_intersect(data, self.data_scaler), xp=self.xp
         )
-        mpars = scale_params(self, mpars)
 
+        out = self.xp.zeros(len(data), dtype=data.dtype)
         with nullcontext() if self.with_grad else xp.no_grad():
-            return self.jacobian_logdet + self.net.log_prob(
-                inputs=data[self.coord_names].array,
-                context=data[self.indep_coord_names].array
+            out[idx] = self.jacobian_logdet + self.net.log_prob(
+                inputs=data[self.coord_names].array[idx],
+                context=data[self.indep_coord_names].array[idx]
                 if self.indep_coord_names is not None
                 else None,
             )
+        return out
 
     def forward(self, data: Data[Array]) -> Array:
         """Forward pass.