Distinguish params from class attributes (only in_4d and ensemble). Could improve documentation for class attributes or just make them non-public since user shouldn't need them

Author: kysolvik

dabench/dacycler/_dacycler.py

@@ -22,10 +22,6 @@ class DACycler():
         system_dim: System dimension
         delta_t: The timestep of the model (assumed uniform)
         model_obj: Forecast model object.
-        in_4d: True for 4D data assimilation techniques (e.g. 4DVar).
-            Default is False.
-        ensemble: True for ensemble-based data assimilation techniques
-            (ETKF). Default is False
         B: Initial / static background error covariance. Shape:
             (system_dim, system_dim). If not provided, will be calculated
             automatically.
@@ -36,14 +32,19 @@ class DACycler():
             If not provided will be calculated automatically.
         h: Optional observation operator as function. More flexible
             (allows for more complex observation operator). Default is None.
+    
+    Attributes:
+        in_4d: True for 4D data assimilation techniques (e.g. 4DVar).
+        ensemble: True for ensemble-based data assimilation techniques
+            (ETKF).
     """
+    in_4d = False
+    ensemble = False
 
     def __init__(self,
                  system_dim: int,
                  delta_t: float,
                  model_obj: Model,
-                 in_4d: bool = False,
-                 ensemble: bool = False,
                  B: ArrayLike | None = None,
                  R: ArrayLike | None = None,
                  H: ArrayLike | None = None,
@@ -54,8 +55,6 @@ def __init__(self,
         self.H = H
         self.R = R
         self.B = B
-        self.in_4d = in_4d
-        self.ensemble = ensemble
         self.system_dim = system_dim
         self.delta_t = delta_t
         self.model_obj = model_obj

dabench/dacycler/_etkf.py

@@ -19,7 +19,7 @@
 class ETKF(dacycler.DACycler):
     """Class for building ETKF DA Cycler
 
-    Attributes:
+    Args:
         system_dim: System dimension.
         delta_t: The timestep of the model (assumed uniform)
         model_obj: Forecast model object.
@@ -38,6 +38,8 @@ class ETKF(dacycler.DACycler):
         multiplicative_inflation: Scaling factor by which to multiply ensemble
             deviation. Default is 1.0 (no inflation).
     """
+    in_4d = False
+    ensemble = True
 
     def __init__(self,
                  system_dim: int,
@@ -57,8 +59,6 @@ def __init__(self,
         super().__init__(system_dim=system_dim,
                          delta_t=delta_t,
                          model_obj=model_obj,
-                         in_4d=False,
-                         ensemble=True,
                          B=B, R=R, H=H, h=h)
 
     def _step_forecast(self,

dabench/dacycler/_var3d.py

@@ -18,7 +18,7 @@
 class Var3D(dacycler.DACycler):
     """Class for building 3DVar DA Cycler
 
-    Attributes:
+    Args:
         system_dim: System dimension.
         delta_t: The timestep of the model (assumed uniform)
         model_obj: Forecast model object.
@@ -33,6 +33,8 @@ class Var3D(dacycler.DACycler):
         h: Optional observation operator as function. More flexible
             (allows for more complex observation operator). Default is None.
         """
+    in_4d = False
+    ensemble = False
 
     def __init__(self,
                  system_dim: int,

dabench/dacycler/_var4d.py

@@ -27,14 +27,10 @@
 class Var4D(dacycler.DACycler):
     """Class for building 4D DA Cycler
 
-    Attributes:
+    Args:
         system_dim: System dimension.
         delta_t: The timestep of the model (assumed uniform)
         model_obj: Forecast model object.
-        in_4d: True for 4D data assimilation techniques (e.g. 4DVar).
-            Always True for Var4D.
-        ensemble: True for ensemble-based data assimilation techniques
-            (ETKF). Always False for Var4D.
         B: Initial / static background error covariance. Shape:
             (system_dim, system_dim). If not provided, will be calculated
             automatically.
@@ -59,6 +55,8 @@ class Var4D(dacycler.DACycler):
             [0, 1, 2, 3, 4, 5]. If None (default), will calculate
             automatically.
     """
+    in_4d = True
+    ensemble = False
 
     def __init__(self,
                  system_dim: int,

dabench/dacycler/_var4d_backprop.py

@@ -27,14 +27,10 @@
 class Var4DBackprop(dacycler.DACycler):
     """Class for building Backpropagation 4D DA Cycler
 
-    Attributes:
+    Args:
         system_dim: System dimension.
         delta_t: The timestep of the model (assumed uniform)
         model_obj: Forecast model object.
-        in_4d: True for 4D data assimilation techniques (e.g. 4DVar).
-            Always True for Var4DBackprop.
-        ensemble: True for ensemble-based data assimilation techniques
-            (ETKF). Always False for Var4DBackprop.
         B: Initial / static background error covariance. Shape:
             (system_dim, system_dim). If not provided, will be calculated
             automatically.
@@ -65,6 +61,8 @@ class Var4DBackprop(dacycler.DACycler):
             return an error. This prevents it from hanging indefinitely
             when loss grows exponentionally. Default is 10.
     """
+    in_4d = True
+    ensemble = False
 
     def __init__(self,
                  system_dim: int,