Merge pull request #240 from lpsinger/astropy-modeling-units

Move input unit conversion and bounds checking to base class

CHANGES.rst

@@ -1,7 +1,16 @@
 1.6.dev (unreleased)
 ================
 
-- none yet
+- Refactor model input unit conversion and wavenumber bounds checking:
+
+  - The models' evaluate methods are no longer meant to be called directly.
+
+  - The BaseExtModel class now overrides the prepare_inputs method so that
+    unit conversion and bounds checking is done prior to calling the subclass'
+    evaluate method. Subclasses are no longer responsible for doing the unit
+    conversion and bounds checking themselves.
+
+  - Use Astropy's built-in input model units handling.
 
 1.5 (2024-08-16)
 ================

docs/dust_extinction/dev_model.rst

@@ -16,7 +16,7 @@ All
 All dust extinction models have at least the following:
 
 * A member variable `x_range` that that define the valid range of wavelengths. These are defined in inverse microns as is common for extinction curve research.
-* A member function `evaluate` that computes the extinction at a given `x` and any model parameter values.  The `x` values are checked to be within the valid `x_range`. The `x` values should have astropy.units.  If they do not, then they are assumed to be in inverse microns and a warning is issued stating such.
+* A member function `evaluate` that computes the extinction at a given `x` and any model parameter values.  The `x` values are checked to be within the valid `x_range`. The `x` values passed to the `evaluate` method have no units; the base class `BaseExtModel` will automatically convert whatever units the user provided to inverse microns prior to calling the `evaulate` method. The `evaluate` method should not be called directly.
 
 All of these classes used in ``dust_extinction`` are based on the 
 `Model <https://docs.astropy.org/en/stable/modeling/>`_ astropy.modeling class.