docs: add _BundleConditionMixin in documentation

docs/api.rst

@@ -21,6 +21,7 @@ API Reference
     :show-inheritance:
     :inherited-members:
     :members:
+    :private-members:
 
 `neurodiffeq.solvers`
 ------------------------------------------------------

neurodiffeq/conditions.py

@@ -76,24 +76,25 @@ def set_impose_on(self, ith_unit):
 
 
 class _BundleConditionMixin:
+    """Mixin class for bundle conditions. Unlike other conditions, parameters of a bundle conditions are not fixed.
+    For example, a regular `neurodiffeq.conditions.IVP` instance only allows fixed t_0 and u_0. But a bundle
+    condition allows for t_0 and/or u_0 to be dynamically sampled by generators during training. One can use this to
+    learn a bundle of solutions (as a function of t_0 and/or u_0).
+
+    :param bundle_param_lookup:
+        A dictionary that maps bundle parameter name to its corresponding index.
+        For example, if bundle_param_lookup == {'t_0': 0, 'u_0': 1}, then the `thetas` passed to self.enforce()`
+        must contain two entries, with thetas[0] being a tensor of `t_0` and thetas[1] being a tensor of `u_0`.
+        Similarly, if bundle_param_lookup == {'t_0': 1, 'u_0': 0}, then the two entries in thetas should be
+        swapped. Defaults to empty dictionary.
+    :type bundle_param_lookup: Dict[str, int]
+    :param allowed_params:
+        A collection of parameter names allowed in ``bundle_param_lookup``. If specified, an error will be raised
+        if ``bundle_param_lookup`` contains names that do not appear in ``allowed_params``.
+    :type allowed_params: Set[str] or List[str] or Tuple[str]
+    """
+
     def __init__(self, bundle_param_lookup=None, allowed_params=None):
-        """Mixin class for bundle conditions. Unlike other conditions, parameters of a bundle conditions are not fixed.
-        For example, a regular `neurodiffeq.conditions.IVP` instance only allows fixed t_0 and u_0. But a bundle
-        condition allows for t_0 and/or u_0 to be dynamically sampled by generators during training. One can use this to
-        learn a bundle of solutions (as a function of t_0 and/or u_0).
-
-        :param bundle_param_lookup:
-            A dictionary that maps bundle parameter name to its corresponding index.
-            For example, if bundle_param_lookup == {'t_0': 0, 'u_0': 1}, then the `thetas` passed to self.enforce()`
-            must contain two entries, with thetas[0] being a tensor of `t_0` and thetas[1] being a tensor of `u_0`.
-            Similarly, if bundle_param_lookup == {'t_0': 1, 'u_0': 0}, then the two entries in thetas should be
-            swapped. Defaults to empty dictionary.
-        :type bundle_param_lookup: Dict[str, int]
-        :param allowed_params:
-            A collection of parameter names allowed in ``bundle_param_lookup``. If specified, an error will be raised
-            if ``bundle_param_lookup`` contains names that do not appear in ``allowed_params``.
-        :type allowed_params: Set[str] or List[str] or Tuple[str]
-        """
         self.bundle_param_lookup = bundle_param_lookup or {}
 
         if isinstance(allowed_params, str):
@@ -113,8 +114,7 @@ def _get_parameter(self, param_name, thetas, override_name=None):
 
         Example:
         - If param_name == 't_0', and self.bundle_param_lookup['t_0'] == 0, the method will return `thetas[0]`.
-        - If `t_0` is not present in self.bundle_param_lookup, it will return `self.t_0`, or, self.some_other_attribute
-          if override_name is set to 'some_other_attribute'.
+        - If `t_0` is not present in self.bundle_param_lookup, it will return `self.t_0`, or, self.some_other_attribute if override_name is set to 'some_other_attribute'.
 
         :param param_name: Name of parameter to be used. E.g. `t_0`, `u_0` in the case of an initial value problem.
         :type param_name: str