User guide for cardinality constraints (#496)

This PR adds explanations for the `ContinuousCardinalityConstraint` and
the `DiscreteCardinalityConstraint` to the user guide.

Author: AdrianSosic

CHANGELOG.md

@@ -32,6 +32,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Utilities `activate_parameter` and `is_cardinality_fulfilled` for enforcing and
   validating cardinality constraints
 - Utility `is_inactive` for determining if parameters are inactive
+- Cardinality constraints sections to the user guide
 
 ### Changed
 - Acquisition function indicator `is_mc` has been removed in favor of new indicators 

docs/userguide/constraints.md

@@ -79,6 +79,46 @@ ContinuousLinearConstraint(
 A more detailed example can be found
 [here](../../examples/Constraints_Continuous/linear_constraints).
 
+### ContinuousCardinalityConstraint
+The {class}`~baybe.constraints.continuous.ContinuousCardinalityConstraint` gives you a
+tool to control the number of active factors (i.e. parameters that take a non-zero
+value) in your designs. This comes handy, for example, when designing mixtures with a
+limited number of components.
+
+To create a constraint of this kind, simply specify the set of parameters on which the
+constraint is to be imposed, together with the corresponding upper and lower cardinality
+limits. For instance, the following constraint would ensure that there is always a
+minimum of one and a maximum of two components in each parameter configuration:
+
+```python
+from baybe.constraints import ContinuousCardinalityConstraint
+
+ContinuousCardinalityConstraint(
+    parameters=["x_1", "x_2", "x_3"],
+    min_cardinality=1,  # defaults to 0
+    max_cardinality=2,  # defaults to the number of affected parameters (here: 3)
+    relative_threshold=0.001,  # optional, defines the range of values considered active
+)
+```
+
+```{admonition} Computational Challenges
+:class: warning
+
+Note that, compared to optimization with [discrete cardinality
+constraints](#DiscreteCardinalityConstraint), finding optimal cardinality-constrained
+solutions in continuous spaces is significantly more challenging due to the
+fractured nature of the resulting space. For larger parameter sets or complex constraint
+settings, searching an optimal parameter configuration can quickly become infeasible,
+creating the need for approximation schemes:
+
+* The
+  {paramref}`BotorchRecommender.max_n_subspaces <baybe.recommenders.pure.bayesian.botorch.BotorchRecommender.max_n_subspaces>`
+  attribute can be used to limit the number of subspaces considered during optimization.
+* When the ranges of cardinality-constrained parameters cover both positive and negative
+  values, minimal cardinality requirements cannot always be guaranteed, potentially
+  resulting in a {class}`~baybe.exceptions.MinimumCardinalityViolatedWarning`.
+```
+
 ## Conditions
 Conditions are elements used within discrete constraints.
 While discrete constraints can operate on one or multiple parameters, a condition
@@ -369,6 +409,20 @@ DiscretePermutationInvarianceConstraint(
 The usage of `DiscretePermutationInvarianceConstraint` is also part of the
 [example on slot-based mixtures](../../examples/Mixtures/slot_based).
 
+### DiscreteCardinalityConstraint
+Like its [continuous cousin](#ContinuousCardinalityConstraint), the
+{class}`~baybe.constraints.discrete.DiscreteCardinalityConstraint` lets you control the
+number of active parameters in your design. The construction works analogously: 
+```python
+from baybe.constraints import DiscreteCardinalityConstraint
+
+DiscreteCardinalityConstraint(
+    parameters=["Fraction_1", "Fraction_2", "Fraction_3"],
+    min_cardinality=1,  # defaults to 0
+    max_cardinality=2,  # defaults to the number of affected parameters (here: 3)
+)
+```
+
 ### DiscreteCustomConstraint
 With a [`DiscreteCustomConstraint`](baybe.constraints.discrete.DiscreteCustomConstraint) 
 constraint, you can specify a completely custom filter: