Constraints Refactor 4: Introduce DiscreteLinearConstraint and Align with Conti Counterpart

CHANGELOG.md

@@ -22,8 +22,6 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
   number of required models instead of the number of transform outputs
 
 ### Added
-- `coefficients` attribute for `DiscreteSumConstraint`, enabling weighted sums. Follows
-  the same pattern as `ContinuousLinearConstraint.coefficients`
 - `simplex_coefficients` keyword argument to `SubspaceDiscrete.from_simplex` for
   weighted simplex sum constraints
 - `Symmetry` concept for expressing symmetries of the optimization problem, including
@@ -43,14 +41,17 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - `DiscreteRepetitionConstraint` for controlling value repetition across parameters
   via `n_max_repetitions` (replaces `DiscreteNoLabelDuplicatesConstraint` and
   `DiscreteLinkedParametersConstraint`)
+- `DiscreteLinearConstraint` for (optionally weighted) sum constraints on discrete
+  parameters, supporting `coefficients` and mirroring `ContinuousLinearConstraint`'s
+  `operator`/`rhs`/`coefficients` interface (replaces `DiscreteSumConstraint`)
 
 ### Changed
 - `BOTORCH` GP preset now includes `BetaPrior(2.5, 1.5)` for the task covariance
   kernel in multi-task scenarios, matching BoTorch's `MultiTaskGP` defaults introduced
   in version `0.18.0`
 - The `BOTORCH` GP preset now requires BoTorch `>= 0.18.0` and raises an
   `IncompatibilityError` if an older version is installed
-- `DiscreteSumConstraint`, `ContinuousLinearConstraint`, and
+- `DiscreteLinearConstraint`, `ContinuousLinearConstraint`, and
   `SubspaceDiscrete.from_simplex` now forbid 0 as coefficients
 - `SubspaceDiscrete.from_simplex` no longer requires non-negative parameter values
 - Bumped polars to `>=0.20.8`
@@ -62,6 +63,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Discrete filtering constraints now uniformly define what is **kept** in the search
   space; set `exclude=True` to invert and keep the complement instead
 - Renamed `exclusion_constraints` example to `selection_constraints`
+- `DiscreteProductConstraint` now uses `operator`/`rhs`/`tolerance` instead of
+  `condition`
 
 ### Deprecations
 - `DiscreteExcludeConstraint` in favor of
@@ -71,6 +74,9 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - `DiscreteLinkedParametersConstraint` in favor of
   `DiscreteRepetitionConstraint(..., n_max_repetitions=len(parameters)-1,
   exclude=True)`
+- `DiscreteSumConstraint` in favor of `DiscreteLinearConstraint`
+- `DiscreteProductConstraint(condition=ThresholdCondition(...))` in favor of
+  `DiscreteProductConstraint(operator=..., rhs=...)`
 
 ## [0.15.0] - 2026-06-11
 ### Breaking Changes

baybe/constraints/__init__.py

@@ -12,6 +12,7 @@
     DiscreteCustomConstraint,
     DiscreteDependenciesConstraint,
     DiscreteExcludeConstraint,
+    DiscreteLinearConstraint,
     DiscreteLinkedParametersConstraint,
     DiscreteNoLabelDuplicatesConstraint,
     DiscretePermutationInvarianceConstraint,
@@ -35,6 +36,7 @@
     "DiscreteCustomConstraint",
     "DiscreteDependenciesConstraint",
     "DiscreteExcludeConstraint",
+    "DiscreteLinearConstraint",
     "DiscreteLinkedParametersConstraint",
     "DiscreteNoLabelDuplicatesConstraint",
     "DiscretePermutationInvarianceConstraint",

baybe/constraints/continuous.py

@@ -50,7 +50,9 @@ class ContinuousLinearConstraint(ContinuousConstraint):
         converter=lambda x: cattrs.structure(x, tuple[float, ...]),
         validator=deep_iterable(member_validator=finite_float),
     )
-    """In-/equality coefficient for each entry in ``parameters``."""
+    """The coefficients for the weighted sum, one per entry in ``parameters``.
+
+    Defaults to all-ones, i.e. an unweighted sum."""
 
     rhs: float = field(default=0.0, converter=float, validator=finite_float)
     """Right-hand side value of the in-/equality."""