Merge pull request #3402 from emma58/lp-dual

Adding (parameterized) linear programming dual transformation!

pyomo/contrib/fme/fourier_motzkin_elimination.py

@@ -30,6 +30,7 @@
 from pyomo.repn.standard_repn import generate_standard_repn
 from pyomo.common.collections import ComponentMap, ComponentSet
 from pyomo.opt import TerminationCondition
+from pyomo.util.config_domains import ComponentDataSet
 
 import logging
 
@@ -57,23 +58,6 @@ def _check_var_bounds_filter(constraint):
     return True
 
 
-def vars_to_eliminate_list(x):
-    if isinstance(x, (Var, VarData)):
-        if not x.is_indexed():
-            return ComponentSet([x])
-        ans = ComponentSet()
-        for j in x.index_set():
-            ans.add(x[j])
-        return ans
-    elif hasattr(x, '__iter__'):
-        ans = ComponentSet()
-        for i in x:
-            ans.update(vars_to_eliminate_list(i))
-        return ans
-    else:
-        raise ValueError("Expected Var or list of Vars.\n\tReceived %s" % type(x))
-
-
 def gcd(a, b):
     while b != 0:
         a, b = b, a % b
@@ -111,7 +95,7 @@ class Fourier_Motzkin_Elimination_Transformation(Transformation):
         'vars_to_eliminate',
         ConfigValue(
             default=None,
-            domain=vars_to_eliminate_list,
+            domain=ComponentDataSet(Var),
             description="Continuous variable or list of continuous variables to "
             "project out of the model",
             doc="""

pyomo/core/plugins/transform/__init__.py

@@ -22,4 +22,5 @@
     add_slack_vars,
     scaling,
     logical_to_linear,
+    lp_dual,
 )

pyomo/core/plugins/transform/lp_dual.py

@@ -0,0 +1,260 @@
+#  ___________________________________________________________________________
+#
+#  Pyomo: Python Optimization Modeling Objects
+#  Copyright (c) 2008-2024
+#  National Technology and Engineering Solutions of Sandia, LLC
+#  Under the terms of Contract DE-NA0003525 with National Technology and
+#  Engineering Solutions of Sandia, LLC, the U.S. Government retains certain
+#  rights in this software.
+#  This software is distributed under the 3-clause BSD License.
+#  ___________________________________________________________________________
+
+from pyomo.common.autoslots import AutoSlots
+from pyomo.common.collections import ComponentMap
+from pyomo.common.config import ConfigDict, ConfigValue
+from pyomo.common.dependencies import scipy
+from pyomo.core import (
+    ConcreteModel,
+    Block,
+    Var,
+    Constraint,
+    Objective,
+    TransformationFactory,
+    NonNegativeReals,
+    NonPositiveReals,
+    maximize,
+    minimize,
+    Reals,
+)
+from pyomo.opt import WriterFactory
+from pyomo.repn.standard_repn import isclose_const
+from pyomo.util.config_domains import ComponentDataSet
+
+
+class _LPDualData(AutoSlots.Mixin):
+    __slots__ = ('primal_var', 'dual_var', 'primal_constraint', 'dual_constraint')
+
+    def __init__(self):
+        self.primal_var = {}
+        self.dual_var = {}
+        self.primal_constraint = ComponentMap()
+        self.dual_constraint = ComponentMap()
+
+
+Block.register_private_data_initializer(_LPDualData)
+
+
+@TransformationFactory.register(
+    'core.lp_dual', 'Generate the linear programming dual of the given model'
+)
+class LinearProgrammingDual(object):
+    CONFIG = ConfigDict("core.lp_dual")
+    CONFIG.declare(
+        'parameterize_wrt',
+        ConfigValue(
+            default=None,
+            domain=ComponentDataSet(Var),
+            description="Vars to treat as data for the purposes of taking the dual",
+            doc="""
+            Optional list of Vars to be treated as data while taking the LP dual.
+
+            For example, if this is the dual of the inner problem in a multilevel
+            optimization problem, then the outer problem's Vars would be specified
+            in this list since they are not variables from the perspective of the 
+            inner problem.
+            """,
+        ),
+    )
+
+    def apply_to(self, model, **options):
+        raise NotImplementedError(
+            "The 'core.lp_dual' transformation does not implement "
+            "apply_to since it is ambiguous what it means to take a dual "
+            "in place. Please use 'create_using' and do what you wish with the "
+            "returned model."
+        )
+
+    def create_using(self, model, ostream=None, **kwds):
+        """Take linear programming dual of a model
+
+        Returns
+        -------
+        ConcreteModel containing linear programming dual
+
+        Parameters
+        ----------
+        model: ConcreteModel
+            The concrete Pyomo model to take the dual of
+
+        ostream: None
+            This is provided for API compatibility with other writers
+            and is ignored here.
+
+        """
+        config = self.CONFIG(kwds.pop('options', {}))
+        config.set_value(kwds)
+
+        if config.parameterize_wrt is None:
+            std_form = WriterFactory('compile_standard_form').write(
+                model, mixed_form=True, set_sense=None
+            )
+        else:
+            std_form = WriterFactory('compile_parameterized_standard_form').write(
+                model, wrt=config.parameterize_wrt, mixed_form=True, set_sense=None
+            )
+        return self._take_dual(model, std_form)
+
+    def _take_dual(self, model, std_form):
+        if len(std_form.objectives) != 1:
+            raise ValueError(
+                "Model '%s' has no objective or multiple active objectives. Can "
+                "only take dual with exactly one active objective!" % model.name
+            )
+        primal_sense = std_form.objectives[0].sense
+
+        dual = ConcreteModel(name="%s dual" % model.name)
+        # This is a csc matrix, so we'll skip transposing and just work off
+        # of the columns
+        A = std_form.A
+        c = std_form.c.todense().ravel()
+        dual_rows = range(A.shape[1])
+        dual_cols = range(A.shape[0])
+        dual.x = Var(dual_cols, domain=NonNegativeReals)
+        trans_info = dual.private_data()
+        for j, (primal_cons, ineq) in enumerate(std_form.rows):
+            # maximize is -1 and minimize is +1 and ineq is +1 for <= and -1 for
+            # >=, so we need to change domain to NonPositiveReals if the product
+            # of these is +1.
+            if primal_sense * ineq == 1:
+                dual.x[j].domain = NonPositiveReals
+            elif ineq == 0:
+                # equality
+                dual.x[j].domain = Reals
+            trans_info.primal_constraint[dual.x[j]] = primal_cons
+            trans_info.dual_var[primal_cons] = dual.x[j]
+
+        dual.constraints = Constraint(dual_rows)
+        for i, primal in enumerate(std_form.columns):
+            lhs = 0
+            for j in range(A.indptr[i], A.indptr[i + 1]):
+                coef = A.data[j]
+                primal_row = A.indices[j]
+                lhs += coef * dual.x[primal_row]
+
+            if primal.domain is Reals:
+                dual.constraints[i] = lhs == c[i]
+            elif primal_sense is minimize:
+                if primal.domain is NonNegativeReals:
+                    dual.constraints[i] = lhs <= c[i]
+                else:  # primal.domain is NonPositiveReals
+                    dual.constraints[i] = lhs >= c[i]
+            else:
+                if primal.domain is NonNegativeReals:
+                    dual.constraints[i] = lhs >= c[i]
+                else:  # primal.domain is NonPositiveReals
+                    dual.constraints[i] = lhs <= c[i]
+            trans_info.dual_constraint[primal] = dual.constraints[i]
+            trans_info.primal_var[dual.constraints[i]] = primal
+
+        dual.obj = Objective(
+            expr=sum(std_form.rhs[j] * dual.x[j] for j in dual_cols),
+            sense=-primal_sense,
+        )
+
+        return dual
+
+    def get_primal_constraint(self, model, dual_var):
+        """Return the primal constraint corresponding to 'dual_var'
+
+        Returns
+        -------
+        Constraint
+
+        Parameters
+        ----------
+        model: ConcreteModel
+            A dual model returned from the 'core.lp_dual' transformation
+        dual_var: Var
+            A dual variable on 'model'
+
+        """
+        primal_constraint = model.private_data().primal_constraint
+        if dual_var in primal_constraint:
+            return primal_constraint[dual_var]
+        else:
+            raise ValueError(
+                "It does not appear that Var '%s' is a dual variable on model '%s'"
+                % (dual_var.name, model.name)
+            )
+
+    def get_dual_constraint(self, model, primal_var):
+        """Return the dual constraint corresponding to 'primal_var'
+
+        Returns
+        -------
+        Constraint
+
+        Parameters
+        ----------
+        model: ConcreteModel
+            A primal model passed as an argument to the 'core.lp_dual' transformation
+        primal_var: Var
+            A primal variable on 'model'
+
+        """
+        dual_constraint = model.private_data().dual_constraint
+        if primal_var in dual_constraint:
+            return dual_constraint[primal_var]
+        else:
+            raise ValueError(
+                "It does not appear that Var '%s' is a primal variable on model '%s'"
+                % (primal_var.name, model.name)
+            )
+
+    def get_primal_var(self, model, dual_constraint):
+        """Return the primal variable corresponding to 'dual_constraint'
+
+        Returns
+        -------
+        Var
+
+        Parameters
+        ----------
+        model: ConcreteModel
+            A dual model returned from the 'core.lp_dual' transformation
+        dual_constraint: Constraint
+            A constraint on 'model'
+
+        """
+        primal_var = model.private_data().primal_var
+        if dual_constraint in primal_var:
+            return primal_var[dual_constraint]
+        else:
+            raise ValueError(
+                "It does not appear that Constraint '%s' is a dual constraint on "
+                "model '%s'" % (dual_constraint.name, model.name)
+            )
+
+    def get_dual_var(self, model, primal_constraint):
+        """Return the dual variable corresponding to 'primal_constraint'
+
+        Returns
+        -------
+        Var
+
+        Parameters
+        ----------
+        model: ConcreteModel
+            A primal model passed as an argument to the 'core.lp_dual' transformation
+        primal_constraint: Constraint
+            A constraint on 'model'
+
+        """
+        dual_var = model.private_data().dual_var
+        if primal_constraint in dual_var:
+            return dual_var[primal_constraint]
+        else:
+            raise ValueError(
+                "It does not appear that Constraint '%s' is a primal constraint on "
+                "model '%s'" % (primal_constraint.name, model.name)
+            )