def _generate_model(self):
self.model = ConcreteModel()
model = self.model
model._name = self.description
model.x = Var(within=Binary)
model.y = Var(within=Binary)
model.z = Var(within=Binary)
model.obj = Objective(expr=model.x, sense=maximize)
model.c0 = Constraint(expr=model.x + model.y + model.z == 1)
model.qc0 = Constraint(expr=model.x**2 + model.y**2 <= model.z**2)
model.qc1 = Constraint(expr=model.x**2 <= model.y * model.z)
def makeBetweenStepsPaperExample():
"""Original example model, implicit disjunction"""
m = ConcreteModel()
m.I = RangeSet(1, 4)
m.x = Var(m.I, bounds=(-2, 6))
m.disjunction = Disjunction(expr=[[sum(
m.x[i]**2
for i in m.I) <= 1], [sum((3 - m.x[i])**2 for i in m.I) <= 1]])
m.obj = Objective(expr=m.x[2] - m.x[1], sense=maximize)
return m
def back_off_constraint_with_calculated_cut_violation(cut, transBlock_rHull,
bigm_to_hull_map, opt,
stream_solver, TOL):
"""Calculates the maximum violation of cut subject to the relaxed hull
constraints. Increases this violation by TOL (to account for optimality
tolerance in solving the problem), and, if it finds that cut can be violated
up to this tolerance, makes it more conservative such that it no longer can.
Parameters
----------
cut: The cut to be made more conservative, a Constraint
transBlock_rHull: the relaxed hull model's transformation Block
bigm_to_hull_map: Dictionary mapping ids of bigM variables to the
corresponding variables on the relaxed hull instance
opt: SolverFactory object for solving the maximum violation problem
stream_solver: Whether or not to set tee=True while solving the maximum
violation problem.
TOL: An absolute tolerance to be added to the calculated cut violation,
to account for optimality tolerance in the maximum violation problem
solve.
"""
instance_rHull = transBlock_rHull.model()
logger.info("Post-processing cut: %s" % cut.expr)
# Take a constraint. We will solve a problem maximizing its violation
# subject to rHull. We will add some user-specified tolerance to that
# violation, and then add that much padding to it if it can be violated.
transBlock_rHull.separation_objective.deactivate()
transBlock_rHull.infeasibility_objective = Objective(
expr=clone_without_expression_components(cut.body,
substitute=bigm_to_hull_map))
results = opt.solve(instance_rHull, tee=stream_solver)
if verify_successful_solve(results) is not NORMAL:
logger.warning("Problem to determine how much to "
"back off the new cut "
"did not solve normally. Leaving the constraint as is, "
"which could lead to numerical trouble%s" % (results,))
# restore the objective
transBlock_rHull.del_component(transBlock_rHull.infeasibility_objective)
transBlock_rHull.separation_objective.activate()
return
# we're minimizing, val is <= 0
val = value(transBlock_rHull.infeasibility_objective) - TOL
if val <= 0:
logger.info("\tBacking off cut by %s" % val)
cut._body += abs(val)
# else there is nothing to do: restore the objective
transBlock_rHull.del_component(transBlock_rHull.infeasibility_objective)
transBlock_rHull.separation_objective.activate()
def solve_LOA_master(solve_data, config):
"""Solve the augmented lagrangean outer approximation master problem."""
m = solve_data.linear_GDP.clone()
GDPopt = m.GDPopt_utils
# Set up augmented Lagrangean penalty objective
GDPopt.objective.deactivate()
sign_adjust = 1 if GDPopt.objective.sense == minimize else -1
GDPopt.OA_penalty_expr = Expression(
expr=sign_adjust * config.OA_penalty_factor *
sum(v for v in m.component_data_objects(ctype=Var,
descend_into=(Block, Disjunct))
if v.parent_component().local_name == 'GDPopt_OA_slacks'))
GDPopt.oa_obj = Objective(expr=GDPopt.objective.expr +
GDPopt.OA_penalty_expr,
sense=GDPopt.objective.sense)
solve_data.mip_iteration += 1
mip_results = solve_linear_GDP(m, solve_data, config)
if mip_results:
if GDPopt.objective.sense == minimize:
solve_data.LB = max(value(GDPopt.oa_obj.expr), solve_data.LB)
else:
solve_data.UB = min(value(GDPopt.oa_obj.expr), solve_data.UB)
solve_data.iteration_log[(solve_data.master_iteration,
solve_data.mip_iteration,
solve_data.nlp_iteration)] = (
value(GDPopt.oa_obj.expr),
value(GDPopt.objective.expr),
mip_results[1] # mip_var_values
)
config.logger.info(
'ITER %s.%s.%s-MIP: OBJ: %s LB: %s UB: %s' %
(solve_data.master_iteration, solve_data.mip_iteration,
solve_data.nlp_iteration, value(
GDPopt.oa_obj.expr), solve_data.LB, solve_data.UB))
else:
# Master problem was infeasible.
if solve_data.master_iteration == 1:
config.logger.warning(
'GDPopt initialization may have generated poor '
'quality cuts.')
# set optimistic bound to infinity
if GDPopt.objective.sense == minimize:
solve_data.LB = float('inf')
else:
solve_data.UB = float('-inf')
# Call the MILP post-solve callback
config.master_postsolve(m, solve_data)
return mip_results
def solve_NLP_feas(solve_data, config):
"""Solves feasibility NLP and copies result to working model
Returns: Result values and dual values
"""
fixed_nlp = solve_data.working_model.clone()
add_feas_slacks(fixed_nlp)
MindtPy = fixed_nlp.MindtPy_utils
next(fixed_nlp.component_data_objects(Objective, active=True)).deactivate()
for constr in fixed_nlp.component_data_objects(
ctype=Constraint, active=True, descend_into=True):
if constr.body.polynomial_degree() not in [0, 1]:
constr.deactivate()
MindtPy.MindtPy_feas.activate()
MindtPy.MindtPy_feas_obj = Objective(
expr=sum(s for s in MindtPy.MindtPy_feas.slack_var[...]),
sense=minimize)
TransformationFactory('core.fix_integer_vars').apply_to(fixed_nlp)
with SuppressInfeasibleWarning():
feas_soln = SolverFactory(config.nlp_solver).solve(
fixed_nlp, **config.nlp_solver_args)
subprob_terminate_cond = feas_soln.solver.termination_condition
if subprob_terminate_cond is tc.optimal:
copy_var_list_values(
MindtPy.variable_list,
solve_data.working_model.MindtPy_utils.variable_list,
config)
elif subprob_terminate_cond is tc.infeasible:
raise ValueError('Feasibility NLP infeasible. '
'This should never happen.')
else:
raise ValueError(
'MindtPy unable to handle feasibility NLP termination condition '
'of {}'.format(subprob_terminate_cond))
var_values = [v.value for v in MindtPy.variable_list]
duals = [0 for _ in MindtPy.constraint_list]
for i, c in enumerate(MindtPy.constraint_list):
rhs = c.upper if c. has_ub() else c.lower
c_geq = -1 if c.has_ub() else 1
duals[i] = c_geq * max(
0, c_geq * (rhs - value(c.body)))
if value(MindtPy.MindtPy_feas_obj.expr) == 0:
raise ValueError(
'Problem is not feasible, check NLP solver')
return fixed_nlp, feas_soln
def obbt_disjunct(orig_model, idx, solver):
model = orig_model.clone()
# Fix the disjunct to be active
disjunct = model._disjuncts_to_process[idx]
disjunct.indicator_var.fix(1)
for obj in model.component_data_objects(Objective, active=True):
obj.deactivate()
# Deactivate nonlinear constraints
for constr in model.component_data_objects(Constraint,
active=True,
descend_into=(Block, Disjunct)):
if constr.body.polynomial_degree() not in linear_degrees:
constr.deactivate()
# Only look at the variables participating in active constraints within the scope
relevant_var_set = ComponentSet()
for constr in disjunct.component_data_objects(Constraint, active=True):
relevant_var_set.update(
identify_variables(constr.body, include_fixed=False))
TransformationFactory('gdp.bigm').apply_to(model)
model._var_bounding_obj = Objective(expr=1, sense=minimize)
for var in relevant_var_set:
model._var_bounding_obj.set_value(expr=var)
var_lb = solve_bounding_problem(model, solver)
if var_lb is None:
return None # bounding problem infeasible
model._var_bounding_obj.set_value(expr=-var)
var_ub = solve_bounding_problem(model, solver)
if var_ub is None:
return None # bounding problem infeasible
else:
var_ub = -var_ub # sign correction
var.setlb(var_lb)
var.setub(var_ub)
# Maps original variable --> (new computed LB, new computed UB)
var_bnds = ComponentMap(
((orig_var, (clone_var.lb if clone_var.has_lb() else -inf,
clone_var.ub if clone_var.has_ub() else inf))
for orig_var, clone_var in zip(orig_model._disj_bnds_linear_vars,
model._disj_bnds_linear_vars)
if clone_var in relevant_var_set))
return var_bnds
def twoSegments_SawayaGrossmann():
m = ConcreteModel()
m.x = Var(bounds=(0, 3))
m.disj1 = Disjunct()
m.disj1.c = Constraint(expr=inequality(0, m.x, 1))
m.disj2 = Disjunct()
m.disj2.c = Constraint(expr=inequality(2, m.x, 3))
m.disjunction = Disjunction(expr=[m.disj1, m.disj2])
# this is my objective because I want to make sure that when I am testing
# cutting planes, my first solution to rBigM is not on the convex hull.
m.obj = Objective(expr=m.x - m.disj2.indicator_var)
return m
def twoDisj_twoCircles_easy():
m = ConcreteModel()
m.x = Var(bounds=(0,8))
m.y = Var(bounds=(0,10))
m.upper_circle = Disjunct()
m.upper_circle.cons = Constraint(expr=(m.x - 1)**2 + (m.y - 6)**2 <= 2)
m.lower_circle = Disjunct()
m.lower_circle.cons = Constraint(expr=(m.x - 4)**2 + (m.y - 2)**2 <= 2)
m.disjunction = Disjunction(expr=[m.upper_circle, m.lower_circle])
m.obj = Objective(expr=m.x + m.y, sense=maximize)
return m
def knapsack(weights: List[Union[int, float]], values: List[Union[int, float]], max_weight: Union[int, float]):
I = list(range(len(weights)))
model = ConcreteModel()
model.x = Var(I, within=Binary)
model.objective = Objective(expr=sum(model.x[i] * values[i] for i in I), sense=maximize)
model.constraint = Constraint(expr=sum(model.x[i] * weights[i] for i in I) <= max_weight)
solver = solver_factory.get_solver(Solver.BONMIN)
solver.solve(model)
for v in model.component_data_objects(Var):
print(str(v), v.value)
def solve_NLP_feas(solve_data, config):
m = solve_data.working_model.clone()
add_feas_slacks(m)
MindtPy = m.MindtPy_utils
next(m.component_data_objects(Objective, active=True)).deactivate()
for constr in m.component_data_objects(ctype=Constraint,
active=True,
descend_into=True):
constr.deactivate()
MindtPy.MindtPy_feas.activate()
MindtPy.MindtPy_feas_obj = Objective(expr=sum(
s for s in MindtPy.MindtPy_feas.slack_var[...]),
sense=minimize)
for v in MindtPy.variable_list:
if v.is_binary():
v.fix(int(round(v.value)))
# m.pprint() #print nlp feasibility problem for debugging
with SuppressInfeasibleWarning():
feas_soln = SolverFactory(config.nlp_solver).solve(
m, **config.nlp_solver_args)
subprob_terminate_cond = feas_soln.solver.termination_condition
if subprob_terminate_cond is tc.optimal:
copy_var_list_values(
MindtPy.variable_list,
solve_data.working_model.MindtPy_utils.variable_list, config)
pass
elif subprob_terminate_cond is tc.infeasible:
raise ValueError('Feasibility NLP infeasible. '
'This should never happen.')
else:
raise ValueError(
'MindtPy unable to handle feasibility NLP termination condition '
'of {}'.format(subprob_terminate_cond))
var_values = [v.value for v in MindtPy.variable_list]
duals = [0 for _ in MindtPy.constraint_list]
for i, constr in enumerate(MindtPy.constraint_list):
# TODO rhs only works if constr.upper and constr.lower do not both have values.
# Sometimes you might have 1 <= expr <= 1. This would give an incorrect rhs of 2.
rhs = ((0 if constr.upper is None else constr.upper) +
(0 if constr.lower is None else constr.lower))
sign_adjust = 1 if value(constr.upper) is None else -1
duals[i] = sign_adjust * max(0, sign_adjust *
(rhs - value(constr.body)))
if value(MindtPy.MindtPy_feas_obj.expr) == 0:
raise ValueError('Problem is not feasible, check NLP solver')
return var_values, duals
def process_objective(solve_data, config):
m = solve_data.working_model
GDPopt = m.GDPopt_utils
# Handle missing or multiple objectives
objs = list(
m.component_data_objects(ctype=Objective,
active=True,
descend_into=True))
num_objs = len(objs)
solve_data.results.problem.number_of_objectives = num_objs
if num_objs == 0:
config.logger.warning(
'Model has no active objectives. Adding dummy objective.')
GDPopt.dummy_objective = Objective(expr=1)
main_obj = GDPopt.dummy_objective
elif num_objs > 1:
raise ValueError('Model has multiple active objectives.')
else:
main_obj = objs[0]
solve_data.working_objective_expr = main_obj.expr
# Move the objective to the constraints
# TODO only move the objective if nonlinear?
GDPopt.objective_value = Var(domain=Reals, initialize=0)
solve_data.objective_sense = main_obj.sense
if main_obj.sense == minimize:
GDPopt.objective_expr = Constraint(
expr=GDPopt.objective_value >= main_obj.expr)
solve_data.results.problem.sense = ProblemSense.minimize
else:
GDPopt.objective_expr = Constraint(
expr=GDPopt.objective_value <= main_obj.expr)
solve_data.results.problem.sense = ProblemSense.maximize
main_obj.deactivate()
GDPopt.objective = Objective(expr=GDPopt.objective_value,
sense=main_obj.sense)
def select_tear_mip_model(self, G):
"""
Generate a model for selecting tears from the given graph
Returns
-------
model
bin_list
A list of the binary variables representing each edge,
indexed by the edge index of the graph
"""
model = ConcreteModel()
bin_list = []
for i in range(G.number_of_edges()):
# add a binary "torn" variable for every edge
vname = "edge%s" % i
var = Var(domain=Binary)
bin_list.append(var)
model.add_component(vname, var)
# var containing the maximum number of times any cycle is torn
mct = model.max_cycle_tears = Var()
_, cycleEdges = self.all_cycles(G)
for i in range(len(cycleEdges)):
ecyc = cycleEdges[i]
# expression containing sum of tears for each cycle
ename = "cycle_sum%s" % i
expr = Expression(expr=sum(bin_list[i] for i in ecyc))
model.add_component(ename, expr)
# every cycle must have at least 1 tear
cname_min = "cycle_min%s" % i
con_min = Constraint(expr=expr >= 1)
model.add_component(cname_min, con_min)
# mct >= cycle_sum for all cycles, thus it becomes the max
cname_mct = mct.name + "_geq%s" % i
con_mct = Constraint(expr=mct >= expr)
model.add_component(cname_mct, con_mct)
# weigh the primary objective much greater than the secondary
obj_expr = 1000 * mct + sum(var for var in bin_list)
model.obj = Objective(expr=obj_expr, sense=minimize)
return model, bin_list
def get_model(self):
m = ConcreteModel()
m.x = Var(bounds=(-100, 100))
m.obj = Objective(expr=m.x)
m.disjunct1 = Disjunct()
m.disjunct1.comp = Complementarity(
expr=complements(m.x >= 0, 4 * m.x - 3 >= 0))
m.disjunct2 = Disjunct()
m.disjunct2.cons = Constraint(expr=m.x >= 2)
m.disjunction = Disjunction(expr=[m.disjunct1, m.disjunct2])
return m
def _generate_model(self):
self.model = ConcreteModel()
model = self.model
model._name = self.description
model.x = Var(within=Binary)
model.y = Var(within=Binary)
model.z = Var(within=Binary)
model.o = Objective(expr=-model.x - model.y - model.z)
model.c1 = Constraint(expr=model.x + model.y <= 1)
model.c2 = Constraint(expr=model.x + model.z <= 1)
model.c3 = Constraint(expr=model.y + model.z <= 1)
model.c4 = Constraint(expr=model.x + model.y + model.z >= 1.5)
def makeNestedDisjunctions_NestedDisjuncts():
m = ConcreteModel()
m.x = Var(bounds=(0, 2))
m.obj = Objective(expr=m.x)
m.d1 = Disjunct()
m.d1.c = Constraint(expr=m.x >= 1)
m.d2 = Disjunct()
m.d2.c = Constraint(expr=m.x >= 1.1)
m.d1.d3 = Disjunct()
m.d1.d3.c = Constraint(expr=m.x >= 1.2)
m.d1.d4 = Disjunct()
m.d1.d4.c = Constraint(expr=m.x >= 1.3)
m.disj = Disjunction(expr=[m.d1, m.d2])
m.d1.disj2 = Disjunction(expr=[m.d1.d3, m.d1.d4])
return m
def _generate_model(self):
self.model = ConcreteModel()
model = self.model
model._name = self.description
model.x = Var()
model.y = Var()
model.obj = Objective(expr=model.y)
model.p = Piecewise(model.y, model.x,
pw_pts=[-1,0,1],
f_rule=[1,0.5,1],
pw_repn='SOS2',
pw_constr_type='LB',
unbounded_domain_var=True)
def test_add_column_exceptions(self):
m = ConcreteModel()
m.x = Var()
m.c = Constraint(expr=(0, m.x, 1))
m.ci = Constraint([1,2], rule=lambda m,i:(0,m.x,i+1))
m.cd = Constraint(expr=(0, -m.x, 1))
m.cd.deactivate()
m.obj = Objective(expr=-m.x)
opt = SolverFactory('cplex_persistent')
# set_instance not called
self.assertRaises(RuntimeError, opt.add_column, m, m.x, 0, [m.c], [1])
opt.set_instance(m)
m2 = ConcreteModel()
m2.y = Var()
m2.c = Constraint(expr=(0,m.x,1))
# different model than attached to opt
self.assertRaises(RuntimeError, opt.add_column, m2, m2.y, 0, [], [])
# pyomo var attached to different model
self.assertRaises(RuntimeError, opt.add_column, m, m2.y, 0, [], [])
z = Var()
# pyomo var floating
self.assertRaises(RuntimeError, opt.add_column, m, z, -2, [m.c, z], [1])
m.y = Var()
# len(coefficents) == len(constraints)
self.assertRaises(RuntimeError, opt.add_column, m, m.y, -2, [m.c], [1,2])
self.assertRaises(RuntimeError, opt.add_column, m, m.y, -2, [m.c, z], [1])
# add indexed constraint
self.assertRaises(AttributeError, opt.add_column, m, m.y, -2, [m.ci], [1])
# add something not a _ConstraintData
self.assertRaises(AttributeError, opt.add_column, m, m.y, -2, [m.x], [1])
# constraint not on solver model
self.assertRaises(KeyError, opt.add_column, m, m.y, -2, [m2.c], [1])
# inactive constraint
self.assertRaises(KeyError, opt.add_column, m, m.y, -2, [m.cd], [1])
opt.add_var(m.y)
# var already in solver model
self.assertRaises(RuntimeError, opt.add_column, m, m.y, -2, [m.c], [1])
def init_max_binaries(solve_data, config):
"""Initialize by turning on as many binary variables as possible.
The user would usually want to call _solve_NLP_subproblem after an
invocation of this function.
"""
m = solve_data.working_model.clone()
m.dual.deactivate()
MindtPy = m.MindtPy_utils
solve_data.mip_subiter += 1
config.logger.info("MILP %s: maximize value of binaries" %
(solve_data.mip_iter))
for c in MindtPy.constraint_list:
if c.body.polynomial_degree() not in (1, 0):
c.deactivate()
objective = next(m.component_data_objects(Objective, active=True))
objective.deactivate()
binary_vars = (v for v in m.component_data_objects(ctype=Var)
if v.is_binary() and not v.fixed)
MindtPy.MindtPy_max_binary_obj = Objective(expr=sum(v
for v in binary_vars),
sense=maximize)
getattr(m, 'ipopt_zL_out', _DoNothing()).deactivate()
getattr(m, 'ipopt_zU_out', _DoNothing()).deactivate()
opt = SolverFactory(config.mip_solver)
if isinstance(opt, PersistentSolver):
opt.set_instance(m)
results = opt.solve(m, options=config.mip_solver_args)
solve_terminate_cond = results.solver.termination_condition
if solve_terminate_cond is tc.optimal:
copy_var_list_values(
MindtPy.variable_list,
solve_data.working_model.MindtPy_utils.variable_list, config)
pass # good
elif solve_terminate_cond is tc.infeasible:
raise ValueError('MILP master problem is infeasible. '
'Problem may have no more feasible '
'binary configurations.')
else:
raise ValueError(
'MindtPy unable to handle MILP master termination condition '
'of %s. Solver message: %s' %
(solve_terminate_cond, results.solver.message))
def define_model(**kwds):
model = ConcreteModel()
model.x1 = Var(bounds=(0, 6)) # domain variable
model.x2 = Var(bounds=(0, 6)) # domain variable
model.x3 = Var(bounds=(0, 6)) # domain variable
model.x4 = Var(bounds=(0, 6)) # domain variable
model.Fx1 = Var() # range variable
model.Fx2 = Var() # range variable
model.Fx3 = Var() # range variable
model.Fx4 = Var() # range variable
model.p = Param(initialize=1.0)
model.obj = Objective(expr=model.Fx1 + model.Fx2 + model.Fx3 + model.Fx4,
sense=kwds.pop('sense', maximize))
model.piecewise1 = Piecewise(model.Fx1,
model.x1,
pw_pts=DOMAIN_PTS,
f_rule=F,
**kwds)
model.piecewise2 = Piecewise(model.Fx2,
model.x2,
pw_pts=DOMAIN_PTS,
f_rule=F,
**kwds)
model.piecewise3 = Piecewise(model.Fx3,
model.x3,
pw_pts=DOMAIN_PTS,
f_rule=F,
**kwds)
model.piecewise4 = Piecewise(model.Fx4,
model.x4,
pw_pts=DOMAIN_PTS,
f_rule=F,
**kwds)
#Fix the answer for testing purposes
model.set_answer_constraint1 = Constraint(expr=model.x1 == 0.0)
model.set_answer_constraint2 = Constraint(expr=model.x2 == 3.0)
model.set_answer_constraint3 = Constraint(expr=model.x3 == 5.5)
model.set_answer_constraint4 = Constraint(expr=model.x4 == 6.0)
return model
def build_mip(self):
"""Build the model <b>object</b>."""
self.model = AbstractModel()
self.model.J = Set()
self.model.K = Set(self.model.J)
def jk_init(m):
return [(j, k) for j in m.J for k in m.K[j]]
self.model.JK = Set(initialize=jk_init, dimen=None)
self.model.y_pred = Param()
self.model.epsilon = Param()
self.model.max_cost = Var()
self.model.w = Param(self.model.J)
self.model.a = Param(self.model.JK)
self.model.c = Param(self.model.JK)
self.model.u = Var(self.model.JK, within=Binary)
# Make sure only one action is on at a time.
def c1Rule(m, j):
return (sum([m.u[j, k] for k in m.K[j]])) == 1
# 2.b: Action sets must flip the prediction of a linear classifier.
def c2Rule(m):
return (sum((m.u[j, k] * m.a[j, k] * m.w[j])
for j, k in m.JK) >= -m.y_pred)
# instantiate max cost
def maxcost_rule(m, j, k):
return m.max_cost >= (m.u[j, k] * m.c[j, k])
# Set up objective for total sum.
def obj_rule_percentile(m):
return sum(m.u[j, k] * m.c[j, k] for j, k in m.JK)
# Set up objective for max cost.
def obj_rule_max(m):
return (sum(m.epsilon * m.u[j, k] * m.c[j, k]
for j, k in m.JK) + (1 - m.epsilon) * m.max_cost)
##
self.model.g = Objective(rule=obj_rule_max, sense=minimize)
self.model.c1 = Constraint(self.model.J, rule=c1Rule)
self.model.c2 = Constraint(rule=c2Rule)
self.model.c3 = Constraint(self.model.JK, rule=maxcost_rule)
self.built = True
def makeBetweenStepsPaperExample_DeclareVarOnDisjunct():
"""Exactly the same model as above, but declaring the Disjuncts explicitly
and declaring the variables on one of them.
"""
m = ConcreteModel()
m.I = RangeSet(1, 4)
m.disj1 = Disjunct()
m.disj1.x = Var(m.I, bounds=(-2, 6))
m.disj1.c = Constraint(expr=sum(m.disj1.x[i]**2 for i in m.I) <= 1)
m.disj2 = Disjunct()
m.disj2.c = Constraint(expr=sum((3 - m.disj1.x[i])**2 for i in m.I) <= 1)
m.disjunction = Disjunction(expr=[m.disj1, m.disj2])
m.obj = Objective(expr=m.disj1.x[2] - m.disj1.x[1], sense=maximize)
return m
def test_divide_by_mutable(self):
#
# Test from https://github.com/Pyomo/pyomo/issues/153
#
m = ConcreteModel()
m.x = Var(bounds=(1,5))
m.p = Param(initialize=100, mutable=True)
m.con = Constraint(expr=exp(5*(1/m.x - 1/m.p))<=10)
m.obj = Objective(expr=m.x**2)
test = gar(m.con.body)
self.assertEqual(test.constant, 0)
self.assertEqual(test.linear_vars, tuple())
self.assertEqual(test.linear_coefs, tuple())
self.assertEqual(set(id(v) for v in test.nonlinear_vars), set([id(m.x)]))
self.assertIs(test.nonlinear_expr, m.con.body)
def generate_norm1_objective_function(model,
setpoint_model,
discrete_only=False):
r"""This function generates objective (PF-OA main problem) for minimum Norm1 distance to setpoint_model.
Norm1 distance of (x,y) = \sum_i |x_i - y_i|.
Parameters
----------
model : Pyomo model
The model that needs new objective function.
setpoint_model : Pyomo model
The model that provides the base point for us to calculate the distance.
discrete_only : bool, optional
Whether to only optimize on distance between the discrete variables, by default False.
Returns
-------
Objective
The norm1 objective function.
"""
# skip objective_value variable and slack_var variables
var_filter = (lambda v: v.is_integer()) if discrete_only \
else (lambda v: 'MindtPy_utils.objective_value' not in v.name and
'MindtPy_utils.feas_opt.slack_var' not in v.name)
model_vars = list(filter(var_filter, model.MindtPy_utils.variable_list))
setpoint_vars = list(
filter(var_filter, setpoint_model.MindtPy_utils.variable_list))
assert len(model_vars) == len(
setpoint_vars
), 'Trying to generate Norm1 objective function for models with different number of variables'
model.MindtPy_utils.del_component('L1_obj')
obj_blk = model.MindtPy_utils.L1_obj = Block()
obj_blk.L1_obj_idx = RangeSet(len(model_vars))
obj_blk.L1_obj_var = Var(obj_blk.L1_obj_idx,
domain=Reals,
bounds=(0, None))
obj_blk.abs_reform = ConstraintList()
for idx, v_model, v_setpoint in zip(obj_blk.L1_obj_idx, model_vars,
setpoint_vars):
obj_blk.abs_reform.add(
expr=v_model - v_setpoint.value >= -obj_blk.L1_obj_var[idx])
obj_blk.abs_reform.add(
expr=v_model - v_setpoint.value <= obj_blk.L1_obj_var[idx])
return Objective(expr=sum(obj_blk.L1_obj_var[idx]
for idx in obj_blk.L1_obj_idx))
def grossmann_oneDisj():
m = ConcreteModel()
m.x = Var(bounds=(0,20))
m.y = Var(bounds=(0, 20))
m.disjunct1 = Disjunct()
m.disjunct1.constraintx = Constraint(expr=inequality(0, m.x, 2))
m.disjunct1.constrainty = Constraint(expr=inequality(7, m.y, 10))
m.disjunct2 = Disjunct()
m.disjunct2.constraintx = Constraint(expr=inequality(8, m.x, 10))
m.disjunct2.constrainty = Constraint(expr=inequality(0, m.y, 3))
m.disjunction = Disjunction(expr=[m.disjunct1, m.disjunct2])
m.objective = Objective(expr=m.x + 2*m.y, sense=maximize)
return m
def add_cut(self, first=False):
self._iter += 1
model = self._model
self._wprod[self._iter] = self._compute_weight_weight_inner_product()
if first is True:
self._alphas[self._iter] = -(
self._compute_objective_term() +
(self._ph._rho / 2.0) * self._wprod[self._iter])
else:
self._alphas[self._iter] = -(self._compute_objective_term(
)) + self._compute_xbar_weight_inner_product()
if self._solved is True:
if self._compute_convergence() is True:
return True
model.del_component('cuts')
model.cuts = Set(initialize=sorted(self._alphas.keys()))
model.del_component('beta')
model.beta = Var(model.cuts, within=NonNegativeReals)
model.del_component('beta_sum_one')
model.beta_sum_one = Constraint(expr=sum_product(model.beta) == 1)
model.del_component('obj')
model.obj = Objective(expr=sum(self._alphas[i] * model.beta[i]
for i in model.cuts))
self._wbars[self._iter] = {}
for stage in self._ph._scenario_tree._stages[:
-1]: # all blended stages
for tree_node in stage._tree_nodes:
self._wbars[self._iter][tree_node._name] = copy.deepcopy(
tree_node._wbars)
block = getattr(model, tree_node._name)
def _c_rule(block, i):
lhs = sum(model.beta[k] * self._wbars[k][tree_node._name][
block.id_to_var[i][0]][block.id_to_var[i][1]]
for k in model.beta.index_set())
if not isinstance(lhs, ExpressionBase):
return Constraint.Skip
return lhs == 0
block.del_component('con')
block.con = Constraint(block.var_index, rule=_c_rule)
return False
def init_max_binaries(solve_data, config):
"""Initialize by maximizing binary variables and disjuncts.
This function activates as many binary variables and disjucts as
feasible.
"""
solve_data.mip_iteration += 1
linear_GDP = solve_data.linear_GDP.clone()
config.logger.info("Generating initial linear GDP approximation by "
"solving a subproblem that maximizes "
"the sum of all binary and logical variables.")
# Set up binary maximization objective
linear_GDP.GDPopt_utils.objective.deactivate()
binary_vars = (v for v in linear_GDP.component_data_objects(
ctype=Var, descend_into=(Block, Disjunct))
if v.is_binary() and not v.fixed)
linear_GDP.GDPopt_utils.max_binary_obj = Objective(expr=sum(binary_vars),
sense=maximize)
# Solve
mip_results = solve_linear_GDP(linear_GDP, solve_data, config)
if mip_results:
_, mip_var_values = mip_results
# use the mip_var_values to create the NLP subproblem
nlp_model = solve_data.working_model.clone()
# copy in the discrete variable values
copy_and_fix_mip_values_to_nlp(nlp_model.GDPopt_utils.working_var_list,
mip_var_values, config)
TransformationFactory('gdp.fix_disjuncts').apply_to(nlp_model)
solve_data.nlp_iteration += 1
nlp_result = solve_NLP(nlp_model, solve_data, config)
nlp_feasible, nlp_var_values, nlp_duals = nlp_result
if nlp_feasible:
update_nlp_progress_indicators(nlp_model, solve_data, config)
add_outer_approximation_cuts(nlp_var_values, nlp_duals, solve_data,
config)
add_integer_cut(mip_var_values,
solve_data,
config,
feasible=nlp_feasible)
else:
config.logger.info(
"Linear relaxation for initialization was infeasible. "
"Problem is infeasible.")
return False
def makeNestedDisjunctions_NestedDisjuncts():
"""Same as makeNestedDisjunctions_FlatDisjuncts except that the disjuncts
of the nested disjunction are declared on the parent disjunct."""
m = ConcreteModel()
m.x = Var(bounds=(0, 2))
m.obj = Objective(expr=m.x)
m.d1 = Disjunct()
m.d1.c = Constraint(expr=m.x >= 1)
m.d2 = Disjunct()
m.d2.c = Constraint(expr=m.x >= 1.1)
m.d1.d3 = Disjunct()
m.d1.d3.c = Constraint(expr=m.x >= 1.2)
m.d1.d4 = Disjunct()
m.d1.d4.c = Constraint(expr=m.x >= 1.3)
m.disj = Disjunction(expr=[m.d1, m.d2])
m.d1.disj2 = Disjunction(expr=[m.d1.d3, m.d1.d4])
return m
def makeNestedDisjunctions_FlatDisjuncts():
"""Two-term SimpleDisjunction where one of the disjuncts contains a nested
SimpleDisjunction, the disjuncts of which are declared on the model"""
m = ConcreteModel()
m.x = Var(bounds=(0, 2))
m.obj = Objective(expr=m.x)
m.d1 = Disjunct()
m.d1.c = Constraint(expr=m.x >= 1)
m.d2 = Disjunct()
m.d2.c = Constraint(expr=m.x >= 1.1)
m.d3 = Disjunct()
m.d3.c = Constraint(expr=m.x >= 1.2)
m.d4 = Disjunct()
m.d4.c = Constraint(expr=m.x >= 1.3)
m.disj = Disjunction(expr=[m.d1, m.d2])
m.d1.disj = Disjunction(expr=[m.d3, m.d4])
return m
def do_setup(self):
global tmpdir
tmpdir = os.getcwd()
os.chdir(currdir)
TempfileManager.sequential_files(0)
self.asl = pyomo.opt.SolverFactory('asl:ipopt', keepfiles=True)
self.ipopt = pyomo.opt.SolverFactory('ipopt', keepfiles=True)
# The sisser CUTEr instance
# Formulated in Pyomo by Carl D. Laird, Daniel P. Word, Brandon C. Barrera and Saumyajyoti Chaudhuri
# Taken from:
# Source:
# F.S. Sisser,
# "Elimination of bounds in optimization problems by transforming
# variables",
# Mathematical Programming 20:110-121, 1981.
# See also Buckley#216 (p. 91)
# SIF input: Ph. Toint, Dec 1989.
# classification OUR2-AN-2-0
sisser_instance = ConcreteModel()
sisser_instance.N = RangeSet(1, 2)
sisser_instance.xinit = Param(sisser_instance.N,
initialize={
1: 1.0,
2: 0.1
})
def fa(model, i):
return value(model.xinit[i])
sisser_instance.x = Var(sisser_instance.N, initialize=fa)
def f(model):
return 3 * model.x[1]**4 - 2 * (model.x[1] *
model.x[2])**2 + 3 * model.x[2]**4
sisser_instance.f = Objective(rule=f, sense=minimize)
self.sisser_instance = sisser_instance
def to_break_constraint_tolerances():
m = ConcreteModel()
m.x = Var(bounds=(0, 130))
m.y = Var(bounds=(0, 130))
m.disjunct1 = Disjunct()
m.disjunct1.constraintx = Constraint(expr=inequality(0, m.x, 2))
m.disjunct1.constrainty = Constraint(expr=inequality(117, m.y, 127))
m.disjunct2 = Disjunct()
m.disjunct2.constraintx = Constraint(expr=inequality(118, m.x, 120))
m.disjunct2.constrainty = Constraint(expr=inequality(0, m.y, 3))
m.disjunction = Disjunction(expr=[m.disjunct1, m.disjunct2])
m.objective = Objective(expr=m.x + 2*m.y, sense=maximize)
return m