def setup_main(solve_data, config, fp, regularization_problem):
"""Set up main problem/main regularization problem for OA, ECP, Feasibility Pump and ROA methods.
Args:
solve_data (MindtPySolveData): data container that holds solve-instance data.
config (ConfigBlock): the specific configurations for MindtPy.
fp (bool): whether it is in the loop of feasibility pump.
regularization_problem (bool): whether it is solving a regularization problem.
"""
MindtPy = solve_data.mip.MindtPy_utils
for c in MindtPy.constraint_list:
if c.body.polynomial_degree() not in {1, 0}:
c.deactivate()
MindtPy.cuts.activate()
sign_adjust = 1 if solve_data.objective_sense == minimize else -1
MindtPy.del_component('mip_obj')
if regularization_problem and config.single_tree:
MindtPy.del_component('loa_proj_mip_obj')
MindtPy.cuts.del_component('obj_reg_estimate')
if config.add_regularization is not None and config.add_no_good_cuts:
if regularization_problem:
MindtPy.cuts.no_good_cuts.activate()
else:
MindtPy.cuts.no_good_cuts.deactivate()
if fp:
MindtPy.del_component('fp_mip_obj')
if config.fp_main_norm == 'L1':
MindtPy.fp_mip_obj = generate_norm1_objective_function(
solve_data.mip,
solve_data.working_model,
discrete_only=config.fp_discrete_only)
elif config.fp_main_norm == 'L2':
MindtPy.fp_mip_obj = generate_norm2sq_objective_function(
solve_data.mip,
solve_data.working_model,
discrete_only=config.fp_discrete_only)
elif config.fp_main_norm == 'L_infinity':
MindtPy.fp_mip_obj = generate_norm_inf_objective_function(
solve_data.mip,
solve_data.working_model,
discrete_only=config.fp_discrete_only)
elif regularization_problem:
if MindtPy.objective_list[0].expr.polynomial_degree() in {1, 0}:
MindtPy.objective_constr.activate()
if config.add_regularization == 'level_L1':
MindtPy.loa_proj_mip_obj = generate_norm1_objective_function(
solve_data.mip,
solve_data.best_solution_found,
discrete_only=False)
elif config.add_regularization == 'level_L2':
MindtPy.loa_proj_mip_obj = generate_norm2sq_objective_function(
solve_data.mip,
solve_data.best_solution_found,
discrete_only=False)
elif config.add_regularization == 'level_L_infinity':
MindtPy.loa_proj_mip_obj = generate_norm_inf_objective_function(
solve_data.mip,
solve_data.best_solution_found,
discrete_only=False)
elif config.add_regularization in {
'grad_lag', 'hess_lag', 'hess_only_lag', 'sqp_lag'
}:
MindtPy.loa_proj_mip_obj = generate_lag_objective_function(
solve_data.mip,
solve_data.best_solution_found,
config,
solve_data,
discrete_only=False)
if solve_data.objective_sense == minimize:
MindtPy.cuts.obj_reg_estimate = Constraint(
expr=MindtPy.objective_value <= (1 - config.level_coef) *
solve_data.UB + config.level_coef * solve_data.LB)
else:
MindtPy.cuts.obj_reg_estimate = Constraint(
expr=MindtPy.objective_value >= (1 - config.level_coef) *
solve_data.LB + config.level_coef * solve_data.UB)
else:
if config.add_slack:
MindtPy.del_component('aug_penalty_expr')
MindtPy.aug_penalty_expr = Expression(
expr=sign_adjust * config.OA_penalty_factor *
sum(v for v in MindtPy.cuts.slack_vars[...]))
main_objective = MindtPy.objective_list[-1]
MindtPy.mip_obj = Objective(
expr=main_objective.expr +
(MindtPy.aug_penalty_expr if config.add_slack else 0),
sense=solve_data.objective_sense)
if config.use_dual_bound:
# Delete previously added dual bound constraint
MindtPy.cuts.del_component('dual_bound')
if solve_data.objective_sense == minimize:
MindtPy.cuts.dual_bound = Constraint(
expr=main_objective.expr +
(MindtPy.aug_penalty_expr if config.add_slack else 0) >=
solve_data.LB,
doc=
'Objective function expression should improve on the best found dual bound'
)
else:
MindtPy.cuts.dual_bound = Constraint(
expr=main_objective.expr +
(MindtPy.aug_penalty_expr if config.add_slack else 0) <=
solve_data.UB,
doc=
'Objective function expression should improve on the best found dual bound'
)
def solve_fp_subproblem(solve_data, config):
"""Solves the feasibility pump NLP subproblem.
This function sets up the 'fp_nlp' by relax integer variables.
precomputes dual values, deactivates trivial constraints, and then solves NLP model.
Parameters
----------
solve_data : MindtPySolveData
Data container that holds solve-instance data.
config : ConfigBlock
The specific configurations for MindtPy.
Returns
-------
fp_nlp : Pyomo model
Fixed-NLP from the model.
results : SolverResults
Results from solving the fixed-NLP subproblem.
"""
fp_nlp = solve_data.working_model.clone()
MindtPy = fp_nlp.MindtPy_utils
# Set up NLP
fp_nlp.MindtPy_utils.objective_list[-1].deactivate()
if solve_data.objective_sense == minimize:
fp_nlp.improving_objective_cut = Constraint(
expr=sum(fp_nlp.MindtPy_utils.objective_value[:]) <= solve_data.UB)
else:
fp_nlp.improving_objective_cut = Constraint(
expr=sum(fp_nlp.MindtPy_utils.objective_value[:]) >= solve_data.LB)
# Add norm_constraint, which guarantees the monotonicity of the norm objective value sequence of all iterations
# Ref: Paper 'A storm of feasibility pumps for nonconvex MINLP' https://doi.org/10.1007/s10107-012-0608-x
# the norm type is consistant with the norm obj of the FP-main problem.
if config.fp_norm_constraint:
generate_norm_constraint(fp_nlp, solve_data, config)
MindtPy.fp_nlp_obj = generate_norm2sq_objective_function(
fp_nlp, solve_data.mip, discrete_only=config.fp_discrete_only)
MindtPy.cuts.deactivate()
TransformationFactory('core.relax_integer_vars').apply_to(fp_nlp)
try:
TransformationFactory(
'contrib.deactivate_trivial_constraints').apply_to(
fp_nlp,
tmp=True,
ignore_infeasible=False,
tolerance=config.constraint_tolerance)
except ValueError:
config.logger.warning(
'infeasibility detected in deactivate_trivial_constraints')
results = SolverResults()
results.solver.termination_condition = tc.infeasible
return fp_nlp, results
# Solve the NLP
nlpopt = SolverFactory(config.nlp_solver)
nlp_args = dict(config.nlp_solver_args)
set_solver_options(nlpopt, solve_data, config, solver_type='nlp')
with SuppressInfeasibleWarning():
with time_code(solve_data.timing, 'fp subproblem'):
results = nlpopt.solve(fp_nlp,
tee=config.nlp_solver_tee,
**nlp_args)
return fp_nlp, results
def solve_fp_subproblem(solve_data, config):
"""
Solves the feasibility pump NLP
This function sets up the 'fp_nlp' by relax integer variables.
precomputes dual values, deactivates trivial constraints, and then solves NLP model.
Parameters
----------
solve_data: MindtPy Data Container
data container that holds solve-instance data
config: ConfigBlock
contains the specific configurations for the algorithm
Returns
-------
fp_nlp: Pyomo model
Fixed-NLP from the model
results: Pyomo results object
result from solving the Fixed-NLP
"""
fp_nlp = solve_data.working_model.clone()
MindtPy = fp_nlp.MindtPy_utils
config.logger.info('FP-NLP %s: Solve feasibility pump NLP subproblem.'
% (solve_data.fp_iter,))
# Set up NLP
fp_nlp.MindtPy_utils.objective_list[-1].deactivate()
if solve_data.objective_sense == minimize:
fp_nlp.improving_objective_cut = Constraint(
expr=fp_nlp.MindtPy_utils.objective_value <= solve_data.UB)
else:
fp_nlp.improving_objective_cut = Constraint(
expr=fp_nlp.MindtPy_utils.objective_value >= solve_data.LB)
# Add norm_constraint, which guarantees the monotonicity of the norm objective value sequence of all iterations
# Ref: Paper 'A storm of feasibility pumps for nonconvex MINLP'
# the norm type is consistant with the norm obj of the FP-main problem.
if config.fp_norm_constraint:
if config.fp_main_norm == 'L1':
# TODO: check if we can access the block defined in FP-main problem
generate_norm1_norm_constraint(
fp_nlp, solve_data.mip, config, discrete_only=True)
elif config.fp_main_norm == 'L2':
fp_nlp.norm_constraint = Constraint(expr=sum((nlp_var - mip_var.value)**2 - config.fp_norm_constraint_coef*(nlp_var.value - mip_var.value)**2
for nlp_var, mip_var in zip(fp_nlp.MindtPy_utils.discrete_variable_list, solve_data.mip.MindtPy_utils.discrete_variable_list)) <= 0)
elif config.fp_main_norm == 'L_infinity':
fp_nlp.norm_constraint = ConstraintList()
rhs = config.fp_norm_constraint_coef * max(nlp_var.value - mip_var.value for nlp_var, mip_var in zip(
fp_nlp.MindtPy_utils.discrete_variable_list, solve_data.mip.MindtPy_utils.discrete_variable_list))
for nlp_var, mip_var in zip(fp_nlp.MindtPy_utils.discrete_variable_list, solve_data.mip.MindtPy_utils.discrete_variable_list):
fp_nlp.norm_constraint.add(nlp_var - mip_var.value <= rhs)
MindtPy.fp_nlp_obj = generate_norm2sq_objective_function(
fp_nlp, solve_data.mip, discrete_only=config.fp_discrete_only)
MindtPy.cuts.deactivate()
TransformationFactory('core.relax_integer_vars').apply_to(fp_nlp)
try:
TransformationFactory('contrib.deactivate_trivial_constraints').apply_to(
fp_nlp, tmp=True, ignore_infeasible=False, tolerance=config.constraint_tolerance)
except ValueError:
config.logger.warning(
'infeasibility detected in deactivate_trivial_constraints')
results = SolverResults()
results.solver.termination_condition = tc.infeasible
return fp_nlp, results
# Solve the NLP
nlpopt = SolverFactory(config.nlp_solver)
nlp_args = dict(config.nlp_solver_args)
set_solver_options(nlpopt, solve_data, config, solver_type='nlp')
with SuppressInfeasibleWarning():
with time_code(solve_data.timing, 'fp subproblem'):
results = nlpopt.solve(
fp_nlp, tee=config.nlp_solver_tee, **nlp_args)
return fp_nlp, results
def setup_main(solve_data, config, fp, regularization_problem):
"""Set up main problem/main regularization problem for OA, ECP, Feasibility Pump and ROA methods.
Parameters
----------
solve_data : MindtPySolveData
Data container that holds solve-instance data.
config : ConfigBlock
The specific configurations for MindtPy.
fp : bool
Whether it is in the loop of feasibility pump.
regularization_problem : bool
Whether it is solving a regularization problem.
"""
MindtPy = solve_data.mip.MindtPy_utils
for c in MindtPy.constraint_list:
if c.body.polynomial_degree(
) not in solve_data.mip_constraint_polynomial_degree:
c.deactivate()
MindtPy.cuts.activate()
sign_adjust = 1 if solve_data.objective_sense == minimize else -1
MindtPy.del_component('mip_obj')
if regularization_problem and config.single_tree:
MindtPy.del_component('loa_proj_mip_obj')
MindtPy.cuts.del_component('obj_reg_estimate')
if config.add_regularization is not None and config.add_no_good_cuts:
if regularization_problem:
MindtPy.cuts.no_good_cuts.activate()
else:
MindtPy.cuts.no_good_cuts.deactivate()
if fp:
MindtPy.del_component('fp_mip_obj')
if config.fp_main_norm == 'L1':
MindtPy.fp_mip_obj = generate_norm1_objective_function(
solve_data.mip,
solve_data.working_model,
discrete_only=config.fp_discrete_only)
elif config.fp_main_norm == 'L2':
MindtPy.fp_mip_obj = generate_norm2sq_objective_function(
solve_data.mip,
solve_data.working_model,
discrete_only=config.fp_discrete_only)
elif config.fp_main_norm == 'L_infinity':
MindtPy.fp_mip_obj = generate_norm_inf_objective_function(
solve_data.mip,
solve_data.working_model,
discrete_only=config.fp_discrete_only)
elif regularization_problem:
# The epigraph constraint is very "flat" for branching rules.
# In ROA, if the objective function is linear(or quadratic when quadratic_strategy = 1 or 2), the original objective function is used in the MIP problem.
# In the MIP projection problem, we need to reactivate the epigraph constraint(objective_constr).
if MindtPy.objective_list[0].expr.polynomial_degree(
) in solve_data.mip_objective_polynomial_degree:
MindtPy.objective_constr.activate()
if config.add_regularization == 'level_L1':
MindtPy.loa_proj_mip_obj = generate_norm1_objective_function(
solve_data.mip,
solve_data.best_solution_found,
discrete_only=False)
elif config.add_regularization == 'level_L2':
MindtPy.loa_proj_mip_obj = generate_norm2sq_objective_function(
solve_data.mip,
solve_data.best_solution_found,
discrete_only=False)
elif config.add_regularization == 'level_L_infinity':
MindtPy.loa_proj_mip_obj = generate_norm_inf_objective_function(
solve_data.mip,
solve_data.best_solution_found,
discrete_only=False)
elif config.add_regularization in {
'grad_lag', 'hess_lag', 'hess_only_lag', 'sqp_lag'
}:
MindtPy.loa_proj_mip_obj = generate_lag_objective_function(
solve_data.mip,
solve_data.best_solution_found,
config,
solve_data,
discrete_only=False)
if solve_data.objective_sense == minimize:
MindtPy.cuts.obj_reg_estimate = Constraint(
expr=sum(MindtPy.objective_value[:]) <=
(1 - config.level_coef) * solve_data.primal_bound +
config.level_coef * solve_data.dual_bound)
else:
MindtPy.cuts.obj_reg_estimate = Constraint(
expr=sum(MindtPy.objective_value[:]) >=
(1 - config.level_coef) * solve_data.primal_bound +
config.level_coef * solve_data.dual_bound)
else:
if config.add_slack:
MindtPy.del_component('aug_penalty_expr')
MindtPy.aug_penalty_expr = Expression(
expr=sign_adjust * config.OA_penalty_factor *
sum(v for v in MindtPy.cuts.slack_vars[...]))
main_objective = MindtPy.objective_list[-1]
MindtPy.mip_obj = Objective(
expr=main_objective.expr +
(MindtPy.aug_penalty_expr if config.add_slack else 0),
sense=solve_data.objective_sense)
if config.use_dual_bound:
# Delete previously added dual bound constraint
MindtPy.cuts.del_component('dual_bound')
if solve_data.objective_sense == minimize:
MindtPy.cuts.dual_bound = Constraint(
expr=main_objective.expr +
(MindtPy.aug_penalty_expr if config.add_slack else 0) >=
solve_data.dual_bound,
doc=
'Objective function expression should improve on the best found dual bound'
)
else:
MindtPy.cuts.dual_bound = Constraint(
expr=main_objective.expr +
(MindtPy.aug_penalty_expr if config.add_slack else 0) <=
solve_data.dual_bound,
doc=
'Objective function expression should improve on the best found dual bound'
)