I am programming in python 3.6 using pyomo 5.3
I wish to modify the body of a non-indexed Constraint (in case it is not in the standard format I require). The problem is that the value of the constraint at the specific point is calculated when subtracting from the body. However, I require the body in the form of a function because I have to construct an Objective which is the maximum of all non-linear constraints leading to a min-max-problem.
I tried to directly set the body of the constraint passed to the function but I get the output that the attribute cannot be set. Is there a function to set the body of a constraint?
Edit: Here is the solution I found:
constraint._body = ...
I wish to use this to change a change the form of an optimization problem.
(Sorry for the non-english comments)
Here are the functions used in this example:
._body to modify the constraints.def cont_relax_model_same_bounds(model_vars):
for var in model_vars:
if str(var.domain) in int_type:
var.domain = Reals
def epgraph_reformulation_without_bounds(model):
#Erstelle Epigraph-Modell
epi_model = model.clone()
epi_model.alpha_epi = Var(within = Reals)
#Speichere alle nichtlinearen Restriktionen des usprünglichen Modells in einer Liste
nonlinear_constrs = []
for constr in model.component_objects(Constraint):
if not (constr.body.polynomial_degree() in [0, 1]):
nonlinear_constrs.append(constr)
#Speichere alle nichtlinearen Restriktionen des umformulierten Modells in einer Liste
epi_nonlinear_constrs = []
for constr in epi_model.component_objects(Constraint):
if not (constr.body.polynomial_degree() in [0, 1]):
epi_nonlinear_constrs.append(constr)
#Kontrollausgabe, ob die Restriktionen richtig in der Liste gespeichert werden
for k, constr in enumerate(epi_nonlinear_constrs):
print(epi_nonlinear_constrs[k].body)
#Formuliere die nichtlinearen Restriktionen neu
for k, constr in enumerate(nonlinear_constrs):
epi_nonlinear_constrs[k]._body = (nonlinear_constrs[k].body - epi_model.alpha_epi)
epi_model.obj = Objective(expr = epi_model.alpha_epi, sense = minimize)
return epi_model
Here is the original model:
model_ESH = ConcreteModel(name = "Example 1")
model_ESH.x1 = Var(bounds=(1,20), domain=Reals)
model_ESH.x2 = Var(bounds=(1,20), domain=Integers)
model_ESH.obj = Objective(expr=(-1)*model_ESH.x1-model_ESH.x2)
model_ESH.g1 = Constraint(expr=0.15*((model_ESH.x1 - 8)**2)+0.1*((model_ESH.x2 - 6)**2)+0.025*exp(model_ESH.x1)*((model_ESH.x2)**(-2))-5<=0)
model_ESH.g2 = Constraint(expr=(model_ESH.x1)**(-1) + (model_ESH.x2)**(-1) - ((model_ESH.x1)**(0.5)) * ((model_ESH.x2) ** (0.5))+4<=0)
model_ESH.l1 = Constraint(expr=2 * (model_ESH.x1) - 3 * (model_ESH.x2) -2<=0)
model_ESH.pprint()
Then I clone the model and relax the integer variables
NLP_model = model_ESH.clone()
#Relaxiere das Problem und deaktiviere die nichtlinearen Restriktionen
#Das funktioniert schonmal
cont_relax_model_same_bounds(get_model_vars(NLP_model))
NLP_model.pprint()
2 Var Declarations
x1 : Size=1, Index=None
Key : Lower : Value : Upper : Fixed : Stale : Domain
None : 1 : None : 20 : False : True : Reals
x2 : Size=1, Index=None
Key : Lower : Value : Upper : Fixed : Stale : Domain
None : 1 : None : 20 : False : True : Reals
1 Objective Declarations
obj : Size=1, Index=None, Active=True
Key : Active : Sense : Expression
None : True : minimize : - x1 - x2
3 Constraint Declarations
g1 : Size=1, Index=None, Active=True
Key : Lower : Body : Upper : Active
None : -Inf : -5 + 0.15*( -8 + x1 )**2.0 + 0.1*( -6 + x2 )**2.0 + 0.025 * exp( x1 ) * x2**-2.0 : 0.0 : True
g2 : Size=1, Index=None, Active=True
Key : Lower : Body : Upper : Active
None : -Inf : 4 + x1**-1.0 + x2**-1.0 - x1**0.5 * x2**0.5 : 0.0 : True
l1 : Size=1, Index=None, Active=True
Key : Lower : Body : Upper : Active
None : -Inf : -2 + 2*x1 - 3*x2 : 0.0 : True
6 Declarations: x1 x2 obj g1 g2 l1
Now I use my function for changing/modifying the model:
epi_model_ESH = epgraph_reformulation_without_bounds(NLP_model)
epi_model_ESH.pprint()
WARNING: Implicitly replacing the Component attribute obj (type=<class
'pyomo.core.base.objective.SimpleObjective'>) on block Example 1 with a
new Component (type=<class 'pyomo.core.base.objective.SimpleObjective'>).
This is usually indicative of a modelling error. To avoid this warning,
use block.del_component() and block.add_component().
3 Var Declarations
alpha_epi : Size=1, Index=None
Key : Lower : Value : Upper : Fixed : Stale : Domain
None : None : None : None : False : True : Reals
x1 : Size=1, Index=None
Key : Lower : Value : Upper : Fixed : Stale : Domain
None : 1 : 8.636750397018059 : 20 : False : False : Reals
x2 : Size=1, Index=None
Key : Lower : Value : Upper : Fixed : Stale : Domain
None : 1 : 12.335071455814422 : 20 : False : False : Reals
1 Objective Declarations
obj : Size=1, Index=None, Active=True
Key : Active : Sense : Expression
None : True : minimize : alpha_epi
3 Constraint Declarations
g1 : Size=1, Index=None, Active=True
Key : Lower : Body : Upper : Active
None : -Inf : -5 + 0.15*( -8 + x1 )**2.0 + 0.1*( -6 + x2 )**2.0 + 0.025 * exp( x1 ) * x2**-2.0 - alpha_epi : 0.0 : True
g2 : Size=1, Index=None, Active=True
Key : Lower : Body : Upper : Active
None : -Inf : 4 + x1**-1.0 + x2**-1.0 - x1**0.5 * x2**0.5 - alpha_epi : 0.0 : True
l1 : Size=1, Index=None, Active=True
Key : Lower : Body : Upper : Active
None : -Inf : -2 + 2*x1 - 3*x2 : 0.0 : True
7 Declarations: x1 x2 g1 g2 l1 alpha_epi obj
However, if I try to use IPOPT to solve the newly created model I get the following error:
opt = SolverFactory('ipopt')
#opt.options['bonmin.algorithm'] = 'Bonmin'
print('using IPOPT')
# Set Options for solver.
#opt.options['bonmin.solution_limit'] = '1'
#opt.options['bonmin.time_limit'] = 1800
results = opt.solve(epi_model_ESH, tee = True)
results.write()
using IPOPT
ERROR: Variable 'x1' is not part of the model being written out, but appears
in an expression used on this model.
ERROR: Variable 'x2' is not part of the model being written out, but appears
in an expression used on this model.
---------------------------------------------------------------------------
KeyError Traceback (most recent call last)
<ipython-input-22-dcc4023897c3> in <module>
5 #opt.options['bonmin.solution_limit'] = '1'
6 #opt.options['bonmin.time_limit'] = 1800
----> 7 results = opt.solve(epi_model_ESH, tee = True)
8 results.write()
~\Anaconda3\envs\seminarorsteinss2019\lib\site-packages\pyomo\opt\base\solvers.py in solve(self, *args, **kwds)
594 initial_time = time.time()
595
--> 596 self._presolve(*args, **kwds)
597
598 presolve_completion_time = time.time()
~\Anaconda3\envs\seminarorsteinss2019\lib\site-packages\pyomo\opt\solver\shellcmd.py in _presolve(self, *args, **kwds)
194 self._keepfiles = kwds.pop("keepfiles", False)
195
--> 196 OptSolver._presolve(self, *args, **kwds)
197
198 #
~\Anaconda3\envs\seminarorsteinss2019\lib\site-packages\pyomo\opt\base\solvers.py in _presolve(self, *args, **kwds)
691 self._problem_format,
692 self._valid_problem_formats,
--> 693 **kwds)
694 total_time = time.time() - write_start_time
695 if self._report_timing:
~\Anaconda3\envs\seminarorsteinss2019\lib\site-packages\pyomo\opt\base\solvers.py in _convert_problem(self, args, problem_format, valid_problem_formats, **kwds)
762 valid_problem_formats,
763 self.has_capability,
--> 764 **kwds)
765
766 def _default_results_format(self, prob_format):
~\Anaconda3\envs\seminarorsteinss2019\lib\site-packages\pyomo\opt\base\convert.py in convert_problem(args, target_problem_type, valid_problem_types, has_capability, **kwds)
108 tmpkw = kwds
109 tmpkw['capabilities'] = has_capability
--> 110 problem_files, symbol_map = converter.apply(*tmp, **tmpkw)
111 return problem_files, ptype, symbol_map
112
~\Anaconda3\envs\seminarorsteinss2019\lib\site-packages\pyomo\solvers\plugins\converter\model.py in apply(self, *args, **kwds)
190 format=args[1],
191 solver_capability=capabilities,
--> 192 io_options=io_options)
193 return (problem_filename,), symbol_map_id
194 else:
~\Anaconda3\envs\seminarorsteinss2019\lib\site-packages\pyomo\core\base\block.py in write(self, filename, format, solver_capability, io_options)
1645 filename,
1646 solver_capability,
-> 1647 io_options)
1648 smap_id = id(smap)
1649 if not hasattr(self, 'solutions'):
~\Anaconda3\envs\seminarorsteinss2019\lib\site-packages\pyomo\repn\plugins\ampl\ampl_.py in __call__(self, model, filename, solver_capability, io_options)
390 skip_trivial_constraints=skip_trivial_constraints,
391 file_determinism=file_determinism,
--> 392 include_all_variable_bounds=include_all_variable_bounds)
393
394 self._symbolic_solver_labels = False
~\Anaconda3\envs\seminarorsteinss2019\lib\site-packages\pyomo\repn\plugins\ampl\ampl_.py in _print_model_NL(self, model, solver_capability, show_section_timing, skip_trivial_constraints, file_determinism, include_all_variable_bounds)
959 ampl_repn,
960 list(self_varID_map[id(var)] for var in ampl_repn._linear_vars),
--> 961 list(self_varID_map[id(var)] for var in ampl_repn._nonlinear_vars))
962 except KeyError as err:
963 self._symbolMapKeyError(err, model, self_varID_map,
~\Anaconda3\envs\seminarorsteinss2019\lib\site-packages\pyomo\repn\plugins\ampl\ampl_.py in <genexpr>(.0)
959 ampl_repn,
960 list(self_varID_map[id(var)] for var in ampl_repn._linear_vars),
--> 961 list(self_varID_map[id(var)] for var in ampl_repn._nonlinear_vars))
962 except KeyError as err:
963 self._symbolMapKeyError(err, model, self_varID_map,
KeyError: (200822968, "Variable 'x1' is not part of the model being written out, but appears in an expression used on this model.", "Variable 'x2' is not part of the model being written out, but appears in an expression used on this model.")
The pprint() of the new model still lists x1 and x2 as variables.
Did me using constraint._body = ... cause this?
I found the mistake.
In
epi_nonlinear_constrs[k]._body = (epi_nonlinear_constrs[k].body - epi_model.alpha_epi)
I used the bodies of the constraints of the other model nonlinear_constrs[k].body instead of those of the same model. Therefore, the constraints had variables which were not referenced in the model. Hence, the error message from the solver.