Pyomo: How do I add all set of a model's constraints to another model?

Question

I want to add all constraints and variables from a model to a different model in Pyomo, but I cannot figure out how. The use case is when I want to transform my model to its dual, and I need to add all dual constraints and variables to the primal problem. This is actually useful when we want to add optimality conditions of a certain model to another problem.

Maybe there are other functional transformations in Pyomo that do the same, and I am not aware of them, so in a case such functionality exists, I would be more than happy if anybody could assist.

Thanks,

Do you actually want to add the constraint and variable objects or do you want to add their computed values into the constraints of another model? — cookesd, Jan 22 '21 at 23:39
I want the actual constraints and variables to be added to the other problem, not their value. — AliRa, Jan 23 '21 at 01:55

score 0 · Answer 1 · answered Jan 23 '21 at 16:00

If you want to copy the constraints and variables to a brand new model, you can use the clone method of the model then remove any un-required model objects (objective functions, parameters, etc.)

import pyomo.environ as pyo

#%% Build initial model
my_set = [1,2,3]
m1 = pyo.ConcreteModel()
m1.x = pyo.Var(my_set,within=pyo.NonNegativeReals)
m1.y = pyo.Var(within=pyo.Binary)
m1.con1 = pyo.Constraint(expr = sum([m1.x[i] for i in my_set]) <= 3)
m1.obj = pyo.Objective(expr = m1.y + sum([m1.x[i] for i in my_set]),
                       sense=-1)

#%% Solve initial model
solver = pyo.SolverFactory('glpk')
res1 = solver.solve(m1)

#%% Clone initial model
m2 = m1.clone()

#%% Verify objects copied to other model
m1.pprint()

# 1 Set Declarations
#     x_index : Dim=0, Dimen=1, Size=3, Domain=None, Ordered=False, Bounds=(1, 3)
#         [1, 2, 3]

# 2 Var Declarations
#     x : Size=3, Index=x_index
#         Key : Lower : Value : Upper : Fixed : Stale : Domain
#           1 :     0 :   3.0 :  None : False : False : NonNegativeReals
#           2 :     0 :   0.0 :  None : False : False : NonNegativeReals
#           3 :     0 :   0.0 :  None : False : False : NonNegativeReals
#     y : Size=1, Index=None
#         Key  : Lower : Value : Upper : Fixed : Stale : Domain
#         None :     0 :   1.0 :     1 : False : False : Binary

# 1 Objective Declarations
#     obj : Size=1, Index=None, Active=True
#         Key  : Active : Sense    : Expression
#         None :   True : maximize : x[1] + x[2] + x[3] + y

# 1 Constraint Declarations
#     con1 : Size=1, Index=None, Active=True
#         Key  : Lower : Body               : Upper : Active
#         None :  -Inf : x[1] + x[2] + x[3] :   3.0 :   True

# 5 Declarations: x_index x y con1 obj

m2.pprint()
# 1 Set Declarations
#     x_index : Dim=0, Dimen=1, Size=3, Domain=None, Ordered=False, Bounds=(1, 3)
#         [1, 2, 3]

# 2 Var Declarations
#     x : Size=3, Index=x_index
#         Key : Lower : Value : Upper : Fixed : Stale : Domain
#           1 :     0 :   3.0 :  None : False : False : NonNegativeReals
#           2 :     0 :   0.0 :  None : False : False : NonNegativeReals
#           3 :     0 :   0.0 :  None : False : False : NonNegativeReals
#     y : Size=1, Index=None
#         Key  : Lower : Value : Upper : Fixed : Stale : Domain
#         None :     0 :   1.0 :     1 : False : False : Binary

# 1 Objective Declarations
#     obj : Size=1, Index=None, Active=True
#         Key  : Active : Sense    : Expression
#         None :   True : maximize : x[1] + x[2] + x[3] + y

# 1 Constraint Declarations
#     con1 : Size=1, Index=None, Active=True
#         Key  : Lower : Body               : Upper : Active
#         None :  -Inf : x[1] + x[2] + x[3] :   3.0 :   True

# 5 Declarations: x_index x y con1 obj

#%% Remove objective if desired
for obj in m2.component_objects(pyo.Objective):
    m2.del_component(obj)

#%% See that objective is removed
m2.pprint()

# 1 Set Declarations
#     x_index : Dim=0, Dimen=1, Size=3, Domain=None, Ordered=False, Bounds=(1, 3)
#         [1, 2, 3]

# 2 Var Declarations
#     x : Size=3, Index=x_index
#         Key : Lower : Value : Upper : Fixed : Stale : Domain
#           1 :     0 :   3.0 :  None : False : False : NonNegativeReals
#           2 :     0 :   0.0 :  None : False : False : NonNegativeReals
#           3 :     0 :   0.0 :  None : False : False : NonNegativeReals
#     y : Size=1, Index=None
#         Key  : Lower : Value : Upper : Fixed : Stale : Domain
#         None :     0 :   1.0 :     1 : False : False : Binary

# 1 Constraint Declarations
#     con1 : Size=1, Index=None, Active=True
#         Key  : Lower : Body               : Upper : Active
#         None :  -Inf : x[1] + x[2] + x[3] :   3.0 :   True

# 4 Declarations: x_index x y con1

Thanks, I think you forgot the main issue I have! I would like to add the constraints and variables of a problem to another well-defined model (with its own constraints and variables), not cloning my first model and renaming it. But I think I took what I wanted, I could iterate over all Vars or Constraints of the initial problem and use add_component to add them to the another model, is that right? — AliRa, Jan 23 '21 at 19:15
I wasn't sure if you wanted to add it to an already defined model. The add_component seemed to work for the constraints but I couldn't quite get it to work for the variables since they're tied to the first model. If you're able to get that working that would be good to see — cookesd, Jan 23 '21 at 22:00
You're right, I couldn't add anything using add_component yet, I first thought it might handle new variables by itself but then it kept asking for a value as this function requires two entries. I needed this to somehow make the dual problem from a primal problem and then add the constraints of each of these models to another model (optimality conditions) — AliRa, Jan 23 '21 at 22:32

score 0 · Accepted Answer · answered Jan 24 '21 at 08:13

I added the constraints of a model "model2" to another model "model1" with first deleting the item from the initial model and then adding it to the other.

Here is a sample code to show that:

from pyomo.environ import *

model1 = ConcreteModel()
model2 = ConcreteModel()

model1.a = Var([1,2,3,4,5,6], within=NonNegativeReals)
model2.b = Var([1,2,3,4,5,6], within=NonNegativeReals)

model1.c = Constraint(expr= model1.a[1] + model1.a[3] == 3)
model2.d = Constraint(expr= model2.b[1] + model2.b[3] == 3)

  
for con in model2.component_objects(Constraint):
    model2.del_component(con)
    model1.add_component('d', con)

for var in model2.component_objects(Var):
    model2.del_component(var)
    model1.add_component('b', var)

model1.objective = Objective(expr = model1.a[1] + model1.b[3], sense=minimize)

solver = SolverFactory('glpk')
res = solver.solve(model1)

print(model1.display())

Pyomo: How do I add all set of a model's constraints to another model?

2 Answers2