Lesson_05_Pyomo_Mutable_Parameters.ipynb

Kernel: Python 3 (system-wide)

Pyomo Mutable Parameters

Sometimes when you are coding a model, you want to be able to run it multiple times with some slight changes. For example, you might want to vary a constraint. Lesson 4 homework has a problem that requires this. There are two ways to do that:

1. Wrap the model in a function, using the bit you want to change as a parameter to the function, and re-instantiate the entire model for each value you want to test.

2. Use a mutable parameter.

Mutable just means that it's a paramter value that can change, instead of being a fixed constant. Let's see a quick example of each. We'll use our old standby - the Wyndor model. We'll vary the constant in the constraint, so that instead of just using 18, we'll try the model with 18, 20, and 22.

Wrapping in a function

# unfold to see Pyomo solution with a vector of decision variables
from pyomo.environ import *
import babel.numbers as numbers  # needed to display as currency

############################
# NEW: we're defining a function that takes 1 parameter - the constraint value
############################
def wyndor(c):
    # Concrete Model - all of this is exactly the same as whatyou did before, except using the parameter in the constraint
    model = ConcreteModel(name="Wyndor")

    products = ['drs', 'wdw']

    bounds_dict = {'drs': (0, 4), 'wdw': (0, 6)}
    def bounds_rule(model, product):
        return (bounds_dict[product])
    model.x = Var(products, domain=Reals, bounds=bounds_rule)

    # Objective
    model.profit = Objective(expr=3.0 * model.x['drs'] + 5.0 * model.x['wdw'],
                             sense=maximize)

    # Constraints
    ######################################
    #NEW: Here's where we use the parameter
    ######################################
    model.Constraint3 = Constraint(
        expr=3.0 * model.x['drs'] + 2.0 * model.x['wdw'] <= c) 

    # Solve
    solver = SolverFactory('glpk')
    solver.solve(model)
    return model

##########################################
# NEW: We'll loop over the 3 values we want to model and call our function for each one
##########################################
for c in [18,20,22]:
    model = wyndor(c)
    # display solution
    
    print("Using Constraint", c)
    print("Profit = ",
          numbers.format_currency(1000 * model.profit(), 'USD', locale='en_US'))
    print("Batches of Doors = ", model.x['drs']())
    print("Batches of Windows = ", model.x['wdw']())
    print("-------------------------------------------------")

Using Constraint 18
Profit =  $36,000.00
Batches of Doors =  2.0
Batches of Windows =  6.0
-------------------------------------------------
Using Constraint 20
Profit =  $38,000.00
Batches of Doors =  2.66666666666667
Batches of Windows =  6.0
-------------------------------------------------
Using Constraint 22
Profit =  $40,000.00
Batches of Doors =  3.33333333333333
Batches of Windows =  6.0
-------------------------------------------------

Mutable Parameter

 # Concrete Model - Not wrapped in a function
model = ConcreteModel(name="Wyndor")

products = ['drs', 'wdw']
############################################
# NEW: This is our mutable parameter
model.c = Param(default=18, mutable=True)
############################################
bounds_dict = {'drs': (0, 4), 'wdw': (0, 6)}
def bounds_rule(model, product):
    return (bounds_dict[product])
model.x = Var(products, domain=Reals, bounds=bounds_rule)

# Objective
model.profit = Objective(expr=3.0 * model.x['drs'] + 5.0 * model.x['wdw'],
                         sense=maximize)

# Constraints
######################################
#NEW: Here's where we use the parameter
######################################
model.Constraint3 = Constraint(
    expr=3.0 * model.x['drs'] + 2.0 * model.x['wdw'] <= model.c) 

# Solve
solver = SolverFactory('glpk')
    
##################################
#NEW: Loop over our constraints
##################################

for c in [18,20,22]:
    model.c = c
    solver.solve(model)
    print("Using Constraint", c)
    print("Profit = ",
          numbers.format_currency(1000 * model.profit(), 'USD', locale='en_US'))
    print("Batches of Doors = ", model.x['drs']())
    print("Batches of Windows = ", model.x['wdw']())
    print("-------------------------------------------------")
    

Using Constraint 18
Profit =  $36,000.00
Batches of Doors =  2.0
Batches of Windows =  6.0
-------------------------------------------------
Using Constraint 20
Profit =  $38,000.00
Batches of Doors =  2.66666666666667
Batches of Windows =  6.0
-------------------------------------------------
Using Constraint 22
Profit =  $40,000.00
Batches of Doors =  3.33333333333333
Batches of Windows =  6.0
-------------------------------------------------