I'm programming a optimization in pyomo.environ for timeseries. Now I have the need for using piecewise function. Using the

pyomo.environ.Piecewise(...)

I get the error

TypeError: Cannot apply a Set operator to an indexed Var component (I)

To find a solution e reduzed the model to a minimized version:

import pyomo.environ as pe
import numpy as np

# Modell erstellen
model = pe.ConcreteModel()

model.steps = range(0,10)

# Variablen
model.P = pe.Var(model.steps, within=pe.Reals)  # P kann positiv oder negativ sein
model.I = pe.Var(model.steps )  # I(P) als abhängige Variable

# Parameter
model.a1 = pe.Param(initialize=1)  # Beispielwerte
model.m1 = pe.Param(initialize=2)
model.a2 = pe.Param(initialize=-1)
model.m2 = pe.Param(initialize=3)

for i in range(0, 10):
    # Stückweise lineare Funktion definieren    
    points_p = np.array([-10000, -0.01, 0.01, 10000])
    points_i = np.where(points_p > 0, 0.01 + 0.5 * points_p / 48,
                        0.01 + 0.45 * points_p / 48)

    model.piecewise_pi = pe.Piecewise(model.I, model.P, list(points_p), list(points_i),
                                        pw_constr_type='EQ')

# Zielfunktion (optional)
model.obj = pe.Objective(expr=model.I, sense=pe.minimize)

# Modell lösen
solver = pe.SolverFactory('cplex')
solver.solve(model)

I allready made a test with loop over the indext Variables:

for i in range(0, 10):
    # Stückweise lineare Funktion definieren    
    points_p = np.array([-10000, -0.01, 0.01, 10000])
    points_i = np.where(points_p > 0, 0.01 + 0.5 * points_p / 48,
                        0.01 + 0.45 * points_p / 48)

    model.piecewise_pi = pe.Piecewise(model.I[i], model.P[i], list(points_p), list(points_i),
                                        pw_constr_type='EQ')

but it still leads to te same error:

TypeError: Cannot apply a Set operator to a non-Set component data (I[0])

Without the dimension of the 11 steps the code is running.

Maybe that there is a solution without usingPiecewise in this case, but I also ned the approximation of a 6th degree polynomial within the code and therefore I need the Piecewise-function.

Does anyone have a hint for me, how to treat with it?

I'm programming a optimization in pyomo.environ for timeseries. Now I have the need for using piecewise function. Using the

pyomo.environ.Piecewise(...)

I get the error

TypeError: Cannot apply a Set operator to an indexed Var component (I)

To find a solution e reduzed the model to a minimized version:

import pyomo.environ as pe
import numpy as np

# Modell erstellen
model = pe.ConcreteModel()

model.steps = range(0,10)

# Variablen
model.P = pe.Var(model.steps, within=pe.Reals)  # P kann positiv oder negativ sein
model.I = pe.Var(model.steps )  # I(P) als abhängige Variable

# Parameter
model.a1 = pe.Param(initialize=1)  # Beispielwerte
model.m1 = pe.Param(initialize=2)
model.a2 = pe.Param(initialize=-1)
model.m2 = pe.Param(initialize=3)

for i in range(0, 10):
    # Stückweise lineare Funktion definieren    
    points_p = np.array([-10000, -0.01, 0.01, 10000])
    points_i = np.where(points_p > 0, 0.01 + 0.5 * points_p / 48,
                        0.01 + 0.45 * points_p / 48)

    model.piecewise_pi = pe.Piecewise(model.I, model.P, list(points_p), list(points_i),
                                        pw_constr_type='EQ')

# Zielfunktion (optional)
model.obj = pe.Objective(expr=model.I, sense=pe.minimize)

# Modell lösen
solver = pe.SolverFactory('cplex')
solver.solve(model)

I allready made a test with loop over the indext Variables:

for i in range(0, 10):
    # Stückweise lineare Funktion definieren    
    points_p = np.array([-10000, -0.01, 0.01, 10000])
    points_i = np.where(points_p > 0, 0.01 + 0.5 * points_p / 48,
                        0.01 + 0.45 * points_p / 48)

    model.piecewise_pi = pe.Piecewise(model.I[i], model.P[i], list(points_p), list(points_i),
                                        pw_constr_type='EQ')

but it still leads to te same error:

TypeError: Cannot apply a Set operator to a non-Set component data (I[0])

Without the dimension of the 11 steps the code is running.

Maybe that there is a solution without usingPiecewise in this case, but I also ned the approximation of a 6th degree polynomial within the code and therefore I need the Piecewise-function.

Does anyone have a hint for me, how to treat with it?

• python
• pyomo
• piecewise
• environ

1 Answer 1

The pyomo.environ.Piecewise component is meant for defining a single piecewise function between two individual (scalar) variables. The error arises because you're attempting to apply it to an indexed variable (model.I) or a specific element of it (model.I[i]).

You need to define a separate Piecewise constraint for each time step. You can define the piecewise function parameters (points_p and points_i) outside the loop.

model.steps = range(0, 10)
# Piecewise constraints for each time step
model.piecewise_pi = pe.ConstraintList()
for i in model.steps:
    model.piecewise_pi.add(
        pe.Piecewise(
            model.I[i],
            model.P[i],
            list(points_p),
            list(points_i),
            pw_constr_type='EQ'
        )
    )

For each time step i in model.steps, we create a Piecewise constraint that links model.I[i] and model.P[i] specifically for that time. This individual constraint is then added to the model.piecewise_pi ConstraintList.