Model a binary variable from two continous variables
AnsweredI want to model a binary value (b) from two continous variables (v_in, v_out). These variables are 0<=v_in, v_out <= v_max.
I want the following three conditions to apply on my binary variable
- v_in>0 & v_out=0 ==> b=1,
- v_in=0 & v_out>0 ==> b=0,
- v_in=0 & v_out=0 ==> b=0.
I have tried the following:
- v_in = b*v_max
- v_out = (1-b)v_max
The above solution works for 1. and 2. but it is nearly always wrong for 3.
Can you please help?
Thanks
-
Official comment
-
Gurobi Staff
This post is more than three years old. Some information may not be up to date. For current information, please check the Gurobi Documentation or Knowledge Base. If you need more help, please create a new post in the community forum, or try Gurobot, our chatbot interface offering instant, expert-level support. -
-
-
Your 3 conditions can be reformulated as
v_in > 0 => b=1 and b=0 else.
This is only if you are OK with the fact that if v_in > 0 & v_out > 0 then b=1.
Modeling such a conditional statement is discussed in the Knowledge Base article How do I model conditional statements in Gurobi?
0 -
-
Thanks Jaromil,
v_in>0 and v_out>0 should not be possible. Only one of these values can be positive.
0 -
-
-
If v_in>0 and v_out>0 is not possible then your 3 conditions can indeed be formulated as
v_in > 0 => b=1 and b=0 else and you can follow the instructions in Knowledge Base article How do I model conditional statements in Gurobi?
0 -
-
I read the article and I followed it. I still don't get the correct answer. The issue might arise from other parts of the model.
I have prepared the following minimal example. It is a basic storage model built mostly on the power generation example on Gurobi case study.
Can you please have a look and let me know if you would change it.
x_vol_init = 0
x_vol_min = 0
x_vol_max = 10
phy_cost = 50
x_max = 5
x_max_mirror = -x_max
n_step = 3
df = pd.read_csv('input_data.csv', parse_dates=['ts'])
# Import Constants
const_out = df['input_out'].values
const_in = df['input_in'].values
m = gp.Model('model_1')
# Define continuous variables
x_out = m.addVars(n_step, vtype=GRB.CONTINUOUS, ub=x_max)
x_in = m.addVars(n_step, vtype=GRB.CONTINUOUS, ub=x_max)
# Define boolean variable
bln = m.addVars(n_step, vtype=GRB.BINARY)
# Define other variables
x_vol = m.addVars(n_step, vtype=GRB.CONTINUOUS, ub=x_vol_max, lb=x_vol_min)
x_net = m.addVars(n_step, vtype=GRB.CONTINUOUS, ub=x_max, lb=x_max_mirror)
# From https://support.gurobi.com/hc/en-us/articles/4414392016529-How-do-
I-model-conditional-statements-in-Gurobi-
eps = 0.00001
m.addConstrs(x_out[t] >= x_in[t] + eps - x_max * (1 - bln[t]) for t in range(n_step))
m.addConstrs(x_out[t] <= x_in[t] + x_max * bln[t] for t in range(n_step))
# Other constraints
m.addConstrs(x_net[t] <= x_max for t in range(n_step))
m.addConstrs(x_net[t] >= x_max_mirror for t in range(n_step))
m.addConstrs(x_net[t] == x_out[t] - x_in[t] for t in range(n_step))
# Volume related constraints
m.addConstrs(x_vol[t] <= x_vol_max for t in range(n_step))
m.addConstrs(x_vol[t] >= x_vol_min for t in range(n_step))
m.addConstrs(x_vol[0] == x_vol_init + x_in[0] - x_out[0])
m.addConstrs(x_vol[t] == x_vol[t - 1] + x_in[t] - x_out[t] for t in range(1, n_step))
# Calculate the objective fucntion
OF = gp.quicksum(
(x_out[t] * const_out[t]) - (x_in[t] * const_in[t]) - (bln[t] * (x_net[t] * phy_cost)) for t in range(n_step)
)
m.setObjective(OF, GRB.MAXIMIZE)
m.optimize()0 -
-
-
It is not possible to reproduce your example due to missing input_data.csv file. You could try hard coding a small data set which still shows the wrong behavior. Please then also elaborate on what exactly is going wrong with the model and what you would expect to happen.
0 -
-
I found the issue. It has to do with including another continous variable in the objective function (x_net) which results in forcing incorrect values on b_in and b_out.
Thanks for your support Jaromil.
Best,
0 -
Post is closed for comments.
Comments
7 comments