from abc import ABC,abstractmethod
import os
from gurobipy import Model, GRB, quicksum
import tempfile
class OptimizationProblem(ABC):
def __init__(self,problem:dict):
self._name = problem["problem"].get("name", "OptimizationProblem")
self._problem = problem
self._objective = problem.get("objective", {})
self._constraints = problem.get("constraints", [])
self._variables = problem.get("variables",{})
self._gurobi_variables = {}
self._model = None
self._solution = None
self._status = None
self._create_model() # Call the method to create the model
@abstractmethod
def _create_model(self):
pass
def getModel(self):
return self._model
def getProblemAsLP(self) -> str:
"""
Write the LP problem to a file in LP format.
"""
if self._model is not None:
with tempfile.NamedTemporaryFile(delete=False, suffix=".lp",mode='w+') as tmp:
try:
self._model.write(tmp.name)
tmp.seek(0) # Move to the beginning of the file
lp_string = tmp.read()
finally:
tmp.close()
os.unlink(tmp.name)
return lp_string
else:
return "Model is not created. Cannot write problem to file."
def _addLinearConstraints(self) -> bool:
if self._model is None:
raise ValueError("Model is not created. Cannot add constraints.")
default_constr_index = 0
linear_constraints = self._constraints.get("linear_constraints", [])
for constraint in linear_constraints:
lhs_expr = quicksum(var["coef"] * self._gurobi_variables[var["var"]] for var in constraint["lhs"])
if constraint["sign"] == "<=":
self._model.addConstr(lhs_expr <= constraint["rhs"], name=constraint.get("name", "Constraint_" + str(default_constr_index)))
elif constraint["sign"] == "<":
self._model.addConstr(lhs_expr < constraint["rhs"], name=constraint.get("name", "Constraint_" + str(default_constr_index)))
elif constraint["sign"] == ">=":
self._model.addConstr(lhs_expr >= constraint["rhs"], name=constraint.get("name", "Constraint_" + str(default_constr_index)))
elif constraint["sign"] == ">":
self._model.addConstr(lhs_expr >= constraint["rhs"], name=constraint.get("name", "Constraint_" + str(default_constr_index)))
elif constraint["sign"] == "=":
self._model.addConstr(lhs_expr == constraint["rhs"], name=constraint.get("name", "Constraint_" + str(default_constr_index)))
else:
raise ValueError(f"Unknown sign {constraint['sign']} in constraint: {constraint.get('name', '')}")
default_constr_index += 1
self._model.update()
return True
def _addQuadraticConstraints(self):
if self._model is None:
raise ValueError("Model is not created. Cannot add quadratic constraints.")
counter=0
quadratic_constraints = self._constraints.get("quadratic_constraints", [])
for constraint in quadratic_constraints:
quad_terms = constraint.get("quadratic_terms", [])
linear_terms = constraint.get("linear_terms", [])
constant = constraint.get("constant", 0)
lhs_expr = quicksum(term["coef"] * self._gurobi_variables[term["var1"]] * self._gurobi_variables[term["var2"]] for term in quad_terms)
lhs_expr += quicksum(var["coef"] * self._gurobi_variables[var["var"]] for var in linear_terms)
lhs_expr += constant
if constraint["sign"] == "<=":
self._model.addQConstr(lhs_expr <= 0, name=constraint.get("name", "QuadraticConstraint_" + str(counter)))
elif constraint["sign"] == "<":
self._model.addQConstr(lhs_expr < 0, name=constraint.get("name", "QuadraticConstraint_" + str(counter)))
elif constraint["sign"] == ">=":
self._model.addQConstr(lhs_expr >= 0, name=constraint.get("name", "QuadraticConstraint_" + str(counter)))
elif constraint["sign"] == ">":
self._model.addQConstr(lhs_expr > 0, name=constraint.get("name", "QuadraticConstraint_" + str(counter)))
elif constraint["sign"] == "=":
self._model.addQConstr(lhs_expr == 0, name=constraint.get("name", "QuadraticConstraint_" + str(counter)))
else:
raise ValueError(f"Unknown sign {constraint['sign']} in quadratic constraint: {constraint.get('name', '')}")
counter += 1
self._model.update()
return True
def _addVariables(self):
if self._model is None:
raise ValueError("Model is not created. Cannot add variables.")
for var in self._variables:
"""
Define all the variables in the model.
Each variable is defined with its type, name, lower bound (lb), upper bound (ub).
Objective coefficient is not included in the variable definition.
The type, name, lb, and ub parameters are optional.
"""
var_info = var
var_type = GRB.CONTINUOUS
#Check if var type is in info
if var_info.get("type")=="binary":
var_type = GRB.BINARY
elif var_info.get("type")=="integer":
var_type = GRB.INTEGER
elif var_info.get("type")=="semicontinuous":
var_type = GRB.SEMICONTINUOUS
elif var_info.get("type")=="semidefinite":
var_type = GRB.SEMIDEFINITE
add_var_parameters = {
"name": var_info.get("name", var),
"vtype": var_type
}
if "lb" in var_info and (type(var_info["lb"]) is int or type(var_info["lb"]) is float):
add_var_parameters["lb"] = var_info["lb"]
if "ub" in var_info and (type(var_info["ub"]) is int or type(var_info["ub"]) is float):
add_var_parameters["ub"] = var_info["ub"]
self._gurobi_variables[var_info["name"]] = self._model.addVar(**add_var_parameters)
self._model.update()
def _addLinearObjective(self):
if self._model is None:
raise ValueError("Model is not created. Cannot add objective.")
# Set the objective function
objective = self._objective
if objective["function_type"] != "linear":
raise ValueError("Objective function type must be linear for LP problems. Use QP or QCP for quadratic objectives.")
linear_coeffs = objective["linear_terms"]
if not linear_coeffs:
raise ValueError("Linear terms must be provided for the objective function.")
sum_list = []
for var in linear_coeffs:
name = var["var"]
coeff = var["coef"]
if name not in self._gurobi_variables:
raise ValueError(f"Variable {name} not found in model variables.")
sum_list.append(coeff * self._gurobi_variables[name])
obj_expr = quicksum(sum_list)
if objective["type"]=="maximize":
self._model.setObjective(obj_expr, GRB.MAXIMIZE)
elif objective["type"]=="minimize":
self._model.setObjective(obj_expr, GRB.MINIMIZE)
else:
raise ValueError(f"Unknown objective function type: {objective['type']}")
self._model.update()
def _addQuadraticObjective(self):
if self._model is None:
raise ValueError("Model is not created. Cannot add objective.")
# Set the objective function
objective = self._objective
if objective["function_type"] != "quadratic":
raise ValueError("Objective function type must be quadratic for QP problems. Use LP for linear objectives.")
linear_coeffs = objective["linear_terms"]
sum_list = []
for var in linear_coeffs:
name = var["var"]
coeff = var["coef"]
if name not in self._gurobi_variables:
raise ValueError(f"Variable {name} not found in model variables.")
sum_list.append(coeff * self._gurobi_variables[name])
obj_expr = quicksum(sum_list)
obj_expr += quicksum(term["coef"] * self._gurobi_variables[term["var1"]] * self._gurobi_variables[term["var2"]] for term in objective["quadratic_terms"])
if objective["type"]=="maximize":
self._model.setObjective(obj_expr, GRB.MAXIMIZE)
elif objective["type"]=="minimize":
self._model.setObjective(obj_expr, GRB.MINIMIZE)
else:
raise ValueError(f"Unknown objective function type: {objective['type']}")
self._model.update()
