Problem

class flopt.problem.Problem(name=None, sense=OptimizationType.Minimize)

Interface between User and Solver

Parameters

• name (str) – name of problem

• sense (str, optional) – minimize, maximize (future satisfiability is added)

name

name of problem

Type

str

sense

minimize, maximize (future satisfiability is added)

Type

str, optional

obj

Type

Expression family

solver

Type

Solver or None

time

solving time

Type

float

Examples

>>> prob = Problem(name='test')

When we want to solve the maximize problem, then

>>> prob = Problem(name='test', sense='maximize')

We solve

>>> prob.solve(solver=solver_name or solver object, timelimit=10)

After solving, we can obtain the objective value.

>>> prob.getObjectiveValue()

addConstraint(const, name=None)

add constraint into problem. __iadd__(), “+=” operations call this function.

Parameters

• const (Constraint) – constraint

• name (str or None) – constraint name

Examples

import flopt
prob = flopt.Problem(algo=...)
x = flopt.Variable("x")
y = flopt.Variable("y")
prob.addConstraint(x + y >= 2)

boundsToIneq()

Create a problem object has bounds constraints of variables as inequal constraints

Returns

prob

Return type

Problem

clone(variable_clone=False)

create clone object :param variable_clone: if it is true, variables are cloned in expression :type variable_clone: bool

Returns

prob

Return type

Problem

getConstraints()

Returns

list of constraints in this problem

Return type

list of Constraint

getObjectiveValue()

Returns

the objective value

Return type

float or int

getSolution(k=1)

get the k-top solution

Parameters

k (int) – return k-top solution

getVariables()

Returns

set of VarElement used in this problem

Return type

set

removeDuplicatedConstraints()

Remove duplicated constraints in problem

Examples

import flopt
a = flopt.Variable("a")
b = flopt.Variable("b")
c = flopt.Variable("c")

prob = flopt.Problem(name="Test")
prob += a + b >= 0
prob += a + b >= 0
prob += a >= -b
prob += 0 >= -a - b
prob += Sum([a, b]) >= 0

len(prob.constraints)
>>> 5

prob.removeDuplicatedConstraints()
len(prob.constraints)
>>> 1

replace(correspondence_dict)

Replace variable to another variables or expression

Parameters

correspondence_dict (dict) – key is Variable and value is Variable or ExpressionElement

Examples

import flopt

# create problem
x = flopt.Variable("x", lowBound=4, upBound=5)
prob = flopt.Problem()
prob += x
prob.show()
>>> Name: None
>>> Type         : Problem
>>> sense        : Minimize
>>> objective    : x+0
>>> #constraints : 0
>>> #variables   : 1 (Continuous 1)
>>>
>>>
>>> V 0, name x, Continuous 4 <= x <= 5

# convert bounds of variables to constraints
prob = prob.clone().boundsToIneq()
prob.show()
>>> Name: None
>>>   Type         : Problem
>>>   sense        : Minimize
>>>   objective    : x+0
>>>   #constraints : 2
>>>   #variables   : 1 (Continuous 1)
>>>
>>>   C 0, name None, 4-x <= 0
>>>   C 1, name None, x-5 <= 0
>>>
>>>   V 0, name x, Continuous None <= x <= None

# replace x with x_plus + x_minus
x_plus = flopt.Variable("x_plus", lowBound=0)
x_minus = flopt.Variable("x_minus", lowBound=0)
prob.replace(correspondence_dict={x: x_plus + x_minus})
prob.show()
>>> Name: None
>>>   Type         : Problem
>>>   sense        : Minimize
>>>   objective    : x_plus+x_minus
>>>   #constraints : 2
>>>   #variables   : 2 (Continuous 2)
>>>
>>>   C 0, name None, 4-(x_plus+x_minus) <= 0
>>>   C 1, name None, x_plus+x_minus-5 <= 0
>>>
>>>   V 0, name x_minus, Continuous 0 <= x_minus <= None
>>>   V 1, name x_plus, Continuous 0 <= x_plus <= None

setBestBound(best_bound)

Parameters

best_bound (float) – best objective value of this problem

setObjective(obj, name=None)

set objective function. __iadd__(), “+=” operations call this function.

Parameters

obj (int, float, Variable family or Expression family) – objective function

setSolution(k=1)

set the k-top solution to variables

Parameters

k (int) – set k-top solution data to variables

solve(solver=None, timelimit=None, lowerbound=None, optimized_variables=None, msg=False, **kwargs)

solve this problem

Parameters

• solver (Solver or None) –

• timelimit (float or None) –

• lowerbound (float or None) – solver terminates when it obtains the solution whose objective value is lower than this value

• optimized_variables (None or list, tuple, np.ndarray or any container of Variable) – if it is specified, solver will optimize only the variables in optimized_variables

• msg (bool) – if true, display the message from solver

Returns

• Status – return the status of solving

• Log – return log object

Examples

import flopt
a = flopt.Variable("a")
b = flopt.Variable("b")
c = flopt.Variable("c")

prob = flopt.Problem(name="Test")
prob += a + b
prob += a + b >= 0

solver = flopt.Solver("auto")
status, logs = prob.solve(solver=solver)

When user want to optimize a part of variables under otherwise variables are fixed, user specify optmized_variables in problem.solve().

# optimize only a
status, log = prob.solve(optimized_variables=[a], timelimit=1)

toEq()

Create a problem object with only equal constraints

Returns

prob

Return type

Problem

toIneq()

Create a problem object with only inequal constraints

Returns

prob

Return type

Problem

toProblemType()

Returns

problem_type – key is “Variable”, “Objective”, “Constraint”

Return type

dict

toRelax()

Create relaxed problem

Returns

prob

Return type

Problem