Skip to content

Latest commit

 

History

History
191 lines (124 loc) · 5.07 KB

Variables.md

File metadata and controls

191 lines (124 loc) · 5.07 KB
Raw

Constrained Variables

The solver supports finite domain integer constrained variables. The class that implements them is called NsIntVar and contains the following methods.

NsIntVar(NsProblemManager& pm, NsInt min, NsInt max)

A constructor function for a constraint variable. Argument pm is the problem manager that the variable belongs to; see the Problem Manager section for more. min and max are the bounds of its domain, that is also designated [min..max].

Data-type NsInt can at least represent the integers that can be represented by data-type long. The minimum value that an NsInt can hold, equals to the constant NsMINUS_INF, and the maximum equals to NsPLUS_INF. (The maximum value of the unsigned data-type NsUInt is NsUPLUS_INF.)

The minimum of a domain must be strictly greater than NsMINUS_INF, and the maximum value must be strictly less than NsPLUS_INF, as those constants represent infinity, as we will see below.

NsIntVar()

This constructor creates a variable that can be initialized afterwords by assigning an expression to it using the overloaded operator =, as on the third line of the following example.

NsIntVar X(pm, 0, 3), Y(pm, -1, 15), Z;
NsIntVar W = X + 3 * Y;
Z = W * W;

On the second line of the example, we used another constructor function, that takes an Expression as argument; here the expression is X + 3 * Y.

void remove(NsInt val)

Removes the value val from the domain of the variable.

void remove(NsInt min, NsInt max)

Removes every value of the domain that is in the range [min,max]. Instead of removing those values one by one using remove(val), it is more efficient to call this method.

void removeAll()

“Empties” the domain of the variable. Practically, the solver only informs the problem manager that an inconsistency occurred, due to an empty domain. This method is useful when we want to make search fail. E.g. when we want a goal to fail during its execution, we call this method for any variable. In order to show that a goal succeeds, we make GOAL() return 0; but in order to show that a goal failed, this is a—less elegant—way to do it. For more information about the goals mechanism see the Search via Goals section.

void set(NsInt val)

Assigns the value val to the variable; thus, the variable becomes bound (instantiated).

The following methods do not alter the variable for which they are called.

bool contains(NsInt val)

Returns true if the domain of the variable contains the value val.

NsInt min()

The minimum value of the domain.

NsInt max()

The maximum value of the domain.

NsUInt size()

The number of the values in the domain.

bool isBound()

Returns true if the variable is bound. “Bound” is synonym to “instantiated,” “singleton,” and “fixed” and means that a value has been assigned to the variable or, in other words, that the domain contains only one value.

NsInt value()

It is used only when the variable is bound and returns its (unique) value. If it is called for an unbound variable, an exception is thrown.

If the variable is bound, this method is equivalent to min() and max(), but we use this method for better code readability.

NsInt next(NsInt val)

Returns the smallest value in the domain, that is strictly greater than val. If such a value does not exist, the function returns NsPLUS_INF. The argument val may be equal to the special values NsMINUS_INF and NsPLUS_INF too.

NsInt previous(NsInt val)

Returns the greatest value in the domain, that is strictly less than val. If such a value does not exist, the function returns NsMINUS_INF. The argument val may be equal to the special values NsMINUS_INF and NsPLUS_INF too.

Two iterators for the class accompanied by examples follow.

NsIntVar::const_iterator

Iterates through all the values of the domain of the variable. E.g. the following code prints the values of the variable Var, in ascending order.

for (NsIntVar::const_iterator v = Var.begin(); v != Var.end(); ++v)
    cout << *v << "\n";

NsIntVar::const_reverse_iterator

Iterates through the values of the domain of the variable in reverse order. E.g. the following code prints the values of the variable in descending order.

for (NsIntVar::const_reverse_iterator v=Var.rbegin(); v != Var.rend(); ++v)
    cout << *v << "\n";

NsIntVar::const_gap_iterator

Iterates through the gaps that exist inside the domain of the variable. Mathematically speaking, it gives all the values (inside [min,max]) of the complement of the domain (where min is the minimum, and max is the maximum of the domain). E.g.

for (NsIntVar::const_gap_iterator g = Var.gap_begin(); g != Var.gap_end(); ++g)
    cout << *g << "\n";

Finally, the operator << has been overloaded in order to print the variable to an output stream, by writing e.g. cout << Var;.