Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd
Upload
102 views

Binary Branch-And-Bound: Implicit Enumeration

This document describes the implicit enumeration algorithm for solving integer programs with only binary variables. It works by using a branch-and-bound approach to fix variables to 0 or 1, ignoring the linear constraints. Variables not fixed are set to 0, which maximizes the objective function due to negative coefficients. The algorithm proceeds by subdividing the problem space based on fixed variable assignments and pruning subdivisions that produce infeasible or suboptimal solutions.

Uploaded by

Ashoka Vanjare
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
102 views

Binary Branch-And-Bound: Implicit Enumeration

This document describes the implicit enumeration algorithm for solving integer programs with only binary variables. It works by using a branch-and-bound approach to fix variables to 0 or 1, ignoring the linear constraints. Variables not fixed are set to 0, which maximizes the objective function due to negative coefficients. The algorithm proceeds by subdividing the problem space based on fixed variable assignments and pruning subdivisions that produce infeasible or suboptimal solutions.

Uploaded by

Ashoka Vanjare
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 14

Implicit enumeration

BINARY BRANCH-AND-BOUND

A special branch-and-bound procedure for


integer programs with only binary variables
The algorithm has the advantage that it
requires no linear-programming solutions.
In the ordinary branch-and-bound procedure,
subdivisions were analyzed by maintaining the
linear constraints and dropping the integrality
restrictions relax the variables
Here, we adopt the opposite tactic of always
maintaining the 01 restrictions, but ignoring
the linear inequalities relax the constraints

General idea
The idea is to utilize a branch-and-bound (or
subdivision) process to fix some of the variables
at 0 or 1.
The variables remaining to be specified are called
free variables.
Note that, when the inequality constraints are
ignored, the objective function is maximized by
setting the free variables to zero, since their
objective function coefficients are negative.
This is mandatory to apply this algorithm.

Example
maximization problem

all coefficients are negative

subject to:

constraints are specified as


less than or equal to
inequalities

We start with no fixed variables, and


consequently every variable is free and set to
zero.
The solution does not satisfy the inequality
constraints, and we must subdivide to search
for feasible solutions.
One subdivision choice might be:
For subdivision 1 : x
For subdivision 2 : x = 0.

Now variable x is fixed in each subdivision. If the


inequalities are ignored, the optimal solution over each
subdivision has
x = x3 = x4 = x5 = 0.
The resulting solution in subdivision 1 gives
z = 8(1) 2(0) 4(0) 5(0) + 10 = 2
and happens to satisfy the inequalities, so the optimal
solution to the original problem is at least 2, z*
Also, subdivision 1 has been fathomed:
The above solution is best among all 01 combinations
with x
No other feasible 01 combination in subdivision 1 needs
to be evaluated explicitly.
These combinations have been considered implicitly.

The solution with x = x3 = x4 = x5 = 0 in


subdivision 2 is the same as the original
solution with every variable at zero, and is
infeasible.
Consequently, the region must be subdivided
further, say with x
or x = 0, giving:
For subdivision 3 : x = 0, x
For subdivision 4 : x = 0, x = 0.

Enumeration tree to this point

Completing it

Complete
tree

Note that
1. At , the solution x = 0, x = x3
with free
variables x4 = x5 = 0, is feasible, with z = 4, thus
providing an improved lower bound on z.
2. At , the solution x = x3 = 0, x = x4
and
free variable x5 = 0, has
so that no
solution in that subdivision can be as good as
that generated at .
3. At and , every free variable is fixed. In each
case, the subdivisions contain only a single
point, which is infeasible, and further
subdivision is not possible.

Note that
4. At , the second inequality (with fixed
variables x = x = 0) reads:
x3 4 + x 5
No 01 values of x3, x4, or x5 completing
the fixed variables x = x = 0 satisfy this
constraint, since the lowest value for the
lefthand
x3 =
and x5 = 0.
x4
The subdivision then has no feasible solution
and need not be analyzed further.

Generalizing
Subdivisions are fathomed if any of three
conditions hold:
i.

the integer program is known to be infeasible over


the subdivision, for example, by the previous
infeasibility test;
ii. the 01 solution obtained by setting free variables
to zero satisfies the linear inequalities; or
iii. the objective value obtained by setting free
variables to zero is no larger than the best feasible
01 solution previously generated.

Once again
This algorithm is designed for problems in
which:
1. the objective is a maximization with all
coefficients negative; and
2. constraints are specified as less than or equal
to inequalities.

If not, then

Problem transformation
Minimization problems are transformed to
maximization by multiplying cost coefficients by
If xj appears in the maximization form with a
positive coefficient, then the variable substitution
xj
xj everywhere in the model leaves the
binary variable xj with a negative objectivefunction coefficient.
Greater than or equal to constraints can be
to constraints.

You might also like