Totally Unimodular Matrices John Mitchell

Totally Unimodular Matrices John Mitchell

Totally Unimodular Matrices John Mitchell See Nemhauser and Wolsey, section III.1.2, for more information. Definition 1 An m × n matrix A is totally unimodular (TU) if the determinant of each square submatrix is equal to 0, 1, or -1. Necessary conditions: n Theorem 1 If A is totally unimodular then all the vertices of fx 2 R+ : Ax = bg are integer for any integer vector b 2 Rm. n Theorem 2 If A is totally unimodular then all the vertices of fx 2 R+ : Ax ≤ bg are integer for any integer vector b 2 Rm. It follows that if A is totally unimodular then the optimal solution to the integer program max cT x subject to Ax ≤ b x ≥ 0; integer can be found by solving its LP relaxation. Theorem 3 The following statements are equivalent: 1. A is totally unimodular. 2. AT is totally unimodular. 3. [A; I] is totally unimodular. 4. A matrix obtained by deleting a unit row or column of A is totally unimodular. 5. A matrix obtained by multiplying a row or column of A by −1 is totally unimodular. 6. A matrix obtained by interchanging two rows or two columns of A is totally unimodular. 7. A matrix obtained by duplicating rows or columns of A is totally unimodular. 8. A matrix obtained by pivot operations on A is totally unimodular. Note: unit rows and columns are rows and columns of the identity matrix. 1 Sufficient conditions: Theorem 4 An integer matrix A with every entry aij = 0 or ±1 is totally unimodular if no more than two nonzero entries appear in any column, and if the rows of A can be partitioned into two sets I1 and I2 such that: 1. If a column has two nonzero entries of the same sign then their rows are in different sets. 2. If a column has two nonzero entries of opposite signs then their rows are in the same set. Corollary 1 Any linear program of the form max cT x max cT x subject to Ax = b or subject to Ax ≤ b x ≥ 0 x ≥ 0 where A is either 1. the node-arc incidence matrix of a directed graph, or 2. the node-edge incidence matrix of an undirected bipartite graph has only integer optimal vertices. Thus, the following problems can be solved by solving linear programs, and the optimal solutions to the LPs are integral: shortest path, max-flow, assignment, weighted bipartite matching, . n m Theorem 5 If all the extreme points of fx 2 R+ : Ax ≤ bg are integral for all b 2 Z then A is totally unimodular. Necessary and sufficient condition: Theorem 6 Let aij = 0; 1; −1 for all entries. The following statements are equivalent: 1. A is totally unimodular. 2. For every J ⊆ N := f1; : : : ; ng, there exists a partition J1;J2 of J such that X X aij − aij ≤ 1 for i = 1; : : : ; m: j2J1 j2J2 Note that Theorem 4 is a special case of this theorem (after transposing the matrix). 2.

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    2 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us