22.08.2013 Views

Edge-connectivity of undirected and directed hypergraphs

Edge-connectivity of undirected and directed hypergraphs

Edge-connectivity of undirected and directed hypergraphs

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Section 6.5. Local <strong>connectivity</strong> requirements 109<br />

corresponding <strong>un<strong>directed</strong></strong> edge in E. The capacities <strong>of</strong> the edges are bounded by f ≡ 0,<br />

g ≡ 1.<br />

For a {0, 1}-valued cost function c, consider the system (S), <strong>and</strong> let the dual solutions<br />

be denoted by (y, z), where y consists <strong>of</strong> the dual variables associated to the constraints<br />

in (6.8). Take an integral dual optimal solution (y ∗ , z ∗ ) ≥ 0, where y ∗ is positive on a<br />

laminar family, <strong>and</strong> |z ∗ | is minimal. Let F be the laminar family where every set X has<br />

multiplicity y ∗ (X). It follows from the minimality <strong>of</strong> |z ∗ | that the objective value <strong>of</strong> this<br />

dual solution is<br />

<br />

p(X) − <br />

(ee(F) − c(e)) + − <br />

<br />

<br />

ρa(X) − c(a)<br />

X∈F<br />

e∈E<br />

a∈A<br />

X∈F<br />

+<br />

=<br />

= <br />

p(X) − eE(F) − <br />

ϱA(X) + µ(F).<br />

Conversely, the value <strong>of</strong> (6.14) corresponds to the value <strong>of</strong> the following dual solution<br />

(y ∗ , z ∗ ). Let F be a laminar family where the maximum is attained in (6.14). For X ⊆ V ,<br />

let y ∗ (X) be the multiplicity <strong>of</strong> X in F. Define the values <strong>of</strong> the dual variables in z ∗ as<br />

required by the dual constraints, always setting a variable corresponding to an edge to 0<br />

if the edge also belongs to an orientation constraint. In this case the dual objective value<br />

is equal to the expression (6.14).<br />

X∈F<br />

6.5 Local <strong>connectivity</strong> requirements<br />

Theorems 6.13 <strong>and</strong> 6.14 show that in relation to the covering <strong>of</strong> intersecting supermodular<br />

set functions, a relatively large class <strong>of</strong> orientation problems can be efficiently solved. The<br />

preceding results <strong>of</strong> the chapter indicate that a somewhat more restricted class is solvable<br />

concerning the covering <strong>of</strong> crossing supermodular set functions. In this section we attempt<br />

to further relax the condition <strong>of</strong> crossing supermodularity. As we have mentioned at the<br />

beginning <strong>of</strong> the chapter, it is open whether some analogue <strong>of</strong> Theorem 6.4 can be proved<br />

for <strong>hypergraphs</strong>. The problems discussed in this section are much less ambitious, but<br />

nevertheless they have some interesting corollaries.<br />

6.5.1 Local requirement for one pair <strong>of</strong> nodes<br />

We consider k-edge-connected orientations <strong>of</strong> graphs <strong>and</strong> <strong>hypergraphs</strong>, where the number<br />

<strong>of</strong> edge-disjoint paths required between two designated special nodes may be more than<br />

k. First we formulate a partition-type condition for the hypergraph case, <strong>and</strong> prove its<br />

sufficiency using a modified uncrossing method.<br />

X∈F

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!