(i) Consider a network with node set N and set of directed arcs A equipped with functions d+:A→Z and d−:A→Z with d−⩽d+. Given u:N→R we define the differential Δu:A→R by Δu(j)=u(i′)−u(i) for j=(i,i′)∈A. We say that Δu is a feasible differential if
d−(j)⩽Δu(j)⩽d+(j) for all j∈A
Write down a necessary and sufficient condition on d+,d−for the existence of a feasible differential and prove its necessity.
Assuming Minty's Lemma, describe an algorithm to construct a feasible differential and outline how this algorithm establishes the sufficiency of the condition you have given.
(ii) Let E⊆S×T, where S,T are finite sets. A matching in E is a subset M⊆E such that, for all s,s′∈S and t,t′∈T,
(s,t),(s′,t)∈M(s,t),(s,t′)∈M implies implies s=s′t=t′
A matching M is maximal if for any other matching M′ with M⊆M′ we must have M=M′. Formulate the problem of finding a maximal matching in E in terms of an optimal distribution problem on a suitably defined network, and hence in terms of a standard linear optimization problem.
[You may assume that the optimal distribution subject to integer constraints is integervalued.]