Scribd
Upload
10 views2 pages

Sheet 12

The document outlines Problem Sheet 12 for a course on Discrete Optimization at the Technical University of Munich, featuring various exercises related to digraphs, b-flows, and cycle-cancelling methods. Exercises include constructing specific digraphs, proving conditions for b-flow existence, and analyzing minimum cost walks and mean cycles. Additionally, it proposes a dynamic programming algorithm to determine minimum mean cycles in a digraph.

Uploaded by

scribd01.28p98
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
10 views2 pages

Sheet 12

The document outlines Problem Sheet 12 for a course on Discrete Optimization at the Technical University of Munich, featuring various exercises related to digraphs, b-flows, and cycle-cancelling methods. Exercises include constructing specific digraphs, proving conditions for b-flow existence, and analyzing minimum cost walks and mean cycles. Additionally, it proposes a dynamic programming algorithm to determine minimum mean cycles in a digraph.

Uploaded by

scribd01.28p98
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

Discrete Optimization

School of Computation, Information and Technology


Technical University of Munich

Problem Sheet 12
Discrete Optimization (CIT413041)

Homework Problems

Exercise H 12.1
Construct a (not completely trivial) digraph 𝐺 = (𝑉 , 𝐸) with edge capacities 𝑢 ∶ 𝐸 → R≥0 and
𝑏 ∶ 𝑉 → R with ∑𝑣∈𝑉 𝑏(𝑣) = 0 such that no feasible 𝑏-flow exists in 𝐺.

Exercise H 12.2
Let 𝐺 = (𝑉 , 𝐸) be a digraph with edge capacities 𝑢 ∶ 𝐸 → R≥0 and 𝑏 ∶ 𝑉 → R with ∑𝑣∈𝑉 𝑏(𝑣) = 0.
a) Show: If a 𝑏-flow exists in 𝐺 then

∑ 𝑢(𝑒) ≥ ∑ (−𝑏(𝑣)) for all 𝑋 ⊂ 𝑉.


𝑒∈𝛿out (𝑋) 𝑣∈𝑋

b) Show that the condition given above is also sufficient for the existence of a 𝑏-flow in 𝐺. Hint:
Use the construction discussed in the lecture and apply the Maxflow-Mincut Theorem.

Problems for the Exercise Class

Exercise T 12.3
Like with maximum flow one might think of a “naive” implementation of cycle-cancelling: In each
step, determine some negative cycle in the residual graph and augment flow along that cycle. Continue
until no more negative cycle exists.
Construct an example to show that this can result in exponentially many augmentation steps.

Exercise T 12.4
Let 𝐺 = (𝑉 , 𝐸) be a digraph on 𝑛 ∶= |𝑉| nodes with edge costs 𝑐 ∶ 𝐸 → R and let 𝑠 ∈ 𝑉 such that each
vertex is reachable from 𝑠. For each 𝑥 ∈ 𝑉 and 𝑘 ∈ Z>0 let 𝐹𝑘 (𝑥) denote the cost of a minimum cost
walk (that may visit vertices and edges repeatedly) from 𝑠 to 𝑥 with exactly 𝑘 edges (where 𝐹𝑘 (𝑥) = ∞
if there is no such walk). Let 𝜇(𝐺) be the mean cost of a minimum mean cycle in 𝐺. In this problem,
we will show that
𝐹𝑛 (𝑥) − 𝐹𝑘 (𝑥)
𝜇(𝐺) = min max{ ∶ 0 ≤ 𝑘 ≤ 𝑛 − 1, 𝐹𝑘 (𝑥) < ∞} .
𝑥∈𝑉 𝑛−𝑘

a) Consider the case 𝜇(𝐺) = 0 and show that for all 𝑥 ∈ 𝑉 the following inequality holds:

𝐹𝑛 (𝑥) − 𝐹𝑘 (𝑥)
max{ ∶ 0 ≤ 𝑘 ≤ 𝑛 − 1 , 𝐹𝑘 (𝑥) < ∞} ≥ 0 .
𝑛−𝑘
b) Assume again 𝜇(𝐺) = 0 and show that there is a vertex 𝑥 ∈ 𝑉 where the above inequality holds
at equality.
c) Prove the general case using a reduction to the case of 𝜇(𝐺) = 0.
d) Using these results, devise a dynamic programming algorithm of running time 𝒪 (𝑚𝑛) to deter-
mine a minimum mean cycle in a digraph 𝐺 = (𝑉 , 𝐸) with edge cost 𝑐 ∶ 𝐸 → R.

You might also like