Scribd
Upload
41 views1 page

hw4 1-Solutions

This document summarizes an algorithm for solving the node-capacitated maximum flow problem by reducing it to an edge-capacitated maximum flow problem. It transforms the node-capacitated network G into an edge-capacitated network G' by replacing each node with an input and output node connected by an edge of the node's capacity. Finding the max flow in G' then gives the max flow in G. It runs in O((m+n)n log n) time using existing max flow algorithms on G'. The max flow value also equals the capacity of the minimum node cut that separates the source and sink in G.

Uploaded by

poo
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
41 views1 page

hw4 1-Solutions

This document summarizes an algorithm for solving the node-capacitated maximum flow problem by reducing it to an edge-capacitated maximum flow problem. It transforms the node-capacitated network G into an edge-capacitated network G' by replacing each node with an input and output node connected by an edge of the node's capacity. Finding the max flow in G' then gives the max flow in G. It runs in O((m+n)n log n) time using existing max flow algorithms on G'. The max flow value also equals the capacity of the minimum node cut that separates the source and sink in G.

Uploaded by

poo
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/ 1

CMPS102: Introduction to Analysis of Algorithms Prof.

Abhradeep Guha Thakurta


University of California Santa Cruz June 7, 2017

1. (Node-Capacitated Networks)

(a) Algorithm Given a node-capacitated network of cities G with source s and sink t, we construct an
edge-capacitated graph G0 as follows. Each node v in G, besides s and t, is transformed to two nodes
vin and vout in G0 . All incoming edges to v are replaced with edges pointing to vin while all outgoing
edges from v are replaced with edges leaving vout . These edges have infinite capacities. G0 also has
an edge from vin to vout with capacity cv .

∞ cv ∞
v vin vout

(a) a node v in G (b) nodes vin and vout in G0

We compute a max flow f 0 from s to t in G0 by running a max flow algorithm. Once f 0 is obtained,
we transform it to a max flow f in G by contracting all edges of the form (vin , vout ) to v.
Correctness The flow f in G satisfies flow conservation because the flow f 0 in G0 does. It satisfies
the node-capacity constraints in G because f 0 satisfies the edge-capacity constraints in G0 . The flow
value (the amount of flow coming out of s) is the same for f and f 0 . So, the value of f is equal to the
value of max flow in G0 .
The value of max flow in G cannot be larger than the value of max flow in G0 because any flow in G
can be be converted to a flow of the same value in G0 . For every node v, other than the source and
the sink in G, the flow going through v in G can pass through the edge (vin , vout ) in G0 . It will not
exceed the capacity of this edge because the capacity of v in G was not exceeded.
Time and Space Complexity Graph G0 has O(m + n) edges and O(n) vertices, and can be
generated in O(m + n) time and space. Infinite capacities can be modeled by choosing a number
C larger than the sum of all node capacities. The capacity-scaling algorithm will do better than
Bellman-Ford because Bellman-Ford has running time proportional to C, while the capacity-scaling
has running time proportional to log C. It is even better to use the max flow algorithms that runs
in O(|V ||E| log |V |) time. Employing this algorithm to find a max flow in G0 takes O((m + n)n log n)
time. It takes O(m + n) time to convert f 0 to f . Hence, the running time is O((m + n)n log n).
It takes O(m + n) space to store a graph and run a max flow algorithm.
(b) By the max-flow min-cut theorem, the max flow value in G0 equals the capacity of its minimum cut.
0 ) have
Only the edges of the form (vin , vout ) can be in this cut, since edges of the form of (vout , vin
infinite capacities. In the original node-capacitated graph G, this corresponds to deleting a subset of
the nodes. The capacity of such subset is defined as the sum of capacities of deleted nodes.
(c) By the explanation in part (b), in a node-capacitated network, the max flow value equals the minimum
capacity of a set of nodes whose deleting disconnects the source and the sink.

You might also like