Sie sind auf Seite 1von 9

Euclidean Travelling Sales Person

Problem

Vivek Vikram Singh


2014H103101P

What is Euclidean TSP problem?


In contrast with general Travelling Sales
Person (TSP) problem, in which there is no
restriction on cost function, Euclidean TSP is
a special case of Metric TSP.
The Metric TSP is a special case of Travelling
Sales Person Problem, all edge costs are
symmetric and the Triangle inequality is
satisfied.
The Euclidean TSP is a variant of Metric TSP,
in which the distance between point is
ordinary Euclidean Distance.

What is Euclidean contd..


Let C1,C2,C3. .Cn be a set of n- points in
the Euclidean space of d- dimensions. For sake
of simplicity, we assume d=2.
The graph is complete and cost function
between these points is given by the
Euclidean Distance.
Euclidean Distance between 2 points
p(p1,p2 ) & q(q1,q2) in 2- dimensional space is
given by
d (p , q) = { ( p1- q1 )2 + ( p2 - q2 )2 }1/2

What is Euclidean contd..


Euclidean TSP :
Given : n-points in the 2-dimensinal space,
corresponding to the location of the n-cities
with Euclid distance.
Find : Shortest tour by the sales person, that
visits all the cities.
Like the general TSP problem, the Euclidean
TSP problem is also NP-hard problem.

Why Euclidean TSP?

As we know that general TSP problem does


not have any relation between distances
among cities i.e. distances can be arbitrary.
There is no approximation possible.
In case of Metric TSP , there are 2approximation and 3/2 approximation
algorithms.
But still there is no Polynomial Time
Approximation Scheme (PTAS) algorithm.
There exist PTAS algorithms for Euclidean TSP
though.

Why Euclidean TSP contd..

Source: http://www.diku.dk/OLD/undervisning/2006-2007/20062007_b2_415/tspappr.pdf

How ??
There exist several algorithms , which are
known to be superpolynomial in time.
However, that time is not reasonable for
vertices being even less than 1000.
Considering an input instance being 15,112
cities of Germany , way back in 2001on
network of 110 workstations. The time taken
was almost equal to 30 years. Scary??
So we need an approximation approach to
this kind of problem.

Source: Recent PTAS Algorithms on the Euclidean TSP by Leonardo Zambito

Approximation Approach
The argument is that if we cant find an exact
feasible solution, then try to find efficient
approximation algorithms , which can be used
in practice.
There exists many approximation algorithms
for Euclidean TSP. These are introduced by
1) Sanjeev Arora
2) Joseph S. B. Mitchell
The goal is to study and analyze these
approaches which exist and come up with a
better understanding and solution for this
kind of problems.

References:

http://algo2.iti.kit.edu/schultes/lebenslauf/p
ublications/euclTSPsummary.pdf

Recent PTAS Algorithms on the


Euclidean TSP by Leonardo Zambito

Das könnte Ihnen auch gefallen