13.04.2014 Views

Mathematical Optimization in Graphics and Vision - Luiz Velho - Impa

Mathematical Optimization in Graphics and Vision - Luiz Velho - Impa

Mathematical Optimization in Graphics and Vision - Luiz Velho - Impa

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

2.5. VARIATIONAL OPTIMIZATION PROBLEMS 27<br />

Example 10 (The Travel<strong>in</strong>g Salesman Problem). This need is well illustrated by<br />

the Travel<strong>in</strong>g Salesman Problem. Given n towns, one wishes to f<strong>in</strong>d the m<strong>in</strong>imum<br />

length route that starts <strong>in</strong> a given town, goes through each one of the others <strong>and</strong><br />

ends at the start<strong>in</strong>g town.<br />

The comb<strong>in</strong>atorial structure of the problem is quite simple. Each possible route<br />

corresponds to one of the (n − 1)! circular permutations of the n towns. Thus, it<br />

suffices to enumerate these permutations, evaluate their lengths, <strong>and</strong> choose the<br />

optimal route.<br />

This, however, becomes unpractical even for moderate values of n. For <strong>in</strong>stance,<br />

for n = 50 there are approximatelly 10 60 permutations to exam<strong>in</strong>e. Even<br />

if 1 billion of them were evaluated per second, exam<strong>in</strong><strong>in</strong>g all would require about<br />

10 51 seconds, or 10 43 years! However, there are techniques that allow solv<strong>in</strong>g this<br />

problem, <strong>in</strong> practice, even for larger values of n.<br />

2.5 Variational optimization problems<br />

An optimization problem is called a variational problem when its solution set S<br />

is an <strong>in</strong>f<strong>in</strong>ite dimensional subset of a space of functions.<br />

Among the most important examples we could mention the path <strong>and</strong> surface<br />

problems. The problems consist <strong>in</strong> f<strong>in</strong>d<strong>in</strong>g the “best path” (“best surface”) satisfy<strong>in</strong>g<br />

some conditions which def<strong>in</strong>e the solution set.<br />

Typical examples of variational problems are:<br />

Example 11 (Geodesic problem). F<strong>in</strong>d the path of m<strong>in</strong>imum length jo<strong>in</strong><strong>in</strong>g two<br />

po<strong>in</strong>ts p 1 <strong>and</strong> p 2 of a given surface.<br />

Example 12 (M<strong>in</strong>imal surface problem). F<strong>in</strong>d the surface of m<strong>in</strong>imum area for a<br />

given boundary curve.<br />

Variational problems are studied <strong>in</strong> more detaill <strong>in</strong> Chapter 4.<br />

2.6 Other classifications<br />

Other forms of classify<strong>in</strong>g optimization problems are based on characteristics<br />

which can be exploited <strong>in</strong> order to devise strategies for the solution.

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

Saved successfully!

Ooh no, something went wrong!