20.05.2014 Views

link to my thesis

link to my thesis

link to my thesis

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.

5.3. EFFICIENCY OF THE ND-SYSTEMS APPROACH 61<br />

is that many computations can be carried out in parallel since their exact order does<br />

not matter, so that many alternatives exist with equal performance. Already for<br />

small values of n, m and |t| this makes that the graph G can become very large. Two<br />

helpful observations for constructing a useful shortest path formulation are: (i) any<br />

sequence of time instants which facilitates the computation of y t1,...,t n<br />

from w 0,...,0<br />

can always be reorganized such that the <strong>to</strong>tal time increases mono<strong>to</strong>nically; (ii) the<br />

computation of values at time instants having the same <strong>to</strong>tal time can be carried out<br />

in any arbitrary order. Therefore, a node v ∈ V can naturally be associated with a<br />

set of time instants all having the same <strong>to</strong>tal time, rather than with individual time<br />

instants. This is formalized in the following definition.<br />

Definition 5.2. For k = 1, 2,..., let T k be the set of all multidimensional time<br />

instants t =(t 1 ,...,t n ) ∈ N n 0 for which |t| = k and max{t 1 ,...,t n }≥m. Let V k be<br />

the power set of T k (i.e., the set of all its subsets). Let V ⋆ be the set of time instants<br />

corresponding <strong>to</strong> w 0,...,0 (i.e., for which max{t 1 ,...,t n }

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

Saved successfully!

Ooh no, something went wrong!