14.07.2013 Views

X - UWSpace - University of Waterloo

X - UWSpace - University of Waterloo

X - UWSpace - University of Waterloo

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.

Chapter 1 Introduction<br />

1.1 Brief History <strong>of</strong> Decomposition<br />

Over the past several decades. linear programming (LP) has been a useful planning and<br />

scheduling tool for economic and management applications. The real world LP problems are very<br />

large and usuaily have some special structures, which could break into several distinct LPs except<br />

for a few linking consuaints. These linking constmints represent relationships among different<br />

periods. regions. stochastic scenarios. etc. The most cornmonly discussed structures are prima1<br />

(or dual) block-anguim. sraircase and block-triangular as shown in Figure 1.1.<br />

l a. bloc k-angular b. staircase c. block-nianmilar<br />

- -<br />

Figure 1.1 Structures <strong>of</strong> mulù-stage problem. from Lm and Fuller [1995]<br />

Based on the ideas <strong>of</strong> utilizing the specid structures, many decomposition algorithms.<br />

which drcompose a very large problem into several subproblems and solve them iteratively<br />

through the exchanges <strong>of</strong> information, have been proposed since the early 1960's,<br />

There have ken several motivations to study decomposition: the prospect <strong>of</strong> reducing<br />

computationd time: a rnethod to solve a huge mode1 within a cornputer's memory limits that do<br />

not permit a straightfonvard solution <strong>of</strong> the whole mode1 (Fragniere et al. [1998]); and to ease

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

Saved successfully!

Ooh no, something went wrong!