X - UWSpace - University of Waterloo
X - UWSpace - University of Waterloo
X - UWSpace - University of Waterloo
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
are first sent and necessary columns for other parts' linking consaaints are sent afterwards.<br />
5.1.2 Solution phase<br />
in the solution phase on PCs. each machine starts with its own process and receives the<br />
subproblem's data from the RS/6000 by the routine pvrn-receive(). Using the CPLEX Callable<br />
Library. each LP subproblem is loaded in each cornputer. For instance in SP3. the columns having<br />
that part's nonlinking consuainu (the columns in L33, &, L3 and B3, A3) are loaded fini. then<br />
necessxy columns for other parts' linking consuainü (the columns in L3 Li and L3?, Li2) are<br />
loaded and finally the unnecessary linking constraints <strong>of</strong> other parts (parts 1 and 3) are delered<br />
since column wise vecton include dl linking constraints' data. Each subproblem is solved<br />
sirnultaneously without any information exchange aat the fint iteration and exchanges necessary<br />
primai or dud information with other machines and solves each new subproblem again until the<br />
gap between the objective function values <strong>of</strong> the upper-upper bound subproblem and the lower-<br />
lower bound subproblem reaches 3 prescribed tolerance. Note that if any <strong>of</strong> the subproblems is<br />
unbounded. the whole pmcess stops at the fint itemtion by checking the optimal statu generated<br />
by CPXSolution().<br />
Since C PW provides the dud values corresponding to the primai constraints by calling<br />
the CPXsolutionO routine in the CPLEX Callable Library, we don't have to solve for the dual<br />
variables sepmtely, so the implementation efforts are bgreatly simplified The subproblems can<br />
be solved by the simplex method or the barrier methd For the bher method the dud and bais<br />
information cm be obtained by crossover at the last step using CPXhybbaropt().<br />
The new primal and duai proposais are multiplied by corresponding matrices and vecton<br />
93