Academia.eduAcademia.edu

Figure 1 – uploaded by Enrique Mota

See full PDFdownloadDownload figure
Therefore, we will call LP, the following linear relaxation of F1 On the other hand, once the integrality conditions have been removed from F2, symmetry conditions (14) alone do not longer guarantee that all the vertices will be even. However, constraints (15) should also be satisfied by any solution x to F2.

Table 1 Therefore, we will call LP, the following linear relaxation of F1 On the other hand, once the integrality conditions have been removed from F2, symmetry conditions (14) alone do not longer guarantee that all the vertices will be even. However, constraints (15) should also be satisfied by any solution x to F2.

Related Figures (12)

In what follows, we show that LPO fg is stronger that LPOr ;. Consider a feasible solution x to LPOpe2 and define again ye = a4; + 2;;, Ve = (i,9) € Er and yj = xj;, V(i,j) € A. If « satisfies the symmetry constraints (28) for every vertex, y satisfies all the system equations and also the balanced-set constraints (20) for each vertex. Then, from any feasible solution to LPOr2 we obtain a feasible solution to LPOr, with the same cost. However, the converse is not true, as the following example shows. The graph in Figure 1, in which all the links are required, represents a feasible solution to LPO; (although the balanced-set conditions are not satisfied for every subset S of vertices). However, there is no way of directing the edges so as, with the same cost, obtaining a symmetric graph. Therefore, the lower bound obtained using LPO; cannot be greater than the lower bound obtained using LPOr2. This fact is reflected in column LBO in the tables in Section 4. Figure 1: Feasible solution to LPOr1
In what follows, we show that LPO fg is stronger that LPOr ;. Consider a feasible solution x to LPOpe2 and define again ye = a4; + 2;;, Ve = (i,9) € Er and yj = xj;, V(i,j) € A. If « satisfies the symmetry constraints (28) for every vertex, y satisfies all the system equations and also the balanced-set constraints (20) for each vertex. Then, from any feasible solution to LPOr2 we obtain a feasible solution to LPOr, with the same cost. However, the converse is not true, as the following example shows. The graph in Figure 1, in which all the links are required, represents a feasible solution to LPO; (although the balanced-set conditions are not satisfied for every subset S of vertices). However, there is no way of directing the edges so as, with the same cost, obtaining a symmetric graph. Therefore, the lower bound obtained using LPO; cannot be greater than the lower bound obtained using LPOr2. This fact is reflected in column LBO in the tables in Section 4. Figure 1: Feasible solution to LPOr1
Table 2: MCPP instances from Madrigueras graph
Table 2: MCPP instances from Madrigueras graph
Table 1: MCPP instances from Albaida graph
Table 1: MCPP instances from Albaida graph
Table 4: MRPP instances from Albaida graph
Table 4: MRPP instances from Albaida graph
Table 3: MCPP instances from Aldaya graph
Table 3: MCPP instances from Aldaya graph
Table 5: MRPP instances from Madrigueras graph
Table 5: MRPP instances from Madrigueras graph
Table 6: MRPP instances from Aldaya graph
Table 6: MRPP instances from Aldaya graph
Table 7: MGRP instances from Albaida graph
Table 7: MGRP instances from Albaida graph
Table 8: MGRP instances from Madrigueras graph
Table 8: MGRP instances from Madrigueras graph
Table 9: MGRP instances from Aldaya graph
Table 9: MGRP instances from Aldaya graph
Related topics:
MathematicsApplied MathematicsComputer ScienceArc RoutingNumerical Analysis and Computational MathematicsLINEAR PROGRAM
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved