---

Benchmarks

Series RAN

Series RAN consists of 8 SMPLD instances obtained from FCTP benchmarks with given m; n and m_ij = 0, constructed by Gottlieb and Paulmann (1998), using a modi¯cation of NETGEN generator (Barr et al., 1981). The original FCTP problems are available at http://plato.la.asu.edu/ftp/fctp/ 

Results for series RAN

All instances series RAN  RAN.zip

Code	fbest	f h cplex	fGAmipr	fGAgrd
ran13×13	3252*	3252	3271	3284
ran12×21	3664*	3675	3664	3671
ran14×18	3712*	3746	3712	3714
ran 4×64	9711*	9711	9711	9726
ran8×32	5247*	5247	5247	5247
ran16×16	3823*	3827	3823	3823
ran10×26	4270*	4270	4270	4270
ran17×17	1373*	1373	1373	1373

 

---

Series H

Series H consists of 14 larger FCTP instances h_0, h_1, ... , h_c and h_e (the instance h d is omitted because it turned out to be identical to h_c) with m = 100; n = 50, m_ij = 0 and the range ofixed costs 6400-25600, produced by the same generator (Sun et al., 1998). The original FCTP problems are available at  http://plato.la.asu.edu/ftp/fctp/ 

Results for series H

All instances series H  H.zip

Code	LB	fbest	f h cplex	fGAmipr	fGAgrd
h_0	1002063	1208672	1260048	1217305	1284070
h_1	990406	1204194	1241812	1231028	1303920
h_2	991110	1192687	1269582	1198609	1305210
h_3	954486	1188904	1236705	1215025	1305520
h_4	992374	1209357	1283998	1238455	1332310
h_5	997677	1202258	1264337	1214291	1311000
h_6	990739	1201097	1258163	1235884	1305000
h_7	967693	1172931	1259481	1204284	1281000
h_8	984973	1198108	1272177	1236018	1306000
h_9	972743	1192529	1267414	1227936	1308000
h_a	994451	1233688	1294036	1259553	1304000
h_b	1003666	1208211	1289744	1237080	1319000
h_c	990923	1209685	1260134	1233882	1338000
h_e	987160	1185378	1257240	1195558	1326000

 

---

Series S

Series S consists of 8 randomly generated SMPLD instances with m = n = 50. All m_ij = 10, the values M_i Î [400; 600], A_j Î [350; 550], a_ij Î [500; 4500] and c_ij Î [20; 1020] are obtained by a uniformly distributed pseudo-random numbers generator. So, on average, the total supply is by 10% greater than the overall demand.

Results for series S

All instances series S  S.zip

Code	LB	fbest	f h cplex	fGAmipr	fGAgrd
s_1	604685	613951	613951	613951	614289
s_2	622470	631219	631219	631219	631540
s_3	617403	627294	627294	627294	627392
s_4	622761	634609	636855	634609	634872
s_5	595627	605145	605145	605380	605432
s_6	655595	669346	669371	669787	670730
s_7	589851	602737	602737	602737	603788
s_8	633324	645843	645843	645843	647242

 

 

---

Series P

Series P consists of 14 randomly generated SMPLD instances with n = m = 50. The values  M_i Î [12; 20], A_j Î [10; 20], a_ij Î [5; 10], c_ij Î [10; 20], and m_ij Î [5; 10] are obtained by a uniformly distributed generator. These ranges of parameters imply that on average the total supply is slightly higher than the overall demand. The values m_ij are big enough to make it particularly difficult to find a feasible solution.

Results for series P

All instances series P  P.zip

Code	fbest	f h cplex	fGAmipr	fGAgrd
p_1	8527	–	8527	8597
p_2	7836	7836	7840	7865
p_3	8423	–	8423	8513
p_4	8409	–	–	8574
p_5	8007	8011	8007	8027
p_6	8320	8369	8320	8392
p_7	7874	7874	7876	7889
p_8	9543	–	–	–
p_9	8831	–	–	8928
p_10	8524	–	8524	8603
p_11	8196	–	8196	8263
p_12	8323	–	8323	8547
p_13	8276	–	8696	8276
p_14	8916	–	–	8916

 

---

Series MIS

Series MIS consists of 6 instances obtained from the maximum independent set problems, using the reduction proposed by Eremeev et al. (2006). In these SMPLD instances m = | E | + 1 and  n = | V |, where V and E are the sets of vertices and edges in the graph of maximum independent set problem. The instances are obtained from the graphs of DIMACS collection of benchmarks (Johnson and Trick, 1996). The values of m, n and the magnitude of coeffcients a_ij are given in Table 1. The graphs from DIMACS set can be found by ftp://dimacs.rutgers.edu/pub/challenge/graph/

Results for series MIS

All instances series MIS  MIS.zip

Code	fbest	f h cplex	fGAmipr	fGAgrd
hamming6–4	60*	60	60	60
c125.9	91*	91	91	94
johnson16.2.4	112*	112	112	112
mann.a27	252*	–	262	259
gen200_p0.9_44	156*	166	164	171
gen200_p0.9_55	145*	164	165	171

 

---

References

1. Barr, R.S., R.S. Glover and D. Klingman (1981). A new optimization method for large scale ¯xed charge transportation problems. Oper. Res. 29(3), 448-463.

2. Eremeev, A.V., A.A. Romanova, V.V. Servakh and S.S. Chauhan (2006).

 

Home