Information Technology Reference
In-Depth Information
Tabl e 2.
Numerical results Class 1 for unknown instances (300-500 products)
sLAHC
LAHC
5000
LAHC
10000
V
LA
V
LA
V
LA
VKf
(
σ
)
f
(
σ
)
f
(
σ
)
33 300 14576
17
14637
15
14346
16
33 300 14712
16
14733
14
14710
14
33 300 14463
16
14695
16
14354
15
33 300 14347
16
14189
19
14410
15
33 300 14530
15
14471
17
14329
16
33 350 15901
17
15990
19
16006
18
33 350 15047
13
14665
17
15035
16
33 350 16383
15
16464
16
16108
17
33 350 16252
17
16400
16
16023
17
33 350 16118
16
16398
17
16249
18
33 400 17790
18
17805
17
17905
19
33 400 16906
18
16836
17
16832
16
33 400 17335
18
17547
19
17480
18
33 400 17313
19
17824
18
17703
19
33 400 17825
17
18029
17
17832
19
33 450 18140
18
18360
18
18309
19
33 450 18164
18
18312
20
18097
18
33 450 18555
17
18677
19
18737
18
33 450 17819
17
17952
20
17875
18
33 450 19021
18
19081
18
18795
18
33 500 19897
19
19821
21
19827
19
33 500 19888
18
19604
19
20028
19
33 500 19774
20
19967
19
19885
19
33 500 19863
18
19925
19
19883
19
33 500 19497
20
19818
19
19643
20
Tabl e 3.
Comparison of results
sLAHC
LAHC
5000
LAHC
10000
Laporte et
al. [12]
VK
%
gap V
LA
%
gap
V
LA
V
LA
%
gap
%
gap
33 50 0,000 8,4 7,507
8
6,135
8,4
6,312
33 100 0,816 10,4 3,215
10,2
3,839
9,8
4,150
33 150 1,434 11,8 3,331
12,4
2,411
12
4,760
33 200 1,954 12,6 4,776
13,6
3,540
13,2
8,567
33 250 2,666
15 3,044
15,2
3,155
16
5,478
is flatter than the LAHC. For both algorithms it is apparent, that the most
improvement in the average objective function value occurs in the first thousand
iterations. After 10000 iterations almost no more significant improvement takes
place. The figure also illustrates that the sLAHC has a lower average objective
function value than the LAHC.
