(4) Evaluate R
Q
t
, and perform trimming operation for R
Q
t
(see Section 3.5). Compute all nondominated cells in R
Q
t
, and
store them to M
Q
, and then delete all dominated cells in M
Q
.
(5) Select new population P
Q
tþ1
from R
Q
t
by using the selection scheme based on truncation algorithm with similar indi-
viduals (TASI) (see Section 3.6).
(6) Randomly select a cell from M
Q
, and use it to update the new population P
Q
tþ1
by Q-gates (see Sectio 3.4).
(7) if t > T
Q
(the maximum number of generations) or another termination condition is satisfied, then terminate the
algorithm and output the nondominated cells in M
Q
. Otherwise, let t ¼ t þ 1 and return to (2).
2. The main procedures of BAIS:
(1) Let t = 0 and combine with the population obtained by QAIS to generate the initial population P
B
0
of size N
B
pop
. Set
the memory M
B
0
¼;with its maximum size N
mem
.
(2) For each cell in the population, reproduce the cell N
B
clones
times and mutate each clone by a reverse mutation oper-
ator (see Section 3.3).
(3) Repair the offspring (see Section 3.7), and then combine the population P
B
t
, the memory M
B
t
and the offspring as R
B
t
.
(4) Evaluate R
B
t
, and perform trimming operation for R
B
t
(see Section 3.5).
(5) Compute nondominated cells among R
B
t
, and copy them to the memory M
B
tþ1
.
(6) if t > T
B
(the maximum number of generations) or another termination condition is satisfied, then terminate the
algorithm and output the nondominated cells in the memory. Otherwise, go on the following step.
(7) If jM
B
tþ1
j > N
mem
, then perform TASI to reduce the size of M
B
tþ1
to N
mem
(see Section 3.5).
(8) If jM
B
tþ1
j > N
B
pop
, copy cells of M
B
tþ1
to a set Z
B
tþ1
, and use TASI to reduce the size of Z
B
tþ1
to N
B
pop
, and then let
P
B
tþ1
¼ Z
B
tþ1
. Otherwise, copy cells of M
B
tþ1
to P
B
tþ1
, and then select ðN
B
pop
jM
B
tþ1
jÞ cells from dominated cells among
R
B
t
by using the selection scheme based on TASI (see Section 3.6), and add them to P
B
tþ1
. Let t ¼ t þ 1, and go to (2).
Next, we fuse the above QAIS and BAIS in an integrated framework to propose a MOQAIS for MKP, which is illustrated in
Fig. 1. From Fig. 1, it can be seen that our approach combines the quantum-inspired search and immune search. QAIS and
Reproduce each cell times in
the population and mutate each clone
If | | > , then truncate it until | | =
i > or another termination
condition is satisified?
Y
N
Generate a random initial population of size . Set two empty memories and , and for
, its maximum size is set to . Repair population , and set t=0
Q
0
P
Q
pop
N
t> or another termination
condition is satisified?
Generate the population of size
together with , Set memory as
and let i=0
Reproduce each cell times in the
population and mutate each clone by a
chaos-based quantum rotation gate
0
Z
B
0
P
Q
t
P
t
M
B
clones
N
Evaluate the offspring, repair them and
combine , and offspring as
i
Z
Perform trimming , compute nondominated
cells and store them to the memory Z
i+1
Q
T
B
T
Output the elite
solutions
mem
N
mem
N
Z
i+1
Z
i+1
Select the new population from using the
selection scheme based on TASI and let
B
pop
N
N
Q
clones
N
Evaluate the offspring, repair them and
combine and offspring as
t
R
Perform trimming ,compute nondominated
cells and add them to the memories and
M
t
Delete dominated cells in memories
andM
t
Select the new population from using
the selection scheme based on TASI
Store to the memory and delete
nondominated cells in
If BAIS and one generation of QAIS are both
completed, then
Y
Z
i+1
M
t+1
B
i
P
M
t+1
t
R
Q
t
P
R
t
M
0
M
0
N
mem
Q
0
P
Q
0
M
R
i
Q
t
M
Q
t
M
Select the best cell from and update the
new population with quantum rotation gate
t=t+ 1
i=i+ 1
R
i
R
i
B
Q
t
M
Fig. 1. The main framework of MOQAIS.
J. Gao et al. /Applied Mathematics and Computation 230 (2014) 120–137
123