T
E
L
KO
M
NI
K
A
,
V
o
l.
1
4
,
N
o.
3
,
S
ept
em
ber
20
1
6
,
pp.
10
99
~
110
4
IS
S
N
:
1
693
-
6
930
,
ac
c
r
edi
t
ed
A
b
y
D
IK
T
I,
D
e
c
r
e
e
N
o
:
58/
D
I
K
T
I
/
K
ep/
2013
D
O
I
:
10.
12928/
T
E
LK
O
M
N
I
K
A
.
v
1
4
i
3
.
3609
10
99
R
ec
ei
v
ed
M
ar
c
h
2
2
,
201
6
;
R
ev
i
s
ed
J
u
ne 2
1
,
2
01
6
;
A
c
c
ept
ed
Ju
l
y 9
,
201
6
A
n
Im
p
r
o
v
e
d
A
r
t
if
icial Bee
C
o
lo
n
y
A
l
g
o
r
it
h
m
f
o
r
S
ta
ge
d
S
ear
ch
S
h
o
u
l
i
n
Y
i
n
1
,
Ji
e L
i
u
*
2
,
L
i
n
T
e
n
g
3
S
of
t
w
ar
e C
ol
l
e
ge
,
S
h
eny
ang N
or
m
al
U
n
i
ve
r
si
t
y
,
No
.
253,
H
uan
gH
e B
ei
S
t
r
ee
t
,
H
uangG
u
D
i
s
t
r
i
c
t
,
S
heny
ang
,
P
.
C
1100
34
-
C
hi
n
a
*
C
or
r
es
pond
i
ng
aut
hor
, e
-
m
ai
l
:
3527
202
14@
qq.
c
o
m
1
,
nan1
2
7@
s
ohu
.
c
o
m
2
,
153
2554
069
@
qq.
c
om
3
A
b
st
r
act
A
r
t
i
f
i
c
i
al
B
ee
C
ol
o
ny
(
A
B
C
)
or
i
t
s
i
m
pr
ov
ed
al
gor
i
t
hm
s
us
ed
i
n
s
o
l
v
i
ng
hi
g
h
d
i
m
ens
i
ona
l
c
o
m
pl
ex
f
unc
t
i
on op
t
i
m
i
z
at
i
on i
s
s
ues
has
s
om
e di
s
adv
ant
a
ges
,
s
uc
h as
l
ow
er
c
o
nv
e
r
gen
c
e,
l
ow
er
s
ol
ut
i
o
n
pr
ec
i
s
i
on,
l
ot
s
of
c
on
t
r
ol
par
a
m
et
er
s
of
i
m
pr
ov
ed al
g
or
i
t
hm
s
,
eas
y
t
o f
al
l
i
nt
o a l
o
c
al
op
t
i
m
um
s
ol
ut
i
on
.
I
n
t
hi
s
l
e
t
t
er
,
w
e pr
op
os
e an i
m
p
r
o
v
ed A
B
C
of
s
t
ag
ed s
e
ar
c
h
.
T
hi
s
new
al
gor
i
t
hm
des
i
gns
s
t
aged em
pl
o
y
e
d
bee
s
ear
c
h
s
t
r
a
t
eg
y
w
hi
c
h
m
ak
e
s
t
h
at
em
p
l
oy
ed
bee
ha
s
di
f
f
er
ent
s
ear
c
h
c
h
ar
ac
t
er
s
i
n
di
f
f
er
ent
s
t
a
ges
.
T
hat
r
ed
uc
e
s
pr
o
babi
l
i
t
y
of
f
a
l
l
i
ng
i
nt
o
l
oc
a
l
ex
t
r
em
e
v
al
ue.
I
t
def
i
n
es
t
he
es
c
ap
e
r
adi
us
w
hi
c
h c
an
gu
i
de
pr
ec
o
c
i
o
us
i
nd
i
v
i
dua
l
t
o j
um
p
l
oc
a
l
e
x
t
r
em
e v
al
u
e an
d av
oi
d t
he b
l
i
n
dne
s
s
of
f
l
i
ght
beh
a
v
i
or
.
M
eanw
hi
l
e,
w
e adopt
i
n
i
t
i
a
l
i
z
at
i
on s
t
r
at
eg
y
c
om
bi
n
i
ng u
ni
f
or
m
di
s
t
r
i
b
ut
i
on and b
ac
k
w
ar
d l
e
ar
ni
ng t
o pr
om
pt
i
ni
t
i
al
s
ol
u
t
i
on w
i
t
h uni
f
or
m
di
s
t
r
i
but
i
on and b
et
t
er
q
ual
i
t
y
.
F
i
nal
l
y
,
w
e m
ak
e s
i
m
ul
at
i
on e
x
per
i
m
ent
s
f
or
ei
g
ht
t
y
pi
c
al
hi
g
h di
m
en
s
i
o
nal
c
om
pl
ex
f
un
c
t
i
o
ns
.
R
e
s
ul
t
s
s
how
t
hat
t
he
i
m
pr
ov
ed al
gor
i
t
hm
has
a h
i
gh
er
s
ol
u
t
i
on
pr
ec
i
s
i
on an
d f
as
t
er
c
onv
er
gen
c
e r
at
e w
hi
c
h i
s
m
or
e
s
ui
t
a
bl
e f
o
r
s
ol
v
i
ng hi
gh di
m
e
ns
i
o
nal
c
om
pl
e
x
f
unc
t
i
on
s
.
K
eyw
o
r
d
s:
a
rt
i
fi
c
i
a
l
be
e c
ol
o
ny
,
s
t
a
ged
s
ear
c
h
,
f
un
c
t
i
on o
pt
i
m
i
z
at
i
on,
e
s
c
ap
e
r
ad
i
us
,
u
ni
f
or
m
d
i
s
t
r
i
but
i
on,
b
ac
k
w
ar
d l
ear
n
i
ng
C
o
p
y
r
i
g
h
t
©
20
16 U
n
i
ver
si
t
a
s A
h
mad
D
ah
l
an
.
A
l
l
r
i
g
h
t
s r
eser
ved
.
1.
I
n
tr
o
d
u
c
ti
o
n
As
w
e
a
ll k
n
o
w
,
A
r
t
if
ic
ia
l B
e
e
C
o
lo
n
y
(A
B
C
) [
1
,
2]
i
s
an
i
nt
el
l
i
gent
opt
i
m
i
z
at
i
o
n
al
g
or
i
t
hm
s
i
m
i
t
at
i
ng b
ees
f
or
agi
ng i
n t
he na
t
ur
e,
w
h
i
c
h has
t
he c
har
ac
t
er
of
eas
i
l
y
i
m
pl
em
ent
and s
e
t
t
i
ng
par
am
et
er
s
.
S
t
at
e
-
of
-
t
he ar
t
of
t
h
e f
i
el
d t
h
e r
ep
or
t
i
s
t
hat
ar
t
i
f
i
c
i
al
be
e
c
ol
o
n
y
c
an
be
us
ed f
or
anom
al
y
-
bas
e
d i
n
t
r
us
i
on
d
et
ec
t
i
on s
y
s
t
em
s
.
A
l
s
o
a
n ar
t
i
f
i
c
i
al
b
ee c
o
l
on
y
al
gor
i
t
hm
i
s
pr
es
ent
e
d
f
or
da
t
a
c
o
l
l
ec
t
i
o
n
pa
t
h
pl
ann
i
n
g
i
n
s
p
ar
s
e
w
i
r
el
es
s
s
ens
or
n
et
w
or
k
s
.
A
B
C
a
l
gor
i
t
hm
i
s
us
ed
f
or
f
unc
t
i
on
op
t
i
m
i
z
at
i
o
n
and
c
om
bi
nes
gen
et
i
c
al
gor
i
t
hm
,
pa
r
t
i
c
l
e
s
w
ar
m
al
gor
i
t
hm
and
f
i
ni
t
e d
i
f
f
er
enc
e al
g
or
i
t
hm
t
o s
ol
v
e s
om
e c
o
m
pl
ex
pr
ob
l
em
s
es
pec
i
al
l
y
T
S
P
pr
o
bl
e
m
s
[
3
,
4
]. A
B
C
al
g
or
i
t
hm
al
s
o i
s
us
ed
i
n
ne
ur
al
n
et
w
or
k
t
r
ai
ni
ng
an
d d
i
gi
t
a
l
I
I
R
f
i
l
t
er
des
i
gni
ng.
H
o
w
e
v
er
,
i
n t
he pr
ac
t
i
c
al
eng
i
ne
er
i
n
g app
l
i
c
a
t
i
o
n,
m
an
y
pr
oduc
t
i
on pr
ac
t
i
c
e pr
obl
em
s
ar
e
t
r
ans
f
or
m
ed
i
nt
o
h
i
gh
di
m
ens
i
ona
l
c
om
pl
ex
f
unc
t
i
on
o
pt
i
m
i
z
a
t
i
o
n
pr
o
bl
em
s
.
B
ut
i
t
h
as
c
har
ac
t
er
s
of
f
unc
t
i
on c
om
pl
ex
,
gr
eat
s
c
al
e,
hi
gh
di
m
ens
i
ons
and no
nl
i
ne
ar
.
W
hen w
e us
e
c
l
as
s
i
c
al
op
t
i
m
i
z
at
i
on m
et
h
od t
o s
ol
v
e t
hi
s
qu
es
t
i
o
n,
i
t
i
s
eas
y
t
o f
al
l
i
n
t
o l
oc
al
ex
t
r
em
u
m
w
i
t
h t
he
i
nc
r
eas
e
of
di
m
ens
i
on.
A
B
C
ex
i
s
t
s
s
om
e di
s
ad
v
ant
ag
es
,
s
uc
h as
pr
em
at
ur
e c
on
v
er
ge
nc
e,
eas
y
t
o
f
al
l
i
nt
o
l
oc
al
ex
t
r
em
u
m
and
l
o
w
s
ol
ut
i
on.
S
o
M
ans
our
i
[
5]
pr
opos
ed
a
n
ov
el
i
t
er
at
i
v
e
m
et
hod
c
o
m
bi
ni
ng
A
B
C
a
nd
B
i
s
ec
t
i
on
m
et
hod
t
o
f
i
n
d
t
he
f
i
x
e
d
poi
nt
of
a
non
l
i
near
f
unc
t
i
on
ef
f
ec
t
i
v
e
l
y
.
I
m
ani
an
[
6]
pr
o
pos
ed
a
m
odi
f
i
ed
A
B
C
al
g
or
i
t
hm
c
al
l
ed
V
A
B
C
t
o
ov
er
c
om
e
t
hi
s
i
n
s
uf
f
i
c
i
enc
y
b
y
app
l
y
i
ng
a
ne
w
s
e
ar
c
h eq
u
at
i
o
n
i
n
t
he
o
nl
o
ok
er
phas
e
.
I
t
us
ed
t
he
P
S
O
s
e
ar
c
h s
t
r
at
eg
y
t
o g
ui
de
t
he s
ear
c
h f
or
c
andi
dat
e s
ol
ut
i
o
ns
.
W
ang
[
7]
pr
o
pos
ed a n
ov
el
m
ul
t
i
-
s
t
r
at
e
g
y
ens
em
bl
e A
B
C
al
g
or
i
t
hm
.
A
po
ol
of
di
s
t
i
nc
t
s
ol
u
t
i
o
n
s
e
ar
c
h
s
t
r
at
e
gi
es
c
oex
i
s
t
ed
t
hr
o
ugh
out
t
he
s
ear
c
h
pr
oc
es
s
and
c
om
pet
ed t
o pr
o
duc
e
o
f
f
s
pr
i
ng.
Mus
t
af
a
[
8]
pr
o
pos
ed i
nt
e
gr
at
i
on
of
m
ul
t
i
p
l
e s
ol
ut
i
o
n up
dat
e
r
ul
es
w
i
t
h
A
B
C
w
h
i
c
h us
e
d f
i
v
e
s
ear
c
h s
t
r
a
t
eg
i
es
a
nd c
ou
nt
er
s
t
o u
pd
at
e
t
he
s
ol
ut
i
o
ns
.
P
.
Mus
t
af
a
[
9]
a
dde
d
d
i
r
ec
t
i
o
nal
i
nf
or
m
at
i
on
t
o
A
B
C
a
l
g
or
i
t
hm
s
.
T
he
ne
w
s
c
hem
e
w
as
c
om
par
ed
w
i
t
h
bas
i
c
A
B
C
a
nd
A
B
C
s
w
i
t
h
MR
.
I
t
ex
am
i
ned
t
he
p
er
f
or
m
anc
e
of
t
hi
s
m
et
hod
on
w
e
l
l
-
k
n
ow
n
ni
n
e num
er
i
c
al
b
enc
hm
ar
k
f
unc
t
i
ons
t
o s
h
o
w
i
t
ef
f
ec
t
i
v
el
y
.
T
he abo
v
e
i
m
pr
ov
ed
A
B
C
al
g
or
i
t
hm
s
ac
hi
ev
e
goo
d
ef
f
ec
t
,
B
ut
t
h
e
y
al
s
o
has
s
om
e
di
s
ad
v
a
nt
ag
es
w
i
t
h m
or
e c
ont
r
ol
par
am
et
er
s
,
pr
em
a
t
ur
e c
on
v
er
g
enc
e
et
a
l
.
T
o s
ol
v
e t
hos
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
16
93
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
1
4
,
N
o
.
3
,
S
ept
em
ber
201
6
:
10
99
–
1
104
1100
ques
t
i
ons
a
nd
i
m
pr
ov
e
s
ol
ut
i
o
n
pr
ec
i
s
i
on
an
d
c
on
v
er
ge
nc
e
s
p
eed
of
A
B
C
al
g
or
i
t
hm
,
w
e
pr
opos
e
an
i
m
pr
ov
ed s
t
a
ge
d s
ear
c
h A
r
t
i
f
i
c
i
a
l
B
ee C
ol
o
n
y
(
S
S
A
B
C
)
al
gor
i
t
hm
.
T
hi
s
ne
w
s
c
hem
e
adop
t
s
i
n
i
t
i
al
i
z
at
i
on
s
t
r
at
e
g
y
b
as
ed
on
un
i
f
or
m
di
s
t
r
i
b
ut
i
o
n a
nd
bac
k
w
ar
d
l
e
ar
ni
ng.
I
t
i
m
pr
ov
es
t
he q
ual
i
t
y
of
i
n
i
t
i
al
hon
e
y
and
enh
anc
es
g
l
ob
al
s
e
ar
c
hi
ng
ab
i
l
i
t
y
of
bee c
o
l
o
n
y
.
W
e di
v
i
de
em
pl
o
y
ed
bee s
e
ar
c
h beh
av
i
or
i
n
t
o t
w
o s
t
ages
.
T
he
t
w
o s
t
ages
us
e
di
f
f
er
ent
s
ear
c
h s
t
r
at
eg
y
w
hi
c
h
i
m
pr
ov
es
t
h
e
s
ol
ut
i
o
n
pr
ec
i
s
i
on
and
c
on
v
er
g
en
c
e
s
peed
of
A
B
C
a
l
g
or
i
t
hm
.
T
he
f
ol
l
o
w
i
n
g
bee us
es
ad
apt
i
v
e opt
i
m
i
z
at
i
o
n s
t
r
at
eg
y
t
o e
nha
nc
e
pr
oduc
t
i
on ab
i
l
i
t
y
of
f
ol
l
o
w
i
ng b
ee.
W
e
def
i
ne
t
he
es
c
ape
r
ad
i
us
t
o gu
i
d
e es
c
ap
e d
i
r
ec
t
i
on
of
pr
e
c
oc
i
ous
i
n
di
v
i
dua
l
w
hi
c
h
r
e
duc
es
bl
i
ndn
es
s
of
i
ndi
v
i
d
ua
l
es
c
ape ef
f
ec
t
i
v
el
y
.
O
ur
m
ai
n c
ont
r
i
b
ut
i
on i
s
t
ha
t
w
e
i
m
pr
ov
e t
he
em
pl
o
y
ee
bees
,
f
ol
l
o
w
i
n
g b
ees
and s
c
ou
t
s
r
es
pec
t
i
v
e
l
y
.
T
hen ex
p
er
i
m
ent
s
r
es
ul
t
s
s
how
t
hat
our
ne
w
m
et
hod
has
hi
g
h
ef
f
i
c
i
enc
y
t
ha
n
or
i
g
i
na
l
A
B
C
.
T
he
f
ol
l
o
w
i
ng
i
s
t
h
e
s
t
r
uc
t
ur
e
of
our
pa
per
.
T
he
nex
t
s
ec
t
i
o
n
i
s
t
he
det
ai
l
ed
i
m
pr
ov
ed
A
B
C
a
l
gor
i
t
hm
.
S
ec
t
i
on
3
i
s
t
he
ex
per
i
m
ent
s
r
es
ul
t
s
.
T
he l
as
t
s
ec
t
i
o
n i
s
a c
onc
l
u
s
i
on.
2.
T
h
e
I
m
p
r
o
v
e
d
S
ta
g
e
d
S
e
a
r
c
h
A
r
ti
fi
ci
al
B
ee C
o
l
o
n
y
2
.
1
U
n
i
fo
r
m
D
e
s
i
g
n
i
n
g
a
n
d
B
a
c
k
w
a
r
d
L
e
a
r
n
i
n
g
f
o
r
I
n
i
ti
a
l
i
z
a
ti
o
n
S
tr
a
te
g
y
T
he i
ni
t
i
al
hon
e
y
i
s
t
he a
l
gor
i
t
hm
s
ear
c
h or
i
gi
n.
I
n
A
B
C
a
l
g
or
i
t
hm
,
i
ni
t
i
a
l
ho
ne
y
i
s
pr
oduc
e
d r
andom
l
y
(
i
.
e.
i
t
g
ener
at
es
s
ev
er
a
l
i
n
di
v
i
dua
l
t
o f
or
m
i
ni
t
i
a
l
gr
ou
p)
.
I
f
t
he i
ni
t
i
al
gr
oup
i
s
gener
at
ed
unr
e
as
ona
bl
y
.
I
t
w
i
l
l
h
av
e
an
ef
f
ec
t
on
t
he
abi
l
i
t
y
of
g
l
ob
al
opt
i
m
i
z
at
i
o
n.
S
o
w
e
m
us
t
i
m
pr
ov
e t
he ge
ner
at
i
on m
et
hod of
i
ni
t
i
a
l
gr
ou
p t
o m
a
k
e
t
he i
n
i
t
i
al
i
ndi
v
i
du
al
d
i
s
t
r
i
b
u
t
i
on u
ni
f
or
m
l
y
and
has
a bet
t
er
q
ual
i
t
y
.
R
ef
er
enc
e
[
1
0]
ad
opt
s
uni
f
or
m
t
o des
i
g
n
a
nd
i
n
i
t
i
a
l
i
z
e t
h
e
gr
oup.
I
t
ens
ur
es
t
h
e i
ni
t
i
a
l
n
ec
t
ar
s
our
c
e di
s
t
r
i
but
ed
w
i
t
h
i
n t
he s
ear
c
h s
p
ac
e u
ni
f
or
m
l
y
.
B
ut
i
t
c
ann
ot
ens
ur
e t
hat
nec
t
ar
s
our
c
e
i
s
g
ood
.
I
f
A
B
C
us
es
i
n
i
t
i
al
i
z
at
i
on s
t
r
at
eg
y
bas
ed
on
bac
k
w
ar
d
l
ear
n
i
n
g,
i
t
w
i
l
l
ens
ur
e
t
h
e qua
l
i
t
y
of
ne
c
t
ar
s
our
c
e
a
n
d
c
an
not
ens
ur
e
t
h
e
uni
f
or
m
di
s
t
r
i
but
i
on.
S
o
t
hi
s
l
et
t
er
c
om
bi
nes
b
a
c
k
w
ar
d
l
ear
n
i
n
g an
d
u
ni
f
or
m
des
i
gni
ng.
T
he
det
ai
l
ed
pr
oc
es
s
es
ar
e
a
s
f
o
llo
w
s
.
a)
W
e
uni
f
or
m
l
y
d
i
v
i
de
t
h
e
v
al
ue
r
an
ge
of
pr
ep
ar
at
i
v
e
op
t
i
m
i
z
at
i
on
i
nt
o
SN
s
u
bs
pac
es
.
I
t
w
i
l
l
r
an
do
m
l
y
pr
oduc
e
a
n i
ni
t
i
a
l
s
o
l
ut
i
on
f
r
om
ev
er
y
s
u
bs
pac
e
and
f
or
m
i
ni
t
i
a
l
i
nd
i
v
i
d
ual
a
s
E
qu
at
i
on (
1)
.
)
)(
1
,
0
(
mi
n
,
ma
x
,
mi
n
,
,
sn
j
sn
j
sn
j
sn
j
i
x
x
r
an
x
x
−
+
=
(
1)
b)
I
t
s
ol
v
es
r
e
v
er
s
e s
ol
ut
i
on
of
eac
h
i
ni
t
i
a
l
s
o
l
ut
i
on
sn
j
i
Ox
,
as
E
q
uat
i
on
(2
).
sn
j
sn
j
sn
j
sn
j
i
x
x
x
Ox
mi
n
,
ma
x
,
mi
n
,
,
−
+
=
(
2)
W
h
er
e
sn
j
i
x
,
denot
es
i
-
t
h
(
1
≤
i≤
S
N
)
nec
t
ar
s
our
c
e a
nd
j
-
di
m
ens
i
on
i
-
t
h(
1≤
j
≤
D
)
c
o
or
di
nat
e at
sn
-
t
h(
1≤
s
n≤
S
N
)
s
ubi
nt
er
v
al
.
j
x
mi
n
,
and
sn
j
x
ma
x
,
ar
e
m
i
ni
m
u
m
nec
t
ar
s
our
c
e and m
ax
i
m
al
nec
t
ar
s
our
c
e r
es
pec
t
i
v
e
l
y
.
2.
2
.
S
ta
g
e
d
E
m
p
l
o
yed
B
e
es S
ea
r
ch
S
t
r
at
eg
y
I
n
t
h
e
ear
l
y
s
t
age
of
s
ol
v
i
n
g
pr
ob
l
em
,
hi
r
e
b
ees
s
ear
c
h
be
ha
v
i
or
s
h
ou
l
d
h
av
e
s
t
r
onge
r
s
ear
c
hi
ng
ab
i
l
i
t
y
a
nd f
ul
l
y
e
x
pl
or
e t
he s
e
ar
c
h s
pac
e.
I
t
m
a
k
e
s
a goo
d pr
e
par
at
i
on
f
or
s
ubs
equ
ent
f
ol
l
o
w
i
ng b
ees
m
i
ni
ng ac
t
i
v
i
t
i
es
.
I
n t
he l
at
e s
t
ag
e of
s
ol
v
i
ng pr
o
bl
em
,
al
gor
i
t
h
m
c
onv
er
ges
t
o
gl
o
bal
opt
i
m
al
s
o
l
ut
i
on
.
H
i
r
e bees
s
ho
ul
d
ha
v
e s
t
r
o
nger
s
ear
c
hi
ng
abi
l
i
t
y
and
i
m
pr
ov
e t
h
e
c
onv
er
g
enc
e
r
at
e
of
pr
ob
l
e
m
.
I
n
or
der
t
o
bet
t
er
ac
c
o
m
m
odat
e
t
he
s
ear
c
h
r
eque
s
t
of
e
m
pl
o
y
ed
bee,
t
h
i
s
paper
des
i
gns
a
s
t
age
d
em
pl
o
y
ed
b
ees
s
ear
c
h
s
t
r
at
eg
y
.
F
i
r
s
t
s
t
age:
h
i
r
e
bees
s
ear
c
h
beha
v
i
or
s
ho
ul
d
ha
v
e s
t
r
o
nger
ex
p
l
or
at
i
o
n c
om
pet
enc
e
w
h
i
c
h c
an f
ul
l
y
ex
pl
o
r
e t
he s
e
ar
c
h
s
pac
e an
d ac
c
e
l
er
at
e t
h
e e
m
er
genc
e of
gl
ob
al
opt
i
m
al
s
ol
ut
i
o
n;
S
ec
on
d s
t
ag
e:
hi
r
e be
es
s
ear
c
h
beha
v
i
or
s
hou
l
d h
av
e s
t
r
onger
m
i
ni
ng a
bi
l
i
t
y
a
nd pr
om
ot
e t
he c
onv
er
genc
e
r
a
t
e of
al
gor
i
t
hm
.
T
he det
ai
l
e
d pr
oc
es
s
es
ar
e
as
f
ol
l
o
w
s
:
a)
A
t
f
i
r
s
t
s
t
age
,
w
e
us
e
E
q
uat
i
on (
3)
t
o s
ear
c
h.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
K
A
I
S
S
N
:
1
693
-
6
930
A
n I
mpr
ov
ed
A
r
t
i
f
i
c
i
al
B
e
e
C
ol
o
ny
A
l
gor
i
t
h
m f
or
S
t
a
ge
d S
e
ar
c
h
(
S
h
o
u
lin
Y
in
)
1101
)
(
)
(
,
,
,
,
,
,
j
r
j
b
es
t
j
k
j
i
j
i
j
i
x
x
x
x
x
v
−
+
−
+
=
ϕ
β
(
3)
W
h
er
e
β
i
s
a
r
andom
nu
m
ber
of
[
-
1,
1]
.
φ
i
s
a
r
a
ndo
m
nu
m
ber
of
[
-
1
,1
].
j
b
es
t
x
,
is
j
-
di
m
ens
i
on
c
oor
di
n
at
e
of
c
ur
r
ent
g
l
o
bal
opt
i
m
al
s
ol
ut
i
o
n,
k
≠r
≠i
.
B
e
c
aus
e
i
t
ad
ds
t
h
e
o
pt
i
m
al
p
os
i
t
i
on
j
b
es
t
x
,
,
i
t
i
m
pr
ov
es
t
he m
i
ni
n
g ab
i
l
i
t
y
t
o s
om
e ex
t
ent
.
b)
S
ec
ond s
t
a
ge.
S
t
r
engt
h
of
hi
r
e bees
m
i
ni
ng ab
i
l
i
t
y
has
an ef
f
ec
t
on t
he c
onv
er
genc
e
r
at
e of
al
gor
i
t
hm
.
S
o t
h
i
s
pa
per
pr
op
os
es
an
ad
apt
i
v
e s
ear
c
h s
t
r
at
eg
y
as
s
ho
w
n
i
n
(
4)
.
)
(
)
1
(
,
,
ma
x
,
,
j
i
j
b
es
t
j
i
j
i
x
x
t
t
x
v
−
×
−
×
+
=
λ
ω
(
4)
W
h
er
e
t
ma
x
i
s
t
he
m
ax
i
m
um
nu
m
ber
of
i
t
er
at
i
on
.
t
i
s
t
he beg
i
n
ni
n
g i
t
er
a
t
i
o
n
nu
m
ber
of
s
ec
ond
s
t
age.
w
i
s
a r
and
om
i
nt
eg
er
num
ber
of
[
-
1,
1]
.
λ
i
s
a
r
andom
num
ber
of
(
0,
1)
.
w
c
an ens
ur
e t
hat
t
he s
e
ar
c
h r
an
ge
i
s
n
ot
l
i
m
i
t
ed t
o t
he
di
r
ec
t
i
on
of
)
(
,
,
j
i
j
b
e
st
x
x
−
.
I
t
c
a
n s
ear
c
h t
he
ne
i
gh
bor
h
oo
d
of
j
i
x
,
r
ound
l
y
.
λ
pr
ev
ent
s
t
e
nd
i
ng
t
o
z
er
o
at
t
h
e
en
d
of
i
t
e
r
at
i
on
al
gor
i
t
hm
.
)
1
(
ma
x
t
t
λ
ω
−
×
w
i
ll
i
nc
r
eas
e
w
i
t
h t
he i
nc
r
eas
i
ng of
t
.
S
e
ar
c
h r
ang
e of
c
ol
on
y
r
ed
uc
es
gr
adu
al
l
y
and pr
o
duc
t
i
o
n
abi
l
i
t
y
of
hi
r
e b
ee s
t
r
en
gt
h
e
ns
gr
adu
al
l
y
.
c)
I
f
t
he f
unc
t
i
on
v
al
u
e of
f
ood
s
our
c
e
V
i
i
s
s
uper
i
or
t
o
X
i
,
t
hen
V
i
w
il
l r
e
p
la
c
e
X
i
.
2.
3
.
F
o
l
l
o
w
i
n
g
B
ee
S
tr
a
te
g
y
o
f
A
d
ap
t
i
v
e L
o
cal
S
ea
r
ch
.
A
t
s
ear
c
h
s
t
age
,
f
ol
l
o
w
i
n
g
bee s
el
ec
t
s
bet
t
er
n
ec
t
ar
s
our
c
e t
o ex
pl
or
e and
d
ev
e
l
o
p
aga
i
n.
S
o
f
ol
l
o
w
i
ng
b
ee
s
ear
c
h
s
houl
d
ha
v
e
s
t
r
on
g
abi
l
i
t
y
of
pr
oduc
t
i
on
.
Mea
n
w
hi
l
e,
i
n
or
d
er
t
o
f
al
l
i
nt
o
l
oc
al
m
i
ni
m
a v
a
l
u
e,
i
t
a
l
s
o s
h
oul
d
ha
v
e
ex
pl
or
at
i
o
n a
bi
l
i
t
y
.
B
as
ed
o
n op
t
i
m
i
z
at
i
on
c
har
ac
t
er
i
s
t
i
c
of
f
ol
l
o
w
i
ng
b
ee,
w
e des
i
gn
a f
ol
l
o
w
i
ng
b
ee s
t
r
at
e
g
y
of
ada
pt
i
v
e
l
oc
a
l
s
ear
c
h.
1)
A
c
c
or
di
n
g t
o t
he pr
o
bab
i
l
i
t
y
s
el
ec
t
i
o
n f
or
m
ul
a
P
i
of
A
B
C
al
gor
i
t
hm
,
f
ol
l
o
w
i
n
g be
e
s
el
ec
t
s
nec
t
ar
s
o
ur
c
e
X
i
t
o
s
ear
c
h opt
i
m
i
z
i
ng
.
f
i
i
s
s
ear
c
h f
unc
t
i
on.
P
i
c
a
n be
c
al
c
u
l
at
e
d b
y
:
∑
=
=
S
N
i
i
i
i
f
f
P
1
/
(
5)
2)
A
c
c
or
di
n
g t
o (
4)
,
w
e s
t
ar
t
l
o
c
al
s
ear
c
h (
t
i
s
c
ur
r
ent
i
t
er
a
t
i
on
num
ber
)
.
3)
W
h
en
f
ol
l
o
w
i
n
g
bee
be
gi
ns
l
oc
al
s
ear
c
h,
hi
r
e
b
ee
r
es
ear
c
hes
nec
t
ar
s
our
c
e
i
n
s
t
ages
t
o
i
m
pr
ov
e c
on
v
er
g
enc
e r
at
e
,
k
eep po
pu
l
at
i
on
di
v
er
s
i
t
y
a
nd j
um
p out
of
l
o
c
al
op
t
i
m
u
m
.
4)
I
t
c
om
par
es
t
he
ne
w
nec
t
a
r
s
our
c
e
V
i
of
f
ol
l
o
w
i
ng
bee
,
ne
w
n
ec
t
ar
s
our
c
e
V'
i
o
f
hi
r
e
bee
and
ol
d
n
ec
t
ar
s
our
c
e
X
i
a
nd
s
e
l
ec
t
s
n
ec
t
ar
s
our
c
e
w
i
t
h
b
et
t
er
f
i
t
nes
s
v
al
u
e
as
ne
w
nec
t
a
r
s
our
c
e.
2.
4
.
E
sc
ap
e
S
co
u
t
er
S
t
r
at
eg
y
I
n
A
B
C
a
l
gor
i
t
hm
,
s
c
out
er
i
s
i
n
c
har
ge
of
f
i
ndi
ng
t
h
e
p
r
em
at
ur
e
c
onv
er
genc
e
i
ndi
v
i
dua
l
and u
pdat
i
n
g al
gor
i
t
hm
w
hi
c
h c
an r
educ
e t
h
e pr
ob
ab
i
l
i
t
y
of
pr
em
at
ur
e c
onv
er
g
e
nc
e.
B
ec
a
us
e
t
he ex
i
s
t
i
ng
A
B
C
al
g
o
r
i
t
hms
[
11
-
13]
ha
v
e t
he d
ef
ec
t
of
r
es
t
r
i
c
t
i
ng t
he
es
c
ap
e of
pr
ec
oc
i
ous
in
d
iv
id
u
a
l
[
13
]
.
W
e
dei
gn
a
ne
w
es
c
ap
e
s
c
out
er
s
t
r
at
eg
y
.
I
f
X
i
i
s
pr
ec
oc
i
ous
i
nd
i
v
i
dua
l
.
T
hen
i
t
s
ho
w
s
t
hat
X
i
f
al
l
s
i
nt
o
l
oc
al
ex
t
r
em
um
w
i
t
h
i
t
s
el
f
as
c
ent
er
and
ε
as
r
a
di
us
.
W
e
def
i
ne
nei
ghb
or
hoo
d r
ang
e as
e
x
t
r
e
m
u
m
nei
ghbor
hoo
d
w
hi
c
h r
es
ul
t
s
i
n i
n
di
v
i
dua
l
f
al
l
i
n
g i
nt
o l
oc
a
l
ex
t
r
em
u
m
.
R
adi
us
ε
of
ex
t
r
em
u
m
nei
ghb
or
ho
od
i
s
es
c
ape
r
ad
i
us
of
X
i
.
X
i
nee
ds
t
o
j
u
m
p
out
of
l
oc
al
ex
t
r
em
u
m
poi
nt
,
i
t
m
us
t
m
a
k
e t
he X
i
es
c
ap
e
ex
t
r
em
al
nei
ghb
or
hoo
d.
T
he
d
et
ai
l
e
d s
t
r
at
e
g
y
is
a
s
b
e
lo
w
:
S
t
ep
1.
S
et
t
i
ng
m
ai
n
i
ni
t
i
a
l
par
am
et
er
s
:
t
he
num
ber
of
popu
l
at
i
o
n
(
SN
)
,
m
ax
i
m
um
c
y
c
l
e
t
im
e
s
(
m
a
xC
ycl
e
)
,
par
am
et
er
di
m
ens
i
on (
D
)
,
m
i
ni
n
g b
ees
an
d obs
er
v
at
i
o
n be
es
r
epr
es
ent
abo
ut
50 p
er
c
ent
of
t
ot
al
.
O
ne s
c
o
ut
b
ee.
S
t
ep
2.
E
x
ec
ut
i
ng
t
he
i
n
i
t
i
al
i
z
at
i
on s
t
r
at
e
g
y
of
un
i
f
or
m
di
s
t
r
i
b
ut
i
on
-
r
e
v
er
s
e l
ear
n
i
n
g.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
16
93
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
1
4
,
N
o
.
3
,
S
ept
em
ber
201
6
:
10
99
–
1
104
1102
S
t
ep
3.
C
al
c
u
l
at
i
n
g f
i
t
nes
s
v
al
ue
of
i
ni
t
i
a
l
pop
ul
a
t
i
o
n an
d r
ec
or
d
i
ng
c
ur
r
ent
o
pt
i
m
al
s
ol
ut
i
on.
S
t
ep
4.
E
x
ec
ut
i
ng
h
i
r
e
b
ee
s
ear
c
h
s
t
r
at
eg
y
i
n
s
t
ages
and
r
ec
or
d
i
ng
t
he
c
ur
r
e
nt
opt
i
m
al
s
ol
ut
i
on.
S
t
ep
5.
E
x
ec
ut
i
ng f
ol
l
o
w
i
n
g
bees
s
t
r
at
e
g
y
of
ada
pt
i
v
e
l
oc
al
s
ear
c
h.
S
t
ep
6.
I
f
t
her
e ex
i
s
t
s
r
en
u
nc
i
at
i
v
e nec
t
ar
s
our
c
e,
t
he
n
m
i
ni
ng be
e
w
i
l
l
be
c
ha
n
ges
as
obs
er
v
a
t
i
on
bees
i
n
t
hi
s
ar
ea.
I
t
w
i
l
l
pr
o
duc
e
ne
w
nec
t
ar
s
our
c
e
ac
c
or
d
i
ng
t
o t
he
es
c
ape s
c
out
s
s
ear
c
h s
t
r
at
eg
y
a
nd c
a
l
c
ul
a
t
e f
i
t
nes
s
v
al
ue.
G
l
ob
al
opt
i
m
al
s
ol
ut
i
o
n
w
i
l
l
be
s
t
or
ed.
S
t
ep
7.
A
d
di
n
g
c
y
c
l
e
t
i
m
es
.
D
et
er
m
i
ni
n
g
i
f
i
t
i
s
gr
e
at
e
r
t
han
m
ax
C
y
c
l
e.
I
f
Y
E
S
,
t
hen
g
o
t
o s
t
ep
8.
I
f
N
O
,
t
he
n b
ac
k
s
t
ep4.
S
t
ep 8
.
R
eac
h
i
ng
m
a
x
C
yc
l
e
an
d di
s
c
ont
i
n
ui
n
g al
gor
i
t
hm
.
I
t
w
i
l
l
o
ut
p
ut
gl
oba
l
o
pt
i
m
al
s
ol
ut
i
on.
3.
S
i
m
u
l
a
ti
o
n
R
e
s
u
l
ts
a
n
d
A
n
al
y
si
s
.
I
n
or
d
er
t
o
v
er
i
f
y
t
h
e
s
up
er
i
or
i
t
y
of
t
h
i
s
al
gor
i
t
hm
,
w
e
s
el
ec
t
E
S
A
B
C
(
E
l
i
t
e
S
w
ar
m
A
B
C
)
al
g
or
i
t
hm
,
MA
B
C
(
Mod
i
f
i
e
d A
r
t
i
f
i
c
i
al
B
ee
C
ol
on
y
)
a
l
gor
i
t
hm
,
A
B
C
M
S
S
(
A
r
t
i
f
i
c
i
al
B
ee
C
o
l
o
n
y
A
l
g
or
i
t
hm
w
i
t
h Mo
di
f
i
e
d S
e
ar
c
h S
t
r
at
e
g
y
)
a
l
gor
i
t
hm
t
o
c
o
m
par
e.
S
N
=
4
0,
max
C
y
c
l
e =
1
000,
D
=
200
.
A
l
g
or
i
t
hm
r
uns
i
ndep
end
e
nt
l
y
5
0 t
i
m
es
under
MA
T
L
A
B
pl
at
f
or
m
.
T
abl
e1 s
ho
w
s
ei
gh
t
h
ig
h
-
di
m
ens
i
on
al
c
om
pl
ex
f
unc
t
i
ons
opt
i
m
i
z
at
i
on
c
om
put
i
ng r
es
ul
t
s
b
y
t
he
f
our
al
g
or
i
t
hm
s
.
W
e
us
e t
he t
ex
t
f
unc
t
i
ons
i
n t
a
bl
e
1 t
o t
es
t
per
f
or
m
anc
e f
or
t
he f
ou
r
al
g
or
i
t
hm
s
.
A
nd
w
e e
v
al
uat
e t
h
i
s
al
g
or
i
t
hm
f
r
o
m
m
ean,
s
t
an
dar
d
d
ev
i
at
i
on
,
t
h
e
opt
i
m
al
v
a
l
ue,
t
he
w
or
s
t
v
al
ue
a
n
d
av
er
ag
e
t
i
m
e
c
os
t
f
i
v
e
as
pec
t
s
.
Mean
v
a
l
ue
a
nd
opt
i
m
al
v
al
ue
c
an
r
epr
es
ent
t
h
e
c
onv
er
genc
e
pr
ec
i
s
i
on
an
d
opt
i
m
i
z
at
i
o
n
c
ap
abi
l
i
t
y
of
al
g
or
i
t
hm
.
F
ro
m
T
abl
e
1,
w
e
c
an
k
no
w
t
hat
w
he
n
s
ol
v
i
n
g
h
i
gh
-
di
m
ens
i
ona
l
(
200)
un
i
m
odal
opt
i
m
i
z
at
i
on
pr
ob
l
em
,
S
S
A
B
C
a
l
g
or
i
t
hm
i
s
r
e
m
ar
k
ab
l
y
h
i
gh
er
t
han
ot
her
t
hr
ee a
l
g
or
i
t
hm
s
.
S
S
A
B
C
al
gor
i
t
hm
al
m
os
t
f
i
nds
t
he t
h
eor
et
i
c
al
opt
i
m
al
s
ol
u
t
i
on f
or
S
pher
e
f
unc
t
i
on (
R
eac
h
i
n
g
1
0
-
95
)
and S
um
s
quar
es
f
unc
t
i
on
(
R
eac
hi
ng 10
-
86
)
.
O
p
t
im
iz
a
t
io
n
p
r
e
c
is
io
n
of
t
hi
s
t
w
o f
unc
t
i
ons
c
an r
e
ac
h 10
-
17
0
.
S
S
A
B
C
get
s
t
he s
am
e ef
f
ec
t
f
or
hi
gh
-
di
m
ens
i
o
nal
m
ul
t
i
-
m
odal
f
unc
t
i
on:
G
r
i
e
w
ank
f
unc
t
i
on,
R
as
t
r
i
g
i
n f
unc
t
i
on an
d A
c
k
l
e
y
f
unc
t
i
on.
I
t
c
onduc
t
s
f
i
fty
opt
i
m
i
z
at
i
o
n ex
per
i
m
ent
s
and t
h
e t
hr
e
e f
unc
t
i
ons
r
eac
h 10
-
1
6.
F
or
R
os
en
br
oc
k
f
unc
t
i
on,
S
c
h
w
ef
el
2.
26
f
unc
t
i
o
n,
Z
ak
har
ov
f
unc
t
i
o
n,
d
ue
t
o t
he
i
r
o
w
n f
eat
ur
es
,
t
h
e a
l
gor
i
t
hm
eas
i
l
y
f
al
l
s
i
nt
o
l
oc
a
l
ex
t
r
em
u
m
v
al
u
e.
B
ut
S
S
A
B
C
s
t
i
l
l
obt
ai
ns
i
dea
l
s
ol
ut
i
on.
S
o
w
e c
an
c
onc
l
ud
e t
h
at
S
S
A
B
C
a
l
gor
i
t
hm
s
how
s
g
ood
abi
l
i
t
y
of
m
i
ni
ng an
d e
x
pl
or
at
i
on
,
i
t
i
s
m
or
e
s
ui
t
a
bl
e f
or
s
ol
v
i
n
g
h
ig
h
-
di
m
ens
i
on
al
c
om
pl
ex
opt
i
m
i
z
at
i
o
n pr
o
bl
em
s
.
S
t
andar
d
de
v
i
at
i
on a
nd
t
he
w
o
r
s
t
v
al
ue r
e
ec
t
s
t
he
al
g
or
i
t
hm
s
r
obus
t
nes
s
and
t
he
ab
i
l
i
t
y
t
o
aga
i
ns
t
t
he
l
oc
a
l
ex
t
r
em
u
m
.
S
t
an
da
r
d
de
v
i
at
i
on
of
S
S
A
B
C
a
l
gor
i
t
hm
i
s
s
m
a
l
l
ex
c
ept
S
c
h
w
ef
e
l
2.
2
6 f
unc
t
i
o
n.
S
t
a
ndar
d d
ev
i
at
i
on of
S
ph
er
e
al
g
or
i
t
hm
and
S
um
s
quar
es
f
unc
t
i
on
r
eac
hes
10
-
94
a
n
d
10
-
83
r
es
pec
t
i
v
el
y
.
S
t
an
dar
d
de
v
i
at
i
on
of
G
r
i
ew
ank
f
unc
t
i
on,
R
as
t
r
i
gi
n
f
unc
t
i
on
an
d
A
c
k
l
e
y
f
unc
t
i
o
n
i
s
z
er
o.
S
o
S
S
A
B
C
al
g
or
i
t
hm
c
an
m
ai
nt
ai
n t
h
e go
od r
o
bus
t
n
es
s
i
n t
he
opt
i
m
i
z
at
i
on a
l
g
or
i
t
hm
.
F
r
o
m
av
er
age t
i
m
e c
os
t
,
t
he f
our
al
g
or
i
t
hm
s
hav
e
t
he
s
am
e
t
i
m
e
c
os
t
.
S
S
A
B
C
al
gor
i
t
hm
does
n
ot
i
nc
r
eas
e
t
he
c
o
m
pl
ex
i
t
y
of
t
he
al
g
or
i
t
hm
and
i
t
i
s
a
m
or
e
ef
f
i
c
i
ent
al
gor
i
t
hm
f
or
s
ol
v
i
n
g
hi
gh
-
d
i
m
ens
i
ona
l
opt
i
m
i
z
at
i
o
n
q
ues
t
i
on.
I
n or
der
t
o v
er
i
f
y
S
S
A
B
C
al
g
or
i
t
hm
'
s
adv
an
t
ag
e i
n
t
ui
t
i
v
el
y
,
w
e gi
v
e
t
he
abo
v
e
ei
g
ht
f
unc
t
i
o
ns
'
i
m
age t
o an
al
y
s
i
s
as
F
ig
ur
e
1(
a
-
h)
.
F
ig
ur
e
1(
a)
and
F
ig
ur
e
1
(
b)
s
how
t
h
at
S
ph
er
e
f
unc
t
i
on
a
nd
S
um
s
quar
es
f
u
nc
t
i
on
c
ons
t
ant
l
y
s
e
ar
c
h bet
t
er
s
ol
ut
i
o
n bas
e
d o
n S
S
A
B
C
a
l
g
or
i
t
hm
w
i
t
h t
h
e i
nc
r
e
as
e of
i
t
er
at
i
ons
,
and
t
he
y
r
eac
h
t
o
ap
pr
ox
i
m
at
el
y
1
0
-
98
an
d
1
0
-
89
r
es
pec
t
i
v
el
y
w
i
t
h
our
ne
w
m
et
hod,
ho
w
ev
er
,
t
he
ot
her
t
hr
ee
m
et
hods
h
av
e
a
b
ad
v
a
l
u
e.
F
i
g
ur
e
1(
c
)
pr
es
ent
s
t
hat
t
he
r
es
ul
t
s
of
E
S
A
B
C
m
et
hod
ar
e c
l
os
el
y
t
o S
S
A
B
C
,
b
ut
S
S
A
B
C
al
gor
i
t
hm
has
a s
hor
t
c
onv
er
g
enc
e t
i
m
e
and i
t
er
at
i
on.
U
ni
q
ue
l
y
t
h
e
f
our
al
gor
i
t
hm
s
hav
e t
he s
i
m
i
l
ar
r
es
ul
t
s
on
S
c
h
w
ef
el
2.
26
f
unc
t
i
on
.
B
ut
t
he r
es
ul
t
w
it
h
S
S
A
B
C
i
s
s
m
al
l
er
t
ha
n
MA
B
C
a
nd
A
B
C
MS
S
.
F
i
g
ur
e
1(
e,
f
)
obv
i
o
us
l
y
pr
es
ent
s
t
hat
us
i
n
g
S
S
A
B
C
al
g
or
i
t
hm
not
onl
y
dec
r
eas
es
t
he c
onv
er
gen
c
e t
i
m
e
s
har
pl
y
,
b
ut
al
s
o h
as
t
he opt
i
m
al
f
unc
t
i
on
v
a
l
ue,
as
w
e
l
l
as
i
n f
i
g1(
g,
h)
.
O
t
her
t
hr
ee
al
g
or
i
t
hm
s
s
t
ar
t
f
al
l
i
ng
i
n
t
o
l
o
c
al
ex
t
r
em
u
m
v
a
l
ue af
t
er
60
0 i
t
er
at
i
ons
.
F
ig
ur
e
1(
e
-
h)
,
S
S
A
B
C
r
ea
c
hes
opt
i
m
al
s
ol
ut
i
o
n af
t
er
250 i
t
er
at
i
ons
and
33
0
i
t
er
at
i
o
ns
r
es
pec
t
i
v
e
l
y
.
T
hus
S
S
A
B
C
al
gor
i
t
h
m
has
t
he
s
up
er
i
or
g
l
ob
al
ex
pl
or
i
ng
ab
i
l
i
t
y
and t
he
abi
l
i
t
y
t
o j
um
p out
o
f
l
oc
al
ex
t
r
em
u
m
.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KO
M
NI
K
A
I
S
S
N
:
1
693
-
6
930
A
n I
mpr
ov
ed
A
r
t
i
f
i
c
i
al
B
e
e
C
ol
o
ny
A
l
gor
i
t
h
m f
or
S
t
a
ge
d S
e
ar
c
h
(
S
h
o
u
lin
Y
in
)
1103
F
ig
ur
e
1
.
F
u
nc
t
i
o
n
V
al
ue w
i
t
h
f
our
A
l
gor
i
t
hm
s
T
abl
e
1
.
C
om
par
i
s
on
o
f
F
u
nc
t
i
on
O
pt
i
m
i
z
at
i
o
n R
es
u
l
t
s
w
i
t
h
F
our
A
l
g
or
i
t
hm
s
F
unc
t
i
on
A
l
gor
i
t
h
m
A
v
er
age
v
al
ue
S
t
andar
d
dev
i
at
i
on
O
pt
i
m
u
m
v
al
ue
W
o
r
st
va
l
u
e
A
v
er
age t
i
m
e
c
o
s
t
S
pher
e
SS
ABC
2.
16e
-
95
1.
401e
-
94
2.
976e
-
169
9.
883e
-
94
2.
77412
ES
ABC
6.
139e
-
16
3.
255e
-
15
4.
621e
-
38
2.
223e
-
14
2.
50038
M
ABC
8.
995e
1
5.
346e
1
4.
069e
0
1.
978e
2
2.
34832
ABC
M
SS
9.
695e
-
3
1.
154e
-
2
4.
833e
-
4
7.
376e
-
2
2.
50149
S
u
m
s
quar
es
SS
ABC
1.
917e
-
86
4.
184e
-
83
2.
963e
-
172
2.
958e
-
77
2.
77561
ES
ABC
5.
784e
-
9
2.
838e
-
9
1.
946e
-
31
1.
901e
-
8
2.
73472
M
ABC
1.
355e
5
4.
928e
4
3.
143e
4
2.
337e
5
2.
91828
ABC
M
SS
4.
284e
4
1.
475e
4
1.
763e
5
8.
472e
4
2.
65432
R
os
enbr
oc
k
SS
ABC
1.
869e
2
1.
232e
-
1
1.
983e
2
1.
989e
2
3.
01124
ES
ABC
1.
867e
2
1.
534e
-
1
1.
984e
2
1.
988e
2
3.
33189
M
ABC
1.
911e
8
1.
194e
8
5.
578e
6
5.
323e
8
3.
26679
ABC
M
SS
3.
197e
6
2.
345e
6
3.
627e
5
1.
083e
7
3.
03054
S
c
hw
ef
el
2.
26
SS
ABC
-
7.
026e
4
9.
664e
2
-
7.
223e
4
-
6.
806e
4
3.
10204
ES
ABC
-
5.
932e
4
1.
452e
3
-
6.
302e
4
-
5.
587e
4
3.
77659
M
ABC
-
5.
729e
4
1.
249e
3
-
5.
976e
4
-
5.
485e
4
2.
90124
ABC
M
SS
-
5.
858e
4
1.
357e
3
-
6.
259e
4
-
5.
586e
4
3.
33647
G
r
i
ew
ank
SS
ABC
0
0
0
0
6.
00122
ES
ABC
4.
115e
-
1
3.
201e
-
1
0
9.
767e
-
1
6.
78912
M
ABC
1.
519e
1
4.
378e
0
5.
049e
0
2.
649e
1
7.
32321
ABC
M
SS
5.
435e
0
1.
519e
0
2.
576e
0
9.
976e
1
6.
67868
R
as
t
r
i
gi
n
SS
ABC
0
0
0
0
3.
12121
ES
ABC
1.
157e
-
6
8.
129e
-
6
0
5.
749e
-
5
3.
12513
M
ABC
5.
621e
4
1.
608e
4
2.
501e
4
1.
015e
5
3.
25687
ABC
M
SS
1.
867e
4
6.
633e
3
6.
479e
3
3.
817e
4
2.
90153
A
ck
l
e
y
SS
ABC
8.
881e
-
16
0
8.
823e
-
16
8.
824e
-
16
3.
98721
ES
ABC
3.
892e
-
4
1.
983e
-
3
3.
334e
-
12
1.
387e
-
2
2.
99234
M
ABC
9.
367e
0
1.
905e
0
5.
843e
0
1.
261e
1
3.
23232
ABC
M
SS
2.
001e
1
7.
308e
-
3
2.
001e
1
2.
004e
1
3.
02364
Z
ak
har
ow
SS
ABC
2.
219e
0
2.
334e
0
8.
262e
-
3
1.
427e
1
4.
13864
ES
ABC
3.
621e
4
9.
472e
3
1.
758e
4
5.
738e
4
4.
16758
M
ABC
3.
854e
2
4.
867e
2
4.
521e
-
1
2.
657e
3
3.
98789
ABC
M
SS
5.
572e
2
1.
628e
2
5.
149e
-
1
6.
102e
3
3.
62426
F
ro
m
T
abl
e
1,
w
e c
an k
no
w
t
ha
t
a
v
er
ag
e v
al
ue,
s
t
andar
d de
v
i
at
i
on,
opt
i
m
u
m
v
a
l
ue,
w
or
s
t
v
al
u
e a
nd
a
v
er
ag
e
t
i
m
e c
os
t
of
ei
ght
f
unc
t
i
ons
w
i
t
h
t
he
f
our
al
gor
i
t
h
m
s
,
w
h
i
c
h c
an
i
l
l
us
t
r
at
e
t
he
ef
f
i
c
i
enc
y
of
o
ur
ne
w
al
gor
i
t
hm
.
F
or
ex
a
m
pl
e,
a
v
er
ag
e
v
al
ue
i
s
2.
16e
-
9
5
w
h
ic
h
is
t
h
e
s
m
al
l
es
t
v
a
l
ue
w
i
t
h
S
S
A
B
C
a
l
g
or
i
t
hm
f
or
S
pher
e
f
unc
t
i
on.
O
n
S
um
s
quar
es
f
unc
t
i
on,
t
h
e
s
t
andar
d
de
v
i
at
i
on,
opt
i
m
um
v
al
ue
and
w
or
s
t
v
a
l
ue
i
s
4.
184
e
-
8
3
,
2.
96
3e
-
1
72
,
2.
958
e
-
7
7
r
es
pec
t
i
v
el
y
t
h
at
s
ho
w
s
S
S
A
B
C
has
a b
et
t
er
o
pt
i
m
i
z
at
i
on r
es
u
l
t
.
E
s
pec
i
a
l
l
y
,
a
v
er
ag
e v
al
u
e
,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SSN
:
16
93
-
6
930
T
E
L
KO
M
NI
K
A
V
o
l.
1
4
,
N
o
.
3
,
S
ept
em
ber
201
6
:
10
99
–
1
104
1104
s
t
andar
d d
ev
i
at
i
o
n,
opt
i
m
u
m
v
al
ue,
w
or
s
t
v
al
ue r
ea
c
h t
o 0 w
i
t
h S
S
A
B
C
f
or
G
r
i
e
w
ank
and
R
as
t
r
i
gi
n
f
unc
t
i
on.
T
he
l
as
t
i
t
em
av
er
age
t
i
m
e
c
os
t
r
ef
l
ec
t
s
t
hat
us
i
n
g
S
S
A
B
C
al
g
or
i
t
hm
f
or
t
he
ei
g
ht
f
unc
t
i
o
ns
c
an be c
o
n
v
er
ge
nt
w
i
t
h
a f
as
t
s
peed
and a s
h
or
t
t
i
m
e.
A
l
l
a
bo
v
e ex
per
i
m
ent
s
v
a
l
ues
d
em
ons
t
r
at
e t
hat
S
S
A
B
C
a
l
g
or
i
t
hm
does
a
go
od j
ob f
or
f
unc
t
i
o
n op
t
i
m
i
z
at
i
on
pr
ob
l
em
s
.
4.
C
o
n
c
l
u
s
i
o
n
T
hi
s
paper
c
ar
r
i
es
out
opt
i
m
i
z
at
i
on
f
or
t
he
di
f
f
er
ent
s
t
ages
of
s
w
ar
m
al
gor
i
t
hm
.
W
i
t
h
t
he
c
ont
i
n
uous
e
v
o
l
ut
i
o
n of
al
g
or
i
t
hm
,
t
he s
w
ar
m
s
ear
c
h s
t
r
at
eg
i
es
ar
e c
ons
t
ant
l
y
c
h
ang
i
ng t
o m
eet
t
he r
eq
ui
r
em
ent
of
opt
i
m
i
z
at
i
o
n pr
o
bl
em
s
.
W
e
r
edef
i
ne
t
he
es
c
ape
beh
av
i
or
of
t
he pr
ec
oc
i
ous
i
nd
i
v
i
d
ual
t
o m
a
k
e i
t
j
u
m
p
out
of
l
oc
al
ex
t
r
em
u
m
v
al
u
e.
S
o
w
e
pr
op
os
e
t
he
i
m
pr
ov
e
d A
r
t
i
f
i
c
i
a
l
B
ee C
ol
o
n
y
i
n s
t
ag
es
.
T
h
e i
ni
t
i
a
l
i
z
a
t
i
on of
t
hi
s
ne
w
al
g
or
i
t
hm
adopt
s
un
i
f
or
m
di
s
t
r
i
b
ut
i
on an
d
bac
k
w
ar
d l
ear
n
i
ng s
t
r
at
eg
y
w
hi
c
h m
ak
es
up t
he
l
ac
k
of
a s
i
ngl
e m
et
hod
.
I
t
m
ak
es
i
ni
t
i
a
l
s
o
l
ut
i
on
uni
f
or
m
di
s
t
r
i
but
e
i
n
t
he
s
e
ar
c
h
s
pac
e
and
i
m
pr
ov
es
t
he
gl
oba
l
ex
p
l
or
i
ng
a
bi
l
i
t
y
.
W
e
des
i
gn
hi
r
e
bee s
e
ar
c
h s
t
r
at
e
g
y
i
n
s
t
a
ge a
nd m
ak
e t
he hi
r
e
be
e has
di
f
f
er
ent
s
ear
c
h a
bi
l
i
t
y
i
n
d
i
f
f
er
ent
s
t
age.
A
t
f
i
r
s
t
s
t
a
ge,
hi
r
e
bee
has
s
t
r
on
g a
bi
l
i
t
y
of
ex
pl
or
at
i
o
n a
nd f
ul
l
y
ex
p
l
o
r
es
t
he s
e
ar
c
h
s
pac
e,
a
v
oi
ds
al
gor
i
t
hm
f
al
l
i
nt
o l
oc
al
ex
t
r
em
u
m
.
A
t
s
ec
ond s
t
a
ge,
i
t
h
as
s
t
r
ong a
bi
l
i
t
y
of
pr
oduc
t
i
on an
d pr
om
pt
s
al
gor
i
t
hm
t
o c
onv
er
ge
nt
t
o gl
o
bal
o
pt
i
m
al
s
ol
ut
i
on q
ui
c
k
l
y
.
F
o
l
l
ow
i
ng
bee
s
ear
c
h us
es
ad
apt
i
v
e
opt
i
m
i
z
at
i
on
s
t
r
at
eg
y
an
d
ac
c
el
er
a
t
es
t
he
al
gor
i
t
hm
'
s
c
onv
er
genc
e
s
peed.
W
e
def
i
ne
t
h
e
es
c
a
pe
r
ad
i
us
t
o
m
a
k
e
t
he
es
c
ape
beh
av
i
or
of
pr
em
at
ur
e
i
nd
i
v
i
d
ua
l
has
di
r
ec
t
i
on.
S
i
m
ul
at
i
o
n r
es
u
l
t
s
al
s
o
pr
ov
e t
hat
S
S
A
B
C
al
g
or
i
t
hm
ha
s
t
h
e c
har
ac
t
e
r
i
s
t
i
c
s
of
hi
g
h
pr
ec
i
s
i
o
n and f
as
t
c
onv
er
genc
e s
pe
ed.
I
n t
he f
ut
ur
e,
w
e
w
i
l
l
m
a
k
e gr
eat
pr
o
gr
es
s
on A
B
C
al
g
or
i
t
hm
bas
ed o
n t
h
i
s
p
aper
'
s
m
et
hod an
d ap
pl
y
ne
w
s
c
hem
e i
nt
o
pr
ac
t
i
c
a
l
en
gi
n
eer
i
ng
app
l
i
c
at
i
on
.
A
c
k
n
o
w
l
e
d
g
e
m
en
t
s
T
he
aut
hor
s
a
l
s
o
gr
at
ef
ul
l
y
ac
k
now
l
edg
e
t
he
he
l
pf
ul
c
om
m
ent
s
and
s
ugges
t
i
ons
of
t
he
r
ev
i
e
w
er
s
,
w
h
i
c
h h
av
e
i
m
pr
ov
e
d t
h
e pr
es
e
nt
at
i
o
n.
R
ef
er
en
ces
[1
]
R
en Y
,
W
u Y
.
A
n ef
f
i
c
i
en
t
al
g
o
r
i
t
hm
f
or
hi
gh
-
d
i
m
e
ns
i
ona
l
f
un
c
t
i
o
n opt
i
m
i
z
at
i
on.
S
of
t
C
om
pu
t
i
ng
.
20
1
3;
17(
6)
:
995
-
1
004.
[2
]
Leni
n
K
a
r
ab
oga
D
,
G
or
k
e
m
l
i
B
,
O
z
t
ur
k
C
,
et
a
l
.
A
c
om
pr
eh
ens
i
v
e
s
ur
v
ey
:
ar
t
i
f
i
c
i
al
bee
c
o
l
ony
(
A
B
C
)
al
gor
i
t
h
m
an
d ap
pl
i
c
at
i
on
s
.
A
r
t
if
ic
ia
l I
n
t
e
lli
g
e
n
c
e
R
e
v
ie
w
.
201
4;
42(
1
)
:
21
-
57.
[3
]
P
ur
bas
ar
i
A
,
S
uw
ar
di
I
S
,
S
an
t
os
o
O
S
,
et
al
.
D
at
a
P
ar
t
i
t
i
on
a
nd
C
o
m
m
u
ni
c
at
i
on
O
n
P
ar
al
l
el
H
e
u
r
is
t
ik
M
odel
B
as
ed
on C
l
onal
S
el
e
c
t
i
on A
l
g
or
i
t
h
m
.
T
el
k
om
ni
k
a
.
20
15
;
1
3
(1
).
[4
]
Y
abo
Luo
,
e
t
a
l
.
A
n
I
m
pr
ov
e
d
N
S
G
A
I
I
A
l
gor
i
t
hm
f
or
M
ul
t
i
obj
e
c
t
i
v
e
T
r
av
el
i
ng
S
al
e
s
m
an
P
r
obl
e
m
.
T
E
LK
O
M
N
I
K
A
I
ndone
s
i
a
n J
ou
r
nal
o
f
E
l
e
c
t
r
i
c
al
E
ngi
neer
i
ng
.
2014;
12
(
6
):
44
13
-
44
18.
[5
]
M
ans
our
i
P,
A
s
ady
B,
G
upt
a
N.
T
he
bi
s
ec
t
i
on
-
ar
t
i
f
i
c
i
al
b
e
e
c
o
l
ony
al
gor
i
t
hm
t
o
s
ol
v
e
f
i
x
ed
poi
nt
pr
obl
e
m
s
.
A
pp
l
i
e
d S
of
t
C
om
pu
t
i
ng
.
201
5;
26(
1)
:
1
43
-
1
48.
[6
]
I
m
an
i
an
N,
Sh
i
ri
ME
,
M
or
adi
P
.
V
el
oc
i
t
y
bas
ed
ar
t
i
f
i
c
i
al
b
e
e
c
ol
ony
al
g
or
i
t
hm
f
or
hi
g
h
di
m
en
s
i
on
al
c
ont
i
nuo
us
opt
i
m
i
z
at
i
on pr
obl
em
s
.
E
n
gi
ne
er
i
n
g A
ppl
i
c
at
i
on
s
of
A
r
t
i
f
i
c
i
al
I
n
t
el
l
i
ge
nc
e
.
20
14;
36(
11)
:
148
-
16
3.
[7
]
W
ang
H
,
W
u
Z,
R
ah
nanay
a
n
S
. M
u
l
ti
-
s
t
r
at
e
gy
en
s
em
bl
e
ar
t
i
f
i
c
i
al
bee
c
o
l
ony
a
l
gor
i
t
h
m
.
I
nf
o
r
m
a
t
i
on
S
c
i
e
nc
es
.
2
014;
279(
9)
:
5
87
-
6
03.
[8
]
Ki
ra
n
M
S
,
Ha
k
l
i
H,
G
unduz
M
,
et
al
.
A
r
t
i
f
i
c
i
al
bee
c
ol
o
ny
al
gor
i
t
h
m
w
i
t
h
v
ar
i
abl
e
s
ear
c
h
s
t
r
at
egy
f
or
c
ont
i
nuo
us
opt
i
m
i
z
at
i
on.
I
nf
or
m
at
i
on S
c
i
en
c
e
s
.
2
015;
300:
1
40
-
157
.
[9
]
Ki
ra
n
M
S
,
F
i
ndi
k
O
.
A
d
i
r
ec
t
ed
ar
t
i
f
i
c
i
al
b
ee
c
ol
o
ny
al
g
or
i
t
h
m
.
A
ppl
i
ed
S
of
t
C
om
put
i
ng
.
2015
;
26:
4
54
-
462.
[
10]
S
har
m
a
T
K
,
P
ant
M
.
E
nhanc
i
ng
t
he
f
ood
l
oc
at
i
o
ns
i
n
a
n
ar
t
i
f
i
c
i
a
l
bee
c
ol
o
ny
al
gor
i
t
hm
.
S
o
ft
C
om
put
i
ng
.
201
3;
1
7(
10)
:
193
9
-
1965
.
[
11]
M
eng L,
Y
i
n S
L,
H
u X
Y
.
A
N
ew
M
et
hod U
s
e
d f
or
T
r
av
el
i
n
g
S
al
es
m
an
P
r
obl
em
B
as
ed on
D
i
s
c
r
e
t
e
A
r
t
i
f
i
c
i
al
B
ee
C
ol
o
ny
A
l
gor
i
t
h
m
.
T
E
LK
O
M
N
I
K
A
(
T
el
ec
om
m
uni
c
at
i
on,
C
om
p
ut
i
n
g,
E
l
e
c
t
r
oni
c
s
an
d
C
ont
r
ol
)
.
201
6;
14
(
1
):
342
-
34
8.
[
12]
M
eng L,
Y
i
n S
L.
A
n i
m
pr
ov
ed
M
am
dani
F
uz
z
y
N
eur
al
N
et
w
or
k
s
B
a
s
ed
on P
S
O
A
l
gor
i
t
hm
and N
ew
P
ar
am
e
t
er
O
pt
i
m
i
z
at
i
on
.
T
E
L
K
O
M
N
I
K
A
I
ndone
s
i
an
J
our
n
a
l
of
E
l
e
c
t
r
i
c
al
E
ng
i
neer
i
ng
.
2
016;
17
(
1
):
201
-
20
6.
[
13]
Yi
n
S
L
,
L
i
u
T
H
,
Li
H
.
A
ppl
i
c
at
i
on
of
K
al
m
an
f
i
l
t
er
i
n
g
i
n
i
nd
oo
r
l
oc
at
i
on
ba
s
ed
o
n
s
i
m
ul
at
ed annea
l
i
n
g
al
gor
i
t
h
m
.
S
he
ny
a
ng N
or
m
al
U
ni
v
er
s
i
t
y
N
at
ur
al
S
c
i
e
nc
e
.
20
15;
33
(
1
)
:
86
-
90.
Evaluation Warning : The document was created with Spire.PDF for Python.