<s>
Multi-swarm	B-Algorithm
optimization	I-Algorithm
is	O
a	O
variant	O
of	O
particle	B-Algorithm
swarm	I-Algorithm
optimization	I-Algorithm
(	O
PSO	O
)	O
based	O
on	O
the	O
use	O
of	O
multiple	O
sub-swarms	O
instead	O
of	O
one	O
(	O
standard	O
)	O
swarm	O
.	O
</s>
<s>
The	O
general	O
approach	O
in	O
multi-swarm	B-Algorithm
optimization	I-Algorithm
is	O
that	O
each	O
sub-swarm	O
focuses	O
on	O
a	O
specific	O
region	O
while	O
a	O
specific	O
diversification	O
method	O
decides	O
where	O
and	O
when	O
to	O
launch	O
the	O
sub-swarms	O
.	O
</s>
<s>
In	O
general	O
,	O
the	O
development	O
of	O
multi-swarm	O
systems	O
leads	O
to	O
design	O
decisions	O
which	O
did	O
not	O
exist	O
during	O
the	O
original	O
development	O
of	O
particle	B-Algorithm
swarm	I-Algorithm
optimization	I-Algorithm
,	O
such	O
as	O
the	O
number	O
of	O
particles	O
to	O
use	O
in	O
each	O
sub-swarm	O
,	O
the	O
optimal	O
value	O
for	O
the	O
constriction	O
factor	O
and	O
the	O
effects	O
of	O
non-random	O
initial	O
positions	O
and	O
initial	O
velocities	O
.	O
</s>
<s>
Multi-swarm	O
systems	O
thus	O
provide	O
a	O
useful	O
framework	O
for	O
the	O
development	O
of	O
hybrid	B-Algorithm
algorithms	I-Algorithm
.	O
</s>
<s>
For	O
example	O
,	O
the	O
UMDA-PSO	O
multi-swarm	O
system	O
effectively	O
combines	O
components	O
from	O
particle	B-Algorithm
swarm	I-Algorithm
optimization	I-Algorithm
,	O
estimation	O
of	O
distribution	O
algorithm	O
,	O
and	O
differential	B-Algorithm
evolution	I-Algorithm
into	O
a	O
multi-swarm	O
hybrid	O
.	O
</s>
<s>
A	O
on	O
Mendeley	B-Language
is	O
available	O
to	O
all	O
interested	O
researchers	O
.	O
</s>
