Evolutionary Computation,
Год журнала:
2011,
Номер
20(1), С. 63 - 89
Опубликована: Май 24, 2011
The
literature
shows
that
one-,
two-,
and
three-dimensional
bin
packing
knapsack
are
difficult
problems
in
operational
research.
Many
techniques,
including
exact,
heuristic,
metaheuristic
approaches,
have
been
investigated
to
solve
these
it
is
often
not
clear
which
method
use
when
presented
with
a
new
instance.
This
paper
presents
an
approach
motivated
by
the
goal
of
building
computer
systems
can
design
heuristic
methods.
overall
aim
explore
possibilities
for
automating
process.
We
present
genetic
programming
system
automatically
generate
good
quality
each
It
necessary
change
methodology
depending
on
problem
type
(one-,
or
problems),
therefore
has
level
generality
unmatched
other
literature.
carry
out
extensive
suite
experiments
compare
best
human
designed
heuristics
Note
our
uses
same
parameters
all
experiments.
contribution
this
more
general
than
those
currently
available,
show
that,
using
methodology,
possible
competitive
from
represents
first
algorithm
able
claim
results
such
wide
variety
domains.
Server
consolidation
using
virtualization
technology
has
become
increasingly
important
for
improving
data
center
efficiency.
It
enables
one
physical
server
to
host
multiple
independent
virtual
machines
(VMs),
and
the
transparent
movement
of
workloads
from
another.
Fine-grained
machine
resource
allocation
reallocation
are
possible
in
order
meet
performance
targets
applications
running
on
machines.
On
other
hand,
these
capabilities
create
demands
system
management,
especially
large-scale
centers.
In
this
paper,
a
two-level
control
is
proposed
manage
mappings
VMs
resources.
The
focus
VM
placement
problem
which
posed
as
multi-objective
optimization
simultaneously
minimizing
total
wastage,
power
consumption
thermal
dissipation
costs.
An
improved
genetic
algorithm
with
fuzzy
evaluation
efficiently
searching
large
solution
space
conveniently
combining
possibly
conflicting
objectives.
simulation-based
power-consumption
thermal-dissipation
models
based
profiling
Blade
Center,
demonstrates
good
performance,
scalability
robustness
our
approach.
Compared
four
well-known
bin-packing
algorithms
two
single-objective
approaches,
solutions
obtained
approach
seek
balance
among
objectives
while
others
cannot.
Evolutionary Computation,
Год журнала:
1994,
Номер
2(2), С. 123 - 144
Опубликована: Июнь 1, 1994
An
important
class
of
computational
problems
are
grouping
problems,
where
the
aim
is
to
group
together
members
a
set
(i.e.,
find
good
partition
set).
We
show
why
both
standard
and
ordering
GAs
fare
poorly
in
this
domain
by
pointing
out
their
inherent
difficulty
capture
regularities
functional
landscape
problems.
then
propose
new
encoding
scheme
genetic
operators
adapted
these
yielding
Grouping
Genetic
Algorithm
(GGA).
give
an
experimental
comparison
GGA
with
other
applied
we
illustrate
approach
two
more
examples
successfully
treated
GGA:
Bin
Packing
Economies
Scale.
