cpt, cp.molecule, dwt, nd.dwt, rob.dwt, rob.dwt.list, wpt, wp.molecule
.
decompose(x, top=NULL, data=NULL, order.crystal="energy")
cpt, cp.molecule, dwt, nd.dwt, rob.dwt,rob.dwt.list, wpt, wp.molecule
.
x
used in the decomposition,
in effect only if
x
is an object of class
"molecule"
.
By default, use all the coefficients in
x
. See details below.
x
is an object of class
"molecule"
.
See details below.
"dwt"
(for the DWT decomposition only),
"energy"
(the decomposition components will be
ordered by energy), and
"frequency"
(in
depth.first
order).
decompose
. See
decompose.object
for details.
The function
decompose
produces a
decomposition which is a sum of components where each k-th component
represents the contribution of the k-th crystal (when
x
is of class
cpt
) or the atom
associated with the k-th largest coefficient (when
x
is of class
cp.molecule
or
wp.molecule
to the signal.
When
x
is of class
"cpt"
or
"wpt"
, all atoms with non-zero coefficients
are included in the decomposition, the sum of the components yields the original
series.
When
x
is of class
"cp.molecule"
or
"wp.molecule"
, if
data
is provided,
the sum of the components yields an approximation of
the
data
with a (non-trivial) residual term attached at the end of the
outcome matrix. The bigger the
top
, the closer the approximation to the
original signal; if
data
is missing, the sum of the components only yields
an approximation to the original signal.
xx <- make.signal("ramp", 1024) par(mfrow=c(2,2)) plot(xx, type="l", xlab="") xx.dwt <- dwt(xx, wavelet="s8", n.levels=4) plot(xx.dwt) xx.dec <- decompose(xx.dwt) plot(xx.dec) xx.top <- top.atoms(xx.dwt, n.atoms=10) xx.dec1 <- decompose(xx.top, data=xx) # decompose signal on the top 10 atoms plot(xx.dec1)