Learn R Programming

ftnonpar (version 0.1-88)

pmspec: Piecewise monotone spectral density approximation with taut strings

Description

Applies the taut string method to spectral densities.

Usage

pmspec(x, pks=0, alpha=0.9, sqzf=0.9, mult=0, lcl=0, ln=0, fig = 0, pow=10^-2)

Arguments

x
data
pks
if pks=0 then the number of peaks is determined automatically. If pks = k >0 then a density with n paeks is returned pks
alpha
If pks=0 then on Gaussian test beds the number of peaks of the true density is at least the returned value with an asymptotic probability of at least alpha. If pks > 0 then the parameter is not operational alpha
sqzf
Squeeze factor for the taut string sqzf
mult
If mult = T then rescaled empirical density is compared with the exponential distribution only on intervals forming a multiresolution scheme. If T = F then all intervals are used. If the sample size of the data is less than 512 then T = F is default mult
lcl
If lcl = T then string is calculated using local squeezing. If lcl = F then global squeezing is used.lcl
fig
If fig = T then the data are automticall plotted fig
ln
If fig = T and ln = T then the densities are plotted on a log scale. If ln = F then no transformation is performed ln
pow
Peaks with a power less than pow * total power will be ignored

Value

edf
Emprical spectral density
df
String density
pks
Number of peaks
ll
Lower boundary for string
uu
Upper boundary for string
str
String

References

Davies, P.L. snd Kovac, A. (2001) Local extremes, runs, strings and multiresolution, Annals of Statistics 29 (1) 1-65. Davies, P. L. and Kovac, A. (2002) Densities, Spectral Densities and Modality

See Also

pmreg,l1pmreg,pmden