mfft_real.Rd

% Generated by roxygen2: do not edit by hand
% Please edit documentation in R/mfft_real.R
\name{mfft_real}
\alias{mfft_real}
\title{Frequency Modified Fourier transform  for real series}
\usage{
mfft_real(
  xdata,
  nfreq = 5,
  minfreq = NULL,
  maxfreq = NULL,
  correction = 1,
  fast = TRUE
)
}
\arguments{
\item{xdata}{The data provided either as a time series (advised), or as a vector.}

\item{nfreq}{is the number of frequencies returned, must be smaller that the length of  xdata.}

\item{minfreq, maxfreq}{If provided, bracket the frequencies to be probed. Note this are
more precisely angular velocities (2\pi / period), expressed in time-inverse units
with the time resolution encoded in `xdata` if the latter is a time series.}

\item{fast}{(default = TRUE) uses analytical formulations for the crossproducts involving sines and cosines. 
note: this is not really faster because the bottleneck is actually the goden section search. But more elegant.}

\item{correction:}{0: no frequency correction (equivalent to Laskar); 1 : frequency correction using linear approximation ; 2: frequency correction using sythetic data; 
3: second order-correction using synthetic data (all documented in the Sidlichovsky and Nesvorny reference)}
}
\value{
a `mfft_deco` object, based on a data.frame with columns "Freq", "Ampl" and "Phases".
}
\description{
R-coded version of the Modified Fourier Transform
with frequency  correction, adapted to R. 
much slower than mfft (for complex numbers) as the latter is
mainly written in C, but is physically
more interpretable if signal is real, because
it is designed to have no imaginary part in the residual
A C-version should be supplied one day.
}
\examples{

data(harmonic_sample)
spectrum <- mfft_real(harmonic_sample$data)
print(spectrum)

}
\references{
\insertRef{sidlichovsky97aa}{gtseries}
}
\author{
Michel Crucifix
}