modifyfv prefilters measured data by the inverse of a known part of the transfer function.
Syntax
Fdatm = modifyfv(pdat,Fdat)
[Fdatm,vdatm] = modifyfv(pdat,Fdat,vdat,plotmode)
Description
If a multiplicative term of the transfer function is known, it is disadvantageous to estimate this part from the measured data, since this increases the variance of the other term, too. A possible solution is to prefilter the input data by the inverse of this term.
The parameter vector is given in pdat (a string if the filename is given).
Fdat contains the Fourier data: it is an array: [freqv,x,y], or a Fourier vector (see expfou), or the name of the Fourier file.
The filtered Fourier data are given in Fdatm, as an array if Fdat is an array, or as a Fourier vector if Fdat is a vector or a filename.
If the known partial transfer function has significant dynamics in the band of interest, the variance data also have to be changed, since the modification preserves the relative variance of the Fourier amplitudes, and in ELiS the absolute variances are given.
vdat is a variance array, [varx,vary] or [varx,vary,covxy]; or a variance vector (see expvar); or the name of the variance file. The filtered variance data are given in vdatm, as an array if vdat is an array, or as a variance vector if vdat is a vector or a filename.
plotmode defines the form of the plots of the results:
'lin', 'linpb' stands for linear frequency scale,
'log', 'logpb' stands for logarithmic frequency scale,
where pb requests plot in the passband only (the points where the transfer function is zero will be excluded from the plot).
If plotmode has any other value, no plot will be shown.
Default Argument Values
plotmode = 'linpb'
Examples
Fdatm = modifyfv('inpchanz.pbn','inpchan.fbn');
Algorithm
The output Fourier amplitudes are divided by the transfer function values of the known term, and output variances by their absolute square values; the covariances are divided again by the transfer function values.
[ Previous | Help Desk | Next ]