adeconv: Absorption band deconvolution

Description Usage Arguments Details Value References See Also

View source: R/deconv.R

Description

This function deconvolves a measured signal into a sum of model functions.

Usage

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
adeconv(
  spec,
  sr,
  start,
  lower,
  upper,
  ef,
  model = c("Gauss", "Lorentz", "Voigt"),
  maxit = 50000,
  reltol = 1e-12
)

Arguments

spec

A vector with the spectra to be deconvolved.

sr

A vector with the spectral reference of the spectra.

start

A matrix with first guess parameters. See details.

lower

A matrix with the lower limit of the parameters to be fitted. See details.

upper

A matrix with the upper limit of the parameters to be fitted. See details.

ef

Error function to be minimized.

model

One of 'Gauss', 'Lorentz' or 'Voigt'. See details.

maxit

Maximum number of iterations.

reltol

Relative tolerance for convergence criteria.

Details

This function is a wrapper for the optim function, with the Nelder-Mead method, for deconvolution of a spectrum as sum of Gaussian, Lorentzian or Voigt curves. It is appropriate for deconvolution of absorption peaks if the peaks are present in or just beyond the limits of the measured spectrum. When applied for deconvolution of pigment absorption, it is important that other absorption signals have been removed form the spectrum for a meaningful result. The spectral reference scale may be any desired (e.g., energy, wavenumber, wavelength), but the fitting space, start values and sr must be specified in the correct scale.

The arguments start, lower, and upper, must be three column matrices for models 'Gauss' and 'Lorentz', with center, sigma ('Gauss') or hwhm ('Lorentz') and scale, respectively and peaks per row. For the 'Voigt' model, it should have four columns, with center, sigma, hwhm and scale (in this order). The number of absorption bands used depends on the number of rows and the fitting space is completely defined by the lower and upper boundaries (may be infinite). The boundaries are enforced by returning NA on the error function evaluation when optim test parameter values outside the fitting space.

The spectra to be deconvolved may contain NA, but not sr, start, lower, and upper. maxit and reltol are passed to optim. A set of error functions is provided: mard, rmsd, mlrt, and rrsd. For more details on the Gaussian, Lorentzian and Voigt parameters, see apeaks.

Value

The return of optim, with fitted coefficients converted to matrix format for convenience.

References

Hoepffner, N.; Sathyendranath, S. 1991. Effect of pigment composition on absorption properties of phytoplankton. Marine Ecology Progress Series 73, 11-23. DOI: 10.3354/meps073011

See Also

apeaks, mard, rmsd, mlrt, and rrsd


AlexCast/rho documentation built on Dec. 14, 2021, 9:47 a.m.