pkdecomp {diffractometry}R Documentation

Decomposition of peaks for the whole data set

Description

Calculates decompositions of peaks for the whole diffractogram

Usage

pkdecomp(baslfit,intnum=0, alpha=0.1, maxiter1=500, maxiter=10000, hmax=5, 
maxsolutions=3,heterosk=TRUE,baselim=c(0.05,5),dispers=1)

Arguments

baslfit Output of baselinefit
intnum Vector of numbers of intervals. If intnum = 0, all intervals are used
alpha Test level for residual criterion
maxiter1 Number of attempts to fit a model with 1 component
maxiter Number of attempts to fit a model with k > 1 components
hmax Maximum number of components
maxsolutions Number of solutions with k components
heterosk If TRUE, the estimate of noise level given in baslfit is used (default); otherwise noise level is taken to be proportional to signal height
baselim Limits for changes in the baseline estimate; first component is given in percent of the baseline height, second in counts/2theta
dispers Additional dispersion factor; not used if heterosk==T

Details

Calls pkdecompint to decompose the peaks found by baselinefit into Pearson Type VII kernels. For every interval, first one kernel is tried. The number of kernels is increased until either a solution accepted by the residual criterion is found or the maximum number of kernels hmax is reached. After a solution is accepted, for maxsolutions greater than 1, further decompositions with the same number of kernels can be produced.

Value

A vector of lists as given by pkdecompint

Author(s)

T. Mildenberger; Algorithm for residual criterion by T. Bernholt and T. Hofmeister

References

P.L. Davies, U. Gather, M. Meise, D. Mergel, T. Mildenberger (2008): "Residual based localization and quantification of peaks in x-ray diffractograms", to appear in the Annals of Applied Statistics. http://arxiv.org/abs/0711.3687

T. Bernholt and T. Hofmeister (2006): "An algorithm for a generalized maximum subsequence problem", in: J. Correa, A. Hevia, M. Kiwi (editors), "Latin 2006: Theoretical Informatics", volume 3887 of Lecture notes in Computer Science, pages 178-189, Berlin, Heidelberg. Springer Verlag

See Also

diffractogram, baselinefit, pkdecompint

Examples

## Decomposition of data in peak interval into two components

par(mfrow=c(2,1))

data(indiumoxide)
indox<-indiumoxide[1901:2400,]
base<-baselinefit(indox)

ind<-c(base$indlsep[1],base$indrsep[1])

plot(indox[ind[1]:ind[2],1], base$baseline$peaks[ind[1]:ind[2]],xlab="",ylab="")

pks<-pkdecomp(base,intnum=1,maxsolutions=1)

lines(indox[ind[1]:ind[2],1],pks[[2]]$fit,col="red")
plot(indox[ind[1]:ind[2],1],pks[[2]]$fitpk[1,],ylim=c(0,1800),type="l",xlab="",ylab="")
lines(indox[ind[1]:ind[2],1],pks[[2]]$fitpk[2,])


[Package diffractometry version 0.1-00 Index]