R: Full pattern summation using linear regression

fps_lm {powdR}

R Documentation

Full pattern summation using linear regression

Description

fps_lm returns a simple fit of a given pattern using linear regression, where coefficients may be either positive or negative. Does not return quantitative data. For quantitative results use fps or afps.

Usage

fps_lm(
  lib,
  smpl,
  harmonise,
  refs,
  std,
  tth_align,
  align,
  manual_align,
  tth_fps,
  shift,
  p,
  ...
)

Arguments

`lib`	A `powdRlib` object representing the reference library. Created using the `powdRlib` constructor function.
`smpl`	A data frame. First column is 2theta, second column is counts
`harmonise`	logical parameter defining whether to harmonise the `lib` and `smpl`. Default = `TRUE`. When `TRUE` the function harmonises the `lib` and `smpl` data to the intersecting 2theta range at the coarsest resolution available using natural splines.
`refs`	A character string of reference pattern IDs or names from the specified library. The IDs or names supplied must be present within the `lib$phases$phase_id` or `lib$phases$phase_name` columns. If missing from the function call then all phases in the reference library will be used.
`std`	The phase ID (e.g. "QUA.1") to be used as internal standard. Must match an ID provided in the `refs` parameter.
`tth_align`	A vector defining the minimum and maximum 2theta values to be used during alignment (e.g. `c(5,65)`). If not defined, then the full range is used.
`align`	The maximum shift that is allowed during initial 2theta alignment (degrees). Default = 0.1.
`manual_align`	A logical operator denoting whether to optimise the alignment within the negative/position 2theta range defined in the `align` argument, or to use the specified value of the `align` argument for alignment of the sample to the standards. Default = `FALSE`, i.e. alignment is optimised.
`tth_fps`	A vector defining the minimum and maximum 2theta values to be used during full pattern summation (e.g. `c(5,65)`). If not defined, then the full range is used.
`shift`	A single numeric value denoting the maximum (positive or negative) shift, in degrees 2theta, that is allowed during the shifting of selected phases. Default = 0.
`p`	a numeric parameter between 0 and 1 specifying the p-value limit for coefficients. Any reference patterns with a p-value greater than this value will be omitted from the linear regression and results recomputed. Must be greater than 0.000001 but no greater than 1.
`...`	Other arguments

Details

Requires a powdRlib library of reference patterns. Mineral concentrations are not quantified and therefore reference intensity ratios are not required.

Value

a powdRlm object with components:

`tth`	a vector of the 2theta scale of the fitted data
`fitted`	a vector of the fitted XRPD pattern
`measured`	a vector of the original XRPD measurement (aligned)
`residuals`	a vector of the residuals (fitted vs measured)
`phases`	a dataframe of the phases used to produce the fitted pattern
`phases_grouped`	the phases dataframe grouped by phase_name and summed
`weighted_pure_patterns`	a dataframe of reference patterns used to produce the fitted pattern. All patterns have been weighted according to the coefficients used in the fit
`coefficients`	a named vector of coefficients used to produce the fitted pattern
`inputs`	a list of input arguments used in the function call

Examples

data(rockjock)
data(rockjock_mixtures)

#Compute the PCA and loadings
x1 <- xrpd_pca(rockjock_mixtures,
               mean_center = TRUE,
               bin_size = 1,
               root_transform = 1)

## Not run: 
fps_lm_out <- fps_lm(rockjock,
                     smpl = data.frame("x" = x1$loadings$tth,
                                       "y" = x1$loadings$Dim.1),
                     refs = rockjock$phases$phase_id,
                     std = "QUARTZ",
                     align = 0.3,
                     p = 0.01)

plot(fps_lm_out,
     wavelength = "Cu",
     interactive = TRUE,
     group = TRUE)


## End(Not run)

[Package powdR version 1.3.0 Index]