EDS spectrum analysis tools within the NeXL toolset
Author usnistgov
5 Stars
Updated Last
1 Year Ago
Started In
September 2019


Microanalytical X-ray Spectrum Analysis

Documentation Build Status


Install NeXLSpectrum using the Julia package manager

julia> ]add NeXLSpectrum


julia> using Pkg
julia> Pkg.add("NeXLSpectrum")


NeXLSpectrum is a library of tools for manipulating EDS spectrum within the NeXL framework. NeXLSpectrum depends on NeXLUncertainties, NeXLCore and NeXLMatrixCorrection and loading NeXLSpectrum will also make these libraries available.

In addition, NeXLSpectrum makes extensive use of the third-party libraries DataFrames for data tables and Gadfly for plotting. You will also want to makes these libraries available.

julia> ]add DataFrames, Gadfly

Primarily, NeXLSpectrum

  • Implements the Spectrum type to represent individual EDS spectra

    • Reads Spectrum objects from disk files (or other streams) in EMSA, Bruker and ASPEX formats
    • Writes Spectrum objects to a disk file in EMSA format
  • Provides utilities and other low level tools to interogate and manipulate Spectrum objects

  • Implements the HyperSpectrum type to represent hyper-spectra (linescan, image, cube, ...)

    • The individual pixels in a hyper-spectrum are visible as Spectrum objects
    • Reads HyperSpectrum objects from LISPIX-style RPL/RAW files
    • Writes HyperSpectrum objects to RPL/RAW files
  • Provides data types to define detector properties

  • Extends Gadfly.jl to plot spectra and spectrum-related items

  • Provides algorithms to perform Schamber-style filter-fitting of spectra

    • Implements a basic weighted LLSQ fit algorithm
      • Fits characteristic, escape, Compton and other features
    • Implements a 'vector-based' quick-quant algorithm for processing hyper-spectra

NeXL and PackageCompiler

Because Julia uses a just-in-time compiler that recompiles libraries each time Julia is restarted and the NeXL libraries are quite large, the first time you perform an operation it can be quite slow. Subsequent calls are much faster. Fortunately, there is a way around this. The PackageCompiler library can be used to build a version of Julia in which NeXLSpectrum and the libraries on which it depends are build into the language much like the base libraries. I recommend that you use the PackageCompiler to rebuild Julia each time you update the packages (using julia> ]up) or upgrade to a new version of Julia.

In Windows, this operation looks like:

julia> using NeXLSpectrum, DataFrames, Gadfly # Load the basic libraries
julia> using PackageCompiler 
julia> PackageCompiler.create_sysimage(
    [ "Gadfly", "DataFrames", "BoteSalvatICX", "NeXLUncertainties", "NeXLCore", "NeXLMatrixCorrection", "NeXLSpectrum" ]; 
    sysimage_path=joinpath(homedir(), ".julia", "NeXLSysimage.dll"),
    precompile_execution_file=joinpath(pkgdir(NeXLSpectrum),"scripts", "precompile.jl"))

This operation may take tens-of-minutes to complete but will be worth the effort in terms of saved time later on.

In other operating systems, you will want to change the extension on the filename in the sysimage_path.

This command creates a sysimage containing the NeXL libraries along with DataFrames and Gadfly. You will now have to tell Julia to use this sysimage. I usually create multiple links to Julia that both use the default sysimage and the NeXLSysimage. The sysimage is selected using the -J command line option. In Windows, this looks like:

## To use the sysimage, start Julia with the `-J` option
# > julia -J"C:\Users\username\.julia\NeXLSysimage.dll"

You can edit the properties of the link used to start Julia to include the -J option. While you are at it, you might also want to add the -t option to start Julia with multiple threads. For example,

## To use the sysimage, start Julia with the `-J` option
# > julia -J"C:\Users\username\.julia\NeXLSysimage.dll" -t 4

will start Julia with the sysimage and 4 threads. Usually, selecting the number of threads equal to the number of cores (not HyperThreads) will produce the optimal performance.