PeakLab v1 Documentation Contents            AIST Software Home            AIST Software Support

Generalized Voigt

Voigt Generalization Convolution Models

PeakLab offers two different models for testing the integrity of Voigt model fits to spectroscopic data:

Gaussian Ä Student's t (Area)

This is an convolution model used to check the validity of Voigt model fits.

a0 = Area

a1 = Center

a2 = Gaussian Width

a3 = Lorentzian Width

a4 = Student's t nu (1=Lorentzian, Infinite=Gaussian)

Built in model: Gauss<S>

User-defined peaks and view functions: Gauss[S]i[amp](x,a0,a1,a2,a3) (Warning: computed as integral, very slow!)

This is a symmetric convolution of a Gaussian and a Student's t which can be used to fit Voigt peaks where the Lorentzian component is estimated with a Student's t that will be a Lorentzian only with the a4 nu=1. If the tails of the non-Gaussian component of the convolution are not perfectly Lorentzian, this model will fit to a value other an a3=1. A Lorentzian has wide tails which may be subject to instrumental sampling, digitization, and filtering. The Lorentzian component of a pure Voigt should fit very close to a4=1.0 if there is a significant Lorentzian component to the Voigt. The peaks below vary from a4=1 to a4=1.5. Lorentzian Ä Student's t (Area)

a0 = Area

a1 = Center

a2 = Lorentzian Width

a3 = Gaussian Width

a4 = Student's t nu (1=Lorentzian, Infinite=Gaussian)

Built in model:Lorentz<S>

User-defined peaks and view functions: Lorentz[S]i(x,a0,a1,a2,a3) (Warning: computed as integral, very slow!)

This is a symmetric convolution of a Lorentzian and a Student's t which can be used to fit Voigt data where the Gaussian component is estimated with a Student's t that is a Gaussian only with nu at infinity (PeakLab's upper bound on nu is 1,000,000). If the tails are not perfectly Gaussian, this model will fit to a value with a nu lower than this maximum. This model can be used to test the integrity of the Gaussian component of a Voigt peak. If any portion of that which contributes to the Gaussian line spread function has wider tails, this will be strongly reflected in a4. If a pure Voigt is fit, you should expect values of a4 10,000 or higher, if there is a significant Gaussian component to the Voigt. The peaks below vary from a4=5 to a4=1,000,000.    