Curve Fit Details
Curve Fit is applied to a single 2D spectral data object. The user may fit peaks with desired shape to match the overall shape of overlapping peaks in the original spectrum. In a calculation step the peak area of the ideal peaks is calculated showing the potential real area of overlapping peaks. The User can manually set ideal peaks and can configure their shapes utilizing Lorentzian, Gaussian or Voigt profiles. Optionally the software calculates the optimal full width half maximum value which might bet fit with the evaluation algorithm. Dependent on the spectrum habit, applying a baseline correction is needed too.
By setting the desired values for the curve, a fitted spectrum of a special peak will be created. This can be done multiple times, preferably for a spectrum range with overlapping signals or the whole spectrum.
After setting all favored peaks the sum of all individual peaks can be displayed to see the resulting overlay compared to the original spectral data object.
The estimated peaks of the curves are shown in a peak table. Of the idealized spectrum, the values of the peak area can be used for further calculations. The peak table shows the idealized peaks and their values, not the actual of the spectrum used for fitting.
Curve Fit parameters
The following options Auto Fit Algorithm is used:
-
Levenberg-Marquardt
The Levenberg-Marquardt algorithm is a well-known numerical optimizing algorithm being used to optimize the width and height of a set of peaks to match the shape of a curve as good as possible. Basically it tries to minimize the error between the spectral original area and the “artificial” area resulting from the different peaks.
An optimal full width at half maximum (FWHM) value is calculated for each peak by setting the Use estimated FWHM flag. In case this flag is set, FWHM starting values for all peaks are identical. Manual adaption is possible.
Several baseline options are available and need to be adapted manually dependent on the overall spectrum baseline shape.
Result and residual curve are calculated too and can be displayed in overlay mode together with the raw data and the estimated peaks.
Peak parameters
Each estimated peak can be optimized interactively by changing the individual parameters.
The following options are available for manipulation:
-
Full Width Half Maximum
Sets the width / shape of the peak. A new value can be entered manually or changed interactively pressing the vertical marker
in the <value> field. -
Maximum X Position
Sets the peak position of the peak. A new value can be entered manually or graphically pressing the vertical marker
in the <value> field. -
Maximum Y Value
Sets the peak height. A new value can be entered manually or changed interactively pressing the vertical marker in the <value> field.
The parameter Response Curve Shape can be adapted manually too. The different options are available from a list box. Each algorithm affects the shape of the curve slightly different:
-
Gauss
The formula of the Gaussian normal distribution function is given below:
-
Cauchy-Lorentz
The formula of the Cauchy-Lorentz normal distribution function is given below:
-
Voigt
The Voigt profile is a linear combination of Gauss and Cauchy-Lorentz.
In order to optimize the look and feel it is possible to define a color scheme for an estimated trace too.