Was agreed to implement an nth order polynomial option in the first instance & to change the GUI to allow different fits for the sample & can.
Alternative fits for transmissions. Log(Trans) can go linearly with wavelength (simple adsorption) or with square of wavelength (strong SANS) or a product of these, it can then also have a wiggle around 6 or 7 angstrom when the wide angle scatter turns off. At present we fit a straight line to Log(trans). Can we have the options to try quadratic or cubic instead of linear, with a separate switch for the sample and the can. Two drop-downs, top right of “reduction settings” tab? Default to linear as now. (I am concerned that letting say a cubic loose on something that only needs a straight line might produce some odd results.) Not sure how to fit the wiggle! Perhaps just try a smoothing function, but keeping those under control at start and end of range may be tricky. Might need two quadratics with a cross-over function in the middle.
Original Reporter: Gesner Passos
This ticket is blocked by :
Was agreed to implement an nth order polynomial option in the first instance & to change the GUI to allow different fits for the sample & can.
Alternative fits for transmissions. Log(Trans) can go linearly with wavelength (simple adsorption) or with square of wavelength (strong SANS) or a product of these, it can then also have a wiggle around 6 or 7 angstrom when the wide angle scatter turns off. At present we fit a straight line to Log(trans). Can we have the options to try quadratic or cubic instead of linear, with a separate switch for the sample and the can. Two drop-downs, top right of “reduction settings” tab? Default to linear as now. (I am concerned that letting say a cubic loose on something that only needs a straight line might produce some odd results.) Not sure how to fit the wiggle! Perhaps just try a smoothing function, but keeping those under control at start and end of range may be tricky. Might need two quadratics with a cross-over function in the middle.