Create time-of-flight fitting function Ikeda-Carpenter

[Laurent Chapon]

This function gives often a better description than the back-to-back exponential function.

[Nick Draper]

Batch move uncompleted tickes to next iteration

[Nick Draper]

dropped in priority until curve fitting algorithm base class work is complete

[Nick Draper]

Moved as part of iteration end

[Nick Draper]

Paper is here ​http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6TJM-470FCSK-126&_user=910841&_coverDate=09%2F15%2F1985&_rdoc=33&_fmt=high&_orig=browse&_srch=doc-info(%23toc%235314%231985%23997609996%23348692%23FLP%23display%23Volume)&_cdi=5314&_sort=d&_docanchor=&view=c&_ct=53&_acct=C000047841&_version=1&_urlVersion=0&_userid=910841&md5=9f540ce286642937a71c7aeb9a578317

[Nick Draper]

Moved as part of iteration 18 end

[Anders Markvardsen]

(In [2897]) Updated doc in code and wiki to be consistent with camel name notation for parameter names defined for the fitting functions. Refs #330.

[Anders Markvardsen]

(In [3048]) Finished first iteration of Ikeda Carpenter, including wiki and test on mock data.

Remaining:

Implemented for now without derivative, and convergence seems to be a bit fragile in some instances (this is apparently not unexpected from conversation with Aziz) For now assumed that ?\226?\128?\152wavelength = 1?\226?\128?\153, i.e. this e.g. means alpha = 1/(alpha0+alpha1) Assume PV is 100% Guassian, need to add Lorentzian part to PV

To do:

A few more test on stability e.g. to probe if some params are more difficult to refine

Implement derivative, which may make this peakshape function more stable

Implement ?\226?\128?\152wavelength?\226?\128?\153 dependence on parameter. Before doing this make sure that the unit for wavelength in Mantid matches the wavelength unit used in the Fullprof manual Include test on a real data peak Add implementation of exponential integral function that takes complex argument and then add Lorentzian part to PV

Desirable:

Allow user to specify ?\226?\128?\152wavelength?\226?\128?\153. That way this peak shape is not tied to data for which the wavelength at each data point is know

Refs #330.

[Anders Markvardsen]

(In [3107]) More testing on simplex Ikeda-Carpenter, and simplex seems to perform better than expected. Implemented derivative code for some of the parameters and somewhat surprisingly simplex-fitting seems to perform better than least-squared-derivative fitting (at least when fitting the I and X0 parameters). Hence for now commented out derivative code.

In this process made some minor improvements in speed to how the other fitting functions calculate derivatives and corrected a small error in the Lorentzian derivative function. Refs #330.

[Anders Markvardsen]

(In [3123]) Modified Fit1D and IkedaCarpenter to allow for fitting parameters to depend on wavelength. Refs #330.

[Anders Markvardsen]

(In [3142]) Modified IkedaCarpenter so workspace plot output also have some dependence on wavelength. Forgot in previous commit to this ticket to add wavelength dependence on Alpha1 parameter also, has now done this. Refs #330.

[Anders Markvardsen]

(In [3173]) As a consequence of the wavelength dependence introduced with IkedaCarpender fitting removed method with signature double FunctionCurve.obj(...). This also reduce dublication. Refs #330.

[Anders Markvardsen]

(In [3180]) Ups my mistake, when fitting a fit in MantidPlot fitted peak wrong way up. Fixed this. Refs #330.

[Nick Draper]

Moved as part of the end of Iteration 19

[Anders Markvardsen]

(In [3232]) Implemented complex Exponential Integral function based on expressions in Abramowitz and Stegun. Tested for a number of point in complex plane. Seem to perform pretty well. Refs #330.

[Anders Markvardsen]

(In [3233]) Removed a few warning signed to unsigned int. Refs #330.

[Anders Markvardsen]

(In [3237]) Added Lorentzian broadening to IC. Refs #330.

[Anders Markvardsen]

(In [3241]) Added test which refine on background, intensity, gaussian and Lorentzian peak-widths simultaniously, while keeping the IC instrument parameters fixed. Refs #330.

[Anders Markvardsen]

Wiki doc updated

[Stuart Campbell]

This ticket has been transferred to github issue ​1178