Estimating correlation length of short range order using box-car r-dependent fitting

Modeling 100K Cu(Ir0.88Cr0.12)2S4 using fd-3m settings 

Cu(Ir0.88Cr0.12)2S4 information:

Space group:			Fd-3m
lattice a:				9.80 Angstroms
	
Fractional coordinates:			Cu	(0,0,0)
					Ir 	(0.625,0.625,0.625)		occ. 0.88
					Cr	(0.625,0.625,0.625)		occ. 0.12
					S	(0.385,0.385,0.385)

Uiso: all				0.01
Qmax:					27.0 Inv. Angstroms
Qdamp:				initially set to 0.045
Qbroad :			initially set to 0.01

Data: X-ray data for Cu(Ir0.88Cr0.12)2S4 at 100K obtained at 11IDC at APS 

GOAL: Familiarize with r-dependent fitting macro in PDFgui

TASKS:
1) set up cubic fit for Cu(Ir0.88Cr0.12)2S4 using Fd-3m. You can preload project file to save time.
2) Set delta parameters to zero and fix. Refine cubic model in 20-40 Angstrom range and refine, including Qdamp and Qbroad. Plot fit.
3) Upon successful fitting adopt values of refined parameters and set them as starting
4) Fix all parameters. change refinement range to 1.5 to 40, let delta1 be the only refinable parameter and refine. 
5) Plot fit. You should clearly see that the model works for high r, but not for low r (just below 10 Angstroms it becomes bad). 
This is because there are local structural footprints of short range ordered electronic texture in the system.
6) You can plot crw as a function of r and see how fit works at different r.
7) Now set delta parameter(s) to zero and fix, and release only structural parameters.
8) highlight the fit tree, select Macros from Fits drop down window, then select r-series
9) window will open up offering you to set fit maximum and fit minimum in format first-last-step
10) for fit maximum type in 5 as first, 25 as last, 1 as step
11) for fit minimum type in 1 as first, 21 as last, 1 as step
12) Hit OK. You will only be able to do it if your fit tree (seed for sequential refinement) is selected.
13) Macro will expand in the fit pane with series of linked fits, each having refinement range 4 Angstroms wide,
and they will be shifted by 1 Angstrrom when you go from one fit tree to the next one. 
14) Now you will select all fit trees in the sequence on the fit pane, and execute all fits at once. They will progress 
sequentially in a box-car fashion.
15) You can plot rw (or any other parameter) as a function of index, and you will see how given parameter/quantity 
evolves as the fit progresses from one link to another. Index simply counts the box-car numbers starting from the first link
16) How does rw evolve with the position of the fitting range? What is your estimate for the correlation length of local disorder?

***Exercise is initiated in 08-cics-fd3m-rdep.ddp and worked out in 08-cics-fd3m-rdep-02.ddp project file