Trace:

# Differences

This shows you the differences between two versions of the page.

y [2009/08/06 15:41] clare created |
y [2009/08/27 12:22] clare |
||
---|---|---|---|

Line 1: | Line 1: | ||

====== y ====== | ====== y ====== | ||

+ | **[//yobs_eqn// !N E]** | ||

+ | |||

+ | The keyword //yobs_eqn// is used in place of the //xdd// keyword. It describes the observed data as an equation. This is very useful for approximating functions, see examples in section **Error! Reference source not found.**. The name given to the equation !N is used for identifying the equation in the GUI. | ||

+ | |||

+ | **[//yobs_out// $file] [//ycalc_out// $file] [//diff_out// $file]** | ||

+ | |||

+ | Outputs the observed, calculated and difference patterns respectively in a format as specified in the extension of $file. Extensions *.xy, *.scr, *.xdd are supported. Further formats are accommodated using any number of macros including Out_Yobs_Ycalc_and_Difference, Out_X_Yobs etc… | ||

+ | |||

+ | **[****//yobs_to_xo_posn_yobs//** **!E]** | ||

+ | |||

+ | At the start of refinement //yobs_to_xo_posn_yobs// decomposes an X-ray diffraction pattern into a new pattern comprising at most one data point per hkl. Fitting to the decomposed pattern in a normal Rietveld refinement manner is then possible due to the ability to refine data of unequal x-axis step sizes. This normal Rietveld manner of fitting is important in structure solution from simulated annealing as the back ground can still be refined and the problem of peak overlap avoided. These new data points are not extracted intensities and thus the problem of peak overlap, as occurs in intensity extraction, is avoided. The much smaller number of data points in the new diffraction pattern can greatly improve speed in structure solution; in other words the calculation time in synthesizing the diffraction pattern becomes close to that of when dealing with single crystal data. | ||

+ | |||

+ | If the distance between two hkls is less than the value of //yobs_to_xo_posn_yobs// then the proposed data point at one of these hkls is discarded. Thus the final decomposed pattern may in fact have less data points than hkls. A reasonable value for //yobs_to_xo_posn_yobs// is Peak_Calculation_Step, or, | ||

+ | |||

+ | yobs_to_xo_posn_yobs = Peak_Calculation_Step; | ||

+ | |||

+ | //yobs_to_xo_posn_yobs// can be a function of the reserved parameter X with X being the value of the x-axis at the hkl//.// | ||

+ | |||

+ | It is important to determine and then fix all peak shape, zero error and lattice parameters before using yobs_to_xo_posn_yobs. Also, if the original diffraction pattern is noisy then it may be best to smooth it using the //smooth// keyword or re-binned using //rebin_with_dx_of//. Alternatively, a calculated pattern could be used as input into the yobs_to_xo_posn_yobs | ||

+ | |||

+ | Note that structure solution can be speed up by preventing graphical output or by increasing the Graphics Response Time in the GUI. | ||

+ | |||

+ | See examples CIME-DECOMPOSE.INP and PBSO4-DECOMPOSE.INP. |