<* $Id: xnd_file.html,v 1.1.1.1 1999/07/31 10:41:00 berar Exp $ *> Files used in xnd program.

Files used in xnd program.


xnd top Next file Previous file

The following files specified, with their current extension, are described below : These other file are described apart:

".lst", the listing file.

This output file is used in the reading part to give user a detailled copy of all data in the ".k" file. This allow the correction of the ".k" file when some mistakes occur.
Then during the refinement procedure, it produces a summary of what append. At the beginning of each cycle, the scale factor of each phase is given after its correction by the phase transition coefficent. The number of used (significant) Bragg lines is given and compared with the number of generated and non zero lines.
The displacement of the parameters are presented in a table containing the new position, the shift and the standard deviation. If the shift can be seen as small according to EpsStop and to the standard deviation a star "*" is added at the end of the line.

Then some statistical ratio are given.

Rwp     standard reliability factor       sqrt (Sum(WdI^2) / Sum(WIo^2));
Rp      unweigthed reliability factor     Sum(Abs(dI)) / Sum(Io)
Rwp_c   modified reliability factor to take into account local correlation
        Berar,Lelann, J.Appl.Cryst. 24 (1991) p1-5
GoF     Goodness of Fit                   Rwp/Rexp
Rwp_b   background modified Rwp           sqrt (Sum(WdI^2) / Sum(W(Io-Iback)^2));
Rp_b    background modified Rp            Sum(Abs(dI)) / Sum(Io-Iback);
N_b     number of points where there is an intensity over the background
        these points are used in calculating Rwp_b et Rp_p,
R_exp   statistical limit expected for Rwp
d-2     Durbin-Watson distance -2, without correlation the limit is 0
Ne      efficient number of points (N decreased of the parameters number)
N       number of data points
Minimised_Sum Raw value of the minimised sum
Penalities    weigth (percent) of penalities like bond length in the minimised sum 
At the last cycle, result on integrated intensities are also given for each phase before values over all phases.
R_Bragg reliability factor on hkl intensities  Sum(Ihkl) / Sum(Ihkl_o)
S_I     sum of the calculated hkl intensities  Sum(Ihkl)
hkl     number of lines used in the calculation
Rw_B    weighted factor                        sqrt(Sum(W Ihkl^2) / Sum(W Ihkl_o^2))
S_wI    weighted sum of the calculated hkl intensities
R_f     reliability factor on structure factors  Sum(Fhkl) / Sum(F_o)
S_f     sum of the calculated Fc  Sum(Ihkl)
Rwf     weighted factor                        sqrt(Sum(W Fhkl^2) / Sum(W F_o^2))
S_wf    weighted sum of the calculated hkl intensities
hkl     number of lines used in the calculation

".new", updated coordinates file.

This is a faithfull copy of the .k file containing the actualized values of the refined parameters. This file also contains all the comments present in the ".k" file.
When the number of data points in the ".k" file does not agree with the data file, the real number is printed in the ".new" file.
The value of prtNew is increased and a comment line with prtNew and the final value of the minimized sum is added at the end of the file.
Without specifications, the ".new" is writen according to the current release input.

".plg", the profile file.

It is a 6 columns file suitable to be graphically displayed. A common way to display pattern on unix station is to use the xmgr software originaly writen by P.Turner. To display all data use "xmgr -nxy file.plg" or read the sets with the XY1Y2... options.
xmgr is now refered as grace ( http://plasma-gate.weizmann.ac.il/Grace ).
After the header lines (#) the 6 columns are:
12theta
2Yobsobserved counts
3Ycalcsum of calculated counts
4Ybkgcalculated background
4dY error or discrepency between observed and calculated counts
4dy/sYrelative error

".hkl" file.

This file contains the list of the calculated hkl, their positions, their integrated intensities and some information on the profile characteristics. For classical samples the columns are described in the following table.
1line number
2pphase number according to xnd or pseudo-phase number
3hMiller indices
4k
5l
6lambda number
72thetaline position including zero shift...
8symbol according to the confidence in i_obs estimation
9i_obs
10i_calc
11f_areal part of structure factor
12f_bimaginary component of structure factor
13w_lFWHM of the Lorenzian component in the Voigt function
14w_gFWHM of the Gaussian component in the Voigt function
15as_1asymetry linear coefficient
16as_3asymetry cubic coefficient
17muline multiplicity
18orienprefered orientation value
19corLorentz-polarisation factor
20d_hklinterplanar distance
21s_iESD on i_obs
22first angle used in the profile function
22last angle used in the profile function
The observed intensities are calculated according to method : the first uses the classical Rietveld formula, it gives a value for the whole intensity seen in the calculation range; the second method uses a weighting scheme and is more sensitive to the intensity near the line center. The difference between both estimations is used to obtain the confidence indicator :
a < 2%; b < 4%; c< 8%; d < 16%; e < 32%; else f.

data files.

No particular extension is necessary. The known formats are described below. They are ascii files and in all cases the only real rule is that all data points have strictly increasing angles. If overlapp are needed, more than one experimental file are needed.
If the weight are not specified, they are assumed to be Poissonian (W=1/Io). If standard deviation are read, the weight are (W=1/s^2)

Simple files with angles and counts

The simpliest files contain the angle and the count on each line, they are recognized according to:
codecolumns
12thetaint
2theta2thetaint
32thetaints
4theta2thetaints
A general coding rule (100-199) allow to read other files belong this kind '1nm' means skip n columns, read 2theta, skip m values and read intensity. Then '1' appears as a short way for '100' and '2' for '110'.
Similarly the coding rule (1000-1999) allows to read file with standard deviations. Then '3' is a short for '1000' and '4' for '1100'.

Fixed step files

This files nead an header which allow to read more than one block. The known files are.
codecontent
101 column
118 columns
1210 columns whitout number of counters
1310 columns whit the number of counters ignored
1410 columns whit the number of counters used
158 columns whit the number of counters ignored
168 columns whit the number of counters used

With their header, these files look like

  BLOCK SCALE WSCALE
  2THETA0 PAS 2THETAF MONO 2T0 2TF
  i0 i1 i2 i3 i4 i5 i6 i7 xxx
  i8 i9 ....
  2THETA1 ...
where
BLOCKnumber of blocks to be read
SCALEvalues are multiplied by SCALE when read
WSCALEstandard deviation are multiplied by WSCALE
2THETA0first used value in the block which begin at 2TO
2THETAFlast used value in the block which end at 2TF
MONOmonitor count or zero
2T0first writen value in the block, not mandatory if equal to 2THETA0
2TFlast writen value in the block, not mandatory if equal to 2THETAF
This allow the use of data recorded with various counting time or monitoring. The effective values are
  INT = Ix * SCALE
  SIGMA = WSCALE * sqrt (INT) + sqrt(MONO)
If the number of counters is specified, this is modified in
  INT = Ix * SCALE
  SIGMA = WSCALE * sqrt (INT/n) + sqrt(MONO)

Multiple files

Few files containing more than one experiment are recognized.
codekind of files
20 ILL D1B, D20 (1990 format)
21 XDMI files from LURE...
 nSETS   nPTS    nPAR  nEFF
 2theta00 2theta01 ....
 2theta10 2theta11 ....
 eff00    eff01    ....
 eff10    eff11    ....
 0        par1(0)  ....
 int00    int01    ....
 int10    int11    ....
 ....
 1        par1(1)  ....
 int0     int1     ....
 ....
Currently this file is presented with 10 values on each line. The reference number of the control value have to be lower than nPAR, which is the number of parameters in the file. The efficiency table is used to correct the data if nEff is equal to 1; if nEff is equal to -1, the correction has soon be done on data but the weight have to be corrected.

optional binary files.

To allow a faster access to data files. These files can be created by xnd, when they are used a very short check is done to ensure that the number of points in the file agrees with the number declared in the ".k" file.

temporary files.

   Le programme utilise deux fichiers  temporaires  binaires.   Le  premier
contient  les  donnees  experimentales  des divers diagrammes, il peut etre
conserve pour eviter la relecture de ces donnees, son nom doit  alors  etre
fournit  par  l'utilisateur.   Le second contient la table des derivees des
parametres sur le diagramme en  cours  d'affinement,  et  les  valeurs  des
intensites.  Il  sert  d'extension  a  la  memoire  si  cela est necessaire
(parametre disk_dydp et fonctions ...dydp). Sur micro-ordinateur de type PC
ou  MAC,  le  compilateur  place les fichiers temporaires dans le directory
courant, on peut utiliser un disque virtuel (ram_disk, memoire  etendue...)
pour  loger ces fichiers en utilisant ce disque comme directory de travail.
En indiquant un nom complet avec path les fichiers coordonnees,  listing...
peuvent  resider  sur un autre disque, le gain de temps peut etre important
et atteindre 30% (cf essai sur MacSE30). La taille necessaire pour loger ce
fichier est indiquer dans le listing a la fin de la phase de lecture.


xnd top Next file Previous file

JFB Aug 25th, 1999