| Author |
 Topic  |
|
|
Lindax
Germany
1 Posts |
 Posted - 01/22/2003 : 04:51:43 AM
|
Hi,
I want to fit 6 graphs with the same fitting function (pseudovoigt1). I wrote a script to do the fitting and it actually works pretty well if I just fit one graph. But as soon as I try to integrate the fitting routine in a loop it does not work anymore. I suppose the problem is due to the "getpts" command I use to go the starting values of the peaks. This command is interrupted so that the quittoolbox macro is executed when I try to do the loop.
My question is: Is there any command that says that first I have to get the points from the graph, then the endtoolbox macro has to be executed and then I want to repeat it for the next graph window.
Here I send you the script I basically used (I left out some details about the fitting routine which should not be of interest for this specific problem)
loop (j,1,6)
{
window -a graph$(j);
nlsf.func$=psdvoigt1;
/*left out something*/
GetNumber -s $Fit.FitNPeaksnumpeaks NumPeaks;
nlsf.numReplica=NumPeaks-1;
/*left out some initialisation*/
def EndToolbox
{
doTool 0;
loop(i,1,NumPeaks)
{
if (NumPeaks==1)
XC=_Peak_A[i];
else
XC$(i)=_Peak_A[i];
}
nlsf.fitdata$=%C;
noFitCurve=-1;
nlsf -v 0 1 3;
nlsf.datastep*=2;
if(nlsf.iterate(2)==-1)
return 1;
nlsf.datastep/=2;
nlsf.iterate(1000000);
nlsf.datastep/=20000;
nlsf.iterate(1000000);
nlsf -v 1 1 3;
nlsf.datastep*=2;
nlsf.iterate(100000);
nlsf.datastep/=5000;
nlsf.fit(1000000);
nlsf.end(100000);
};
GetPts -n _Peak_Y NumPeaks [Double-click (or strike Enter) to set Peak$(1+GetPts.count)];
}; /*end of loop*/
I would appreciate any suggestions how to solve that problem.
Linda
|
|
|
CStorey
Canada
137 Posts |
Posted - 01/22/2003 : 2:23:04 PM
|
Hi Linda,
I think your problem is going to be with the doTool command not causing the loop to pause and wait for your input.
The doTool Macro is a wierd beast. I've tried to do something similar before and run into problems. I suggest that you modify the pointProc Macro to do the work for you once you make a selection with the doTool data selector.
Here's an example of multiple Linear Regressions using the data selector and a redifinied PointProc Macro. In the end it worked pretty well. I think the secret was breaking the code up into sections...but I didn't write myself enough documentation after the fact to be able to remember all the details.
Hope this helps!
[main]
type.redirection=1; // All "type" statements are printed to the 'Script Window'
system.debug=1; // Debugging is turned ON!
echo=1;
////// Ask User to pick a graph //////
ijk=1;
%Z="";
doc -e P { %Z=%Z %H;};
getn
(Graph List:) ijk:Z
(Select the Graph to find Ft/Fmax...); // end of getnumber
%V=%[%Z, #ijk];
////// Create a Worksheet for the Output //////
win -t wks Origin %Vftfmx; // Create a new worksheet
%N=%H; // Remember worksheet name
wks.col1.format=2; // Set Colum A as Text
wks.col1.name$="Curve"; // Rename A to Dataset
wks.col2.name$="FtFmax";
wks.col1.width=20; // Widen Column #1 to 20 chars
wks.col2.width=12; // Widen Column #2 to 12 chars
wks.addcol(B);
wks.col3.name$="Ib";
wks.addcol(C);
wks.col4.name$="Ic";
win -a %V; // Make selected the selected Graph the Active Graph
win -z; // Makes Active graph Maximized!
type -b "Sect the Data Range for LR using the mouse then press 'Enter'";
// Layer -c command:
// Count how many datasets are plotted in the active layer.
// Put the result into the count variable, and concatenate
// the names of all datasets into %Z.
type "***** Ft/Fmax Analysis for Graph %V ******";
type "====================================================";
layer -c;
loop(ii,1,3)
{
%C=%[%Z,#ii];
run.section(,LRAnalysis);
}
[LRAnalysis]
dotool -dh 4;
def pointproc{
type " $(ii)) %C curve";
%U=xof(%C);
lr -x %U %C -b mks1 -e mks2;
xintercept=-1*(lr.a)/(lr.b);
type " --> Range = ($(ptbx), $(ptby)) to ($(ptex), $(ptey))"; /// LR Range in XY for each curve
type " --> Slope = $(lr.b,%9.4g) +- $(lr.sdB,%9.4g)";
type " --> Y-intercept = $(lr.a,%9.4g) +- $(lr.sda,%9.4g)";
type " --> Correlation R = $(lr.r,%9.4g)";
type " --> X-intercept = $(xintercept,%9.4g)";
type "====================================================";
done=1;
dotool 0;
};
for(done=0;done==0; ) { sec -p .05; }; ///// Wait until Done=1!
Return
Craig Storey
Origin WebRing Member - http://g.webring.com/hub?ring=originwebring |
 |
|
| |
Topic |
|
|
|
loop in multiple curve fitting