Changeset 795 for sans/Dev/trunk/NCNR_User_Procedures/Reduction/SANS/ANSTO_DataReadWrite.ipf

Timestamp:

Apr 4, 2011 12:18:33 PM (11 years ago)

Author:

srkline

Message:

Changes to SANS reduction that apply to other Facilities:

These changes are related to the propagation of errors in 2D, on a
per-pixel basis. These changes only affect the errors that are reported in
the QxQy? ASCII file output. The 1D code is unaffected.

If these changes are not implemented, then errors of zero will be substitued as defaults
for these experimental errors.

Upon data loading, an error matrix, linear_data_error is generated and filled with
error values appropriate for Poisson statistics (not simply sqrt(n)).

4 functions in FACILITY_DataReadWrite.ipf have been added, and they are rather
self-explanatory:

• WriteBoxCountsErrorToHeader?()
• WriteTransmissionErrorToHeader?()
• getBoxCountsError()
• getSampleTransError()

In FACILITY_Utils.ipf, the AttenuatorTransmission?() function now returns
an additional parameter, atten_err, which is one standard deviation of the
attenuator transmission value. It returns a default error=0 (which is
correct if no attenuation is used). Facilities can fill this function in
with their own estimates for the uncertainty in the attenutator transmission.

File:

• sans/Dev/trunk/NCNR_User_Procedures/Reduction/SANS/ANSTO_DataReadWrite.ipf (modified) (3 diffs)

sans/Dev/trunk/NCNR_User_Procedures/Reduction/SANS/ANSTO_DataReadWrite.ipf

@@ -202 +202 @@
         String/G root:Packages:NIST:RAW:fileList = textw[0]
         
-        Duplicate/O data $"root:Packages:NIST:RAW:linear_data"                  //data read in is on linear scale, copy it now
-
+        Duplicate/O data linear_data                    //data read in is on linear scale, copy it now
+
+        // proper error for counting statistics, good for low count values too
+        // rather than just sqrt(n)
+        // see N. Gehrels, Astrophys. J., 303 (1986) 336-346, equation (7)
+        // for S = 1 in eq (7), this corresponds to one sigma error bars
+        Duplicate/O linear_data linear_data_error
+        linear_data_error = 1 + sqrt(linear_data + 0.75)                                
+        //
+        
         Return 0
 
@@ -409 +417 @@
         data=temp0
         Redimension/N=(pixelsX,pixelsY) data            //,linear_data
+
+        Duplicate/O data linear_data_error
+        linear_data_error = 1 + sqrt(data + 0.75)
+        
+        //just in case there are odd inputs to this, like negative intensities
+        WaveStats/Q linear_data_error
+        linear_data_error = numtype(linear_data_error[p]) == 0 ? linear_data_error[p] : V_avg
+        linear_data_error = linear_data_error[p] != 0 ? linear_data_error[p] : V_avg
         
         //linear_data = data
@@ -517 +533 @@
 // (3) close the file
 
-
+// new, April 2011 for error propagation. fill these in with the facility-
+// specific versions, if desired.
+Function WriteTransmissionErrorToHeader(fname,transErr)
+        String fname
+        Variable transErr
+        
+
+        return(0)
+End
+
+Function WriteBoxCountsErrorToHeader(fname,rel_err)
+        String fname
+        Variable rel_err
+        
+        return(0)
+End
+
+Function getSampleTransError(fname)
+        String fname
+        
+        return(0)
+end
+
+Function getBoxCountsError(fname)
+        String fname
+        
+        return(0)
+end
+
+
+// end April 2011 additions
 
 //whole transmission is NCNR-specific right now