Ignore:
Timestamp:
Oct 18, 2018 12:09:16 PM (4 years ago)
Author:
srkline
Message:

extensive changes to accomodate 1x1 binning of the HighRes? detector. It is implemented as a global flag. Currently only 4x4 and 1x1 are allowed. 1x1 has never been tested in reality, only simulated data - so my assumed dimensions may not be correct. look for TODOHIGHRES in the file for places that may need to be updated for different file dimensions. Testing of the simulated data is proving to be excruciatingly slow, but passable for a test. Speed optimization will be needed if this is the final solution. Memory management will also be an issue since every "copy" of the highRes matrix is enormous. Carry as few of these around as possible in the future to keep the experiment size to something reasonable.

File:
1 edited

Legend:

Unmodified
Added
Removed
  • sans/Dev/trunk/NCNR_User_Procedures/Reduction/VSANS/V_Utilities_Comparisons.ipf

    r1105 r1117  
    1515// so passing 0.01*val_1 = 1% tolerance, as long as val_1 can't be zero 
    1616// 
     17// SEP 2018 -- increased the tolerance to 2%, since I was getting false differences 
     18// especially for the lateral offset after switching from trans->scatter configs. Panel 
     19// was returning to postion, but within 2% (since the value was near zero) 
    1720// 
    1821 
     
    302305 
    303306        if(val1 != 0) 
    304                 tol = abs(0.01 * val1) 
     307                tol = abs(0.02 * val1) 
    305308        else 
    306                 tol = abs(0.01 * val2) 
     309                tol = abs(0.02 * val2) 
    307310        endif 
    308311         
     
    328331         
    329332        if(val1 != 0) 
    330                 tol = abs(0.01 * val1) 
     333                tol = abs(0.02 * val1) 
    331334        else 
    332                 tol = abs(0.01 * val2) 
     335                tol = abs(0.02 * val2) 
    333336        endif 
    334337         
Note: See TracChangeset for help on using the changeset viewer.