Ignore:
Timestamp:
03/25/11 20:25:53 (9 years ago)
Author:
aivar
Message:

Got rid of the global old_chisquare, it's passed in as a parameter now. Some test code, commented out, bails out if the delta params go to zero. This gives about a 10% boost in performance. At least, in this case, we should avoid making new alpha and beta matrices.

File:
1 edited

Legend:

Unmodified
Added
Removed
  • trunk/projects/slim-curve/src/main/c/EcfGlobal.c

    r7611 r7675  
    32453245////                                    yfit[i], dy[i], alpha_scratch, beta_scratch, 
    32463246                                        yfit[0], dy[0], alpha_scratch, beta_scratch, 
    3247                                         &chisq_trans[i], (i == 0) ? alambda : 0.0, 
     3247                                        &chisq_trans[i], 0.0f, (i == 0) ? alambda : 0.0, //TODO ARG added 0.0f here for new old_chisq parameter 
    32483248                                        pfnvals, pdy_dparam_pure, pdy_dparam_conv, 
    32493249                                        pfnvals_len, pdy_dparam_nparam_size); 
Note: See TracChangeset for help on using the changeset viewer.