User: hinder Date: 2013/10/09 12:57 PM

Modified: /trunk/src/ integrate.cc

Log: integrate.cc: Add a comment on the lack of symmetry of the "midpoint" integration method

File Changes:

Directory: /trunk/src/ ======================

File [modified]: integrate.cc Delta lines: +7 -0 =================================================================== --- trunk/src/integrate.cc 2013-10-09 17:57:15 UTC (rev 102) +++ trunk/src/integrate.cc 2013-10-09 17:57:23 UTC (rev 103) @@ -41,6 +41,13 @@

assert(nx > 0); assert(ny > 0); assert (f);

+ // This sum should probably exclude the upper end points. As + // currently implemented, it does not naturally map to the "sum of + // areas" definition of the integral, as there is an additional + // contribution from the last term in the sum. This introduces an + // additional first order error to a method which is already only + // first order. + for (iy = 0; iy <= ny; iy++) for (ix = 0; ix <= nx; ix++) integrand_sum += f[idx(ix,iy)];