The plots show the second-order coefficient of the perturbative expansion for the EEC, including a factor of \sin^2\chi which makes the curves finite. ie, the definition is exactly the same as in Clay and Ellis. The errors come from Monte Carlo statistics (there are no parameters, so that is the only error). The phase-space generation was optimised for the logarithmic regions - if it is useful to have smaller errors in the central region I can do so.
Using the subtraction method, as we do, fully integrated quantities have
much smaller errors than differential quantities. Therefore the moments
of B, defined by:
B[m,n] = \int_{-1}^{1} d\cos(\chi) B(\chi) \cos^m(\chi) \sin^n(\chi)
should give a very precise check that the three Monte Carlo programs
give exactly the same B(\chi). (or not...?) Although we expect to disagree
with the Clay-Ellis calculation, we include their numbers for comparison.
The moments are tabulated separately
for each group factor.
Mike Seymour