Synthetic-data-sets-via-hemisphere-mixing

Hemisphere mixing used to create synthetic dataset with three important properties: 1) it has kinematics similar to the expected background, 2) it is suppressed in signal and 3) is larger than the original 4b dataset. With these three properties, the synthetic dataset can be used to measure the expected variance and bias of the data-driven prediction. And can be additionally used to check for spurious signal.