The workhorse of the RECAST framework is the back-end associated with a particular analysis. The back-end will
- supplement the signal sample with a tuned parton-shower and hadronization, if the input is provided at the parton-level;
- simulate the signal events in the detector, including effects of pile-up, if necessary;
- reconstruct the events with the corresponding algorithms used for the original search;
- apply all event selection criteria used in the original search;
- obtain the selection efficiency and necessary distributions of the selected signal events;
- determine the upper- and lower-bounds on the rate of events of this type consistent with the original analysis’s background estimates and observation with data; and
- obtain formal approval of the new results and communicate them to the front-end.
Initially, we anticipate that these steps will all be done manually and recognize that it will require a real investment from experimentalists to answer a RECAST request. We hope that this will evolve, and that the collaborations will develop systems so that analyses can be archived.