class RooAcceptReject: public TNamed, public RooPrintable

 Initialize an accept-reject generator for the specified distribution function,
 which must be non-negative but does not need to be normalized over the
 variables to be generated, genVars. The function and its dependents are
 cloned and so will not be disturbed during the generation process.

 Reattach original parameters to function clone

 Return a pointer to a generated event. The caller does not own the event and it
 will be overwritten by a subsequent call. The input parameter 'remaining' should
 contain your best guess at the total number of subsequent events you will request.

 Scan through events in the cache which have not been used yet,
 looking for the first accepted one which is added to the specified
 container. Return a pointer to the accepted event, or else zero
 if we use up the cache before we accept an event. The caller does
 not own the event and it will be overwritten by a subsequent call.

 Add a trial event to our cache and update our estimates
 of the function maximum value and integral.

 Generate the minimum required number of samples for a reliable maximum estimate