Online Single Machine Batch Scheduling

We are concerned with the problem of safely storing a history of actions that happen rapidly in real time, such as in “buy” and “sell” orders in stock exchange trading. This leads to a single-family scheduling problem with batching on a single machine, with a setup time and job release times, under batch availability. We investigate the objective of minimizing the total flow time in an online setting. On the positive side, we propose a 2-competitive algorithm for the case of identical job processing times, and we prove a lower bound that comes close. With general processing times, our lower bound shows that online algorithms are inevitably bad in the worst case.