Effective policies for workflow scheduling in stochastic environments

In both business and computing, complex projects are often defined as workflows, which describe the relationships between the project's tasks. A fundamental problem is how to schedule these tasks, that is to say who and when has to perform them, to achieve specific performance targets. This becomes difficult especially when the tasks require an uncertain amount of time or when there is a queue of overlapping projects. Additionally, in many projects some tasks are executed, some are discarded and some can be repeated many times, based on conditions evaluated at runtime. My work presents algorithms which solve such problems effectively.