In order to measure the run time of varying implementations without too much code modification, we use the time utility, available in -based operating systems. This brings portability to the framework, but has a consequence on the time resolution which can be not very fine and depends on the hardware. It's common for the internal processor clock to have a resolution somewhere between hundredths and millionths of a second. In our case, we have millionth of a second.
When the run time of one operator is small, e.g. the satisfiability test, we repeat the operation many times, e.g. times, to increase the accuracy of time measurement. This also reduces uninteresting overheads such as input/output run time, since these factors are divided by , which is chosen as . However, given that run times of complex operators such as convex hull are too long, we do not repeat these operations, i.e. is .