This directory contains suites from the 20 domains we have used for
evaluating FF's performance. The suites are either taken from the
competition distributions, from the literature or from the IPP domain
collection, and are sometimes adapted to show scaling behavior. Every
subdirectory contains a README file describing the origin of the
instances, and the modifications made, if any. The instances are
exactly those that we used for comparing FF's performance to that of
HSP, Section 8.3. Where possible, we avoided very small instances,
i.e. such instances that get solved by FF in less than 0.2 seconds,
and favored tasks that take FF some dozen seconds to solve. Thus, the
tasks are generally quite large. In Briefcaseworld, there is also a
suite with smaller tasks that we used for obtaining some runtime and
solution length results (as are shown in Section 8.2) with IPP
compared to FF.
