The data collection was by recording cell counts and timings of various CAD algorithms. A full explanation can be found in the associated paper.

Regarding the example set in Section82-ExampleSet.txt, we note that the timings for Maple reported in the paper were from running Maple in command line mode.

The same example set was tested in Qepcad. Here explicit ECs for a parent formula were entered in dynamically as products of the individual sub-formulae ECs, in cases where an explicit EC exists.

Finally, the example set was also tested in Mathematica. Mathematica's CAD command does not return cell counts - these were obtained upon request to a Mathematica developer. Hence they are not recreatable using the information here (something outside the control of the present authors).