The rport [pdf]

Curved Kernel Circle Traits

The following results show the running times of the Curved-Kernel conic traits with various number types. The first 2 data sets consist of degenerate circles and the third consists is random circles. The configuration with the gmpz number-type offered by CORE shows the best running times. Therefore, we chose it to represent the Curved-Kernel circle traits in consequences experiments.

Benchmark Results

Data - Full Ellipses and Circles

Databases of Experiments

All Traits

Benchmark Results

Data - Circles and Ellipses

Databases of Experiments

High Vertex Degree

Only the Exacus conic traits seems to handle these cases (The curved-kernel) can handle some, but not all, and it hasn't been enabled yet for these cases.

Benchmark Results

Data - Ellipses (high vertex degree)

Databases of Experiments

Benchmark Results

Data - Full Ellipses and Circles

Databases of Experiments

Circular Arcs

The single test case consists of degenerate circular arcs. The input format at this time requires the 2 endpoints to lie on the supporting circle. This will be extended to a more general format where the endpoints define a line (and a direction, which in turns defines the arc).

Benchmark Results