The bench.py script is a modified version of Andy's code (changes include additional models and csv output), and should be run as follows:

python ./bench.py >../results.csv

ALL models were run with the same tolerance parameters, which are set in bench.py:

absolute: 1.000000e-007
relative: 0.0001

In both roadrunner and Jarnac tests, the number of steps was set to 50.

The Jarnac benchmarking script is jarnac_bench.jan, and is run in the Jarnac UI.

*The jean_marie, biomod22, and biomod09 models caused errors in Jarnac and were excluded from benchmarking. *The biomod09 model also resulted in errors in roadrunner. *Memory exhaustion prevented Jarnac from completing three trials of the 450/500 Brusselator models. These were completed on a separate run of just the 450/500 Brusselator models.

• The timer resolution in Jarnac is milliseconds