Configurable SIERRA Variables#

Variables configure experiments uniformly across an entire batch — every experiment in the batch gets the same value. This distinguishes them from Batch Criteria, which define axes of variation that produce different experiments. For the conceptual distinction, see Variables vs. Batch Criteria.

Experiment Setup#

Configures Experiment duration, controller cadence (Tick duration/timestep), and the number of datapoints captured per Experimental Run. Supported by the ARGoS and ROS1-based engines; if your engine does not support it, it has no effect.

Cmdline Syntax#

exp_setup.T{duration}[.K{ticks_per_sec}]

  • duration — Duration of the experiment in seconds (not timesteps/ticks).

  • ticks_per_sec — How many times per second each controller outputs data. Controllers may execute more frequently internally, but SIERRA uses this value to determine how many datapoints to expect per run. The number of captured datapoints is approximately duration * ticks_per_sec.

duration must always be specified. ticks_per_sec is optional.

Important

All Experimental Runs must produce the expected number of datapoints. If a run produces fewer, SIERRA's graph generation will produce incorrect results because it computes axis ranges and statistics from the expected count.

Examples#

  • exp_setup.T1000: Run is 1,000 seconds long with the default 5 ticks/sec, giving 1,000 × 5 = 5,000 timesteps, assuming the engine default is capturing one datapoint per 100 timesteps, giving approximately 50 datapoints per run.

  • exp_setup.T10000.K10: Run is 10,000 seconds long with 10 ticks/sec, giving 10,000 × 10 = 100,000 timesteps and approximately 50 datapoints (same default capture interval of one per 100 timesteps).

Note

If you are writing a new engine plugin and your engine models experiment time in terms of duration and controller cadence, adopting --exp-setup gives users a consistent interface across engines. See New Engine Plugin (--engine) for details.