Run it and read the result
Running is the part that should be boring. exp.run() picks a backend, executes the generated code, and hands back a result container whose axes are named — time, node, variable, and one axis per swept parameter — so a selection is written by label and cannot silently land on the wrong region. The same container is what figures bind to and what a report reads its numbers from.
Where it runs is a deployment detail, not a modelling one. A single subject on a laptop and a cohort on a GPU partition execute the same emitted code.
Choosing a backend, what each one is good at, and how to pin one.
The experiment resultThe result container: its axes, its coordinates, and how to select from it.
Plot a result directlyQuick looks at a run — raster, timeseries, phase plane, network.
HPC patterns & containersTaking the same study to a cluster without hand-writing a job script.
Next
A finished run is worth publishing: ⑤ Share.