Output structure

Each experiment run creates a timestamped output directory under:

experiment_output/<experiment_name>/experiment_<timestamp>/

The folder is created automatically if it does not exist. Inside it, the client stores the experiment under the experiment name and writes the files produced during acquisition.

For a non-swept experiment, the output typically includes:

• config.json: the full configuration used for the run,

• data.pkl: the main data output,

• end_message.json: summary information reported by the server.

For swept experiments, the client writes one pickle data file per sweep step (for example, data_0.pkl), a per-step status file (for example, exp_0.json), configuration.json, var_order.json, and end_message.json. The swept experiment also contains the initial config.json created for the experiment.

If the experiment contains a single sweep variable with 100 values, the output contains files such as:

• data_0.pkl, exp_0.json

• …

• data_99.pkl, exp_99.json

If the experiment contains two sweep variables, such as gain and frequency, the output uses a two-index naming convention such as:

• data_0_0.pkl, exp_0_0.json

• …

• data_99_99.pkl, exp_99_99.json

The numbering reflects the sweep index associated with each variable. The file var_order.json records the order of the sweep variables so that each index can be mapped back to the corresponding parameter.