Getting Output

An optimization can return its best result, report evaluations as they finish, store results for later analysis, and write progress messages. This page helps you choose how to get the output you need; the linked pages explain each option.

What was the best result, and why did the optimization stop?

optimize returns an OptimizationResult with the best feasible result and an exit code. The result is None if the optimization did not produce a feasible point. Read the exit code to distinguish convergence, limits, a callback stop, and insufficient successful evaluations. An abort raises an exception instead.

See The result, Exit Codes, and Working with Results for the returned values and their fields.

How can I follow progress or stop early?

Pass a result_callback to receive each function result as the optimization runs. Set report_gradients=True to receive gradient results too. A callback can also stop the optimization by returning True.

See Reporting progress and Stopping early from the callback.

How can I keep evaluation results?

Pass a handler with the handlers argument. HistoryHandler keeps every result it receives; ResultsHandler keeps the best or most recent valid result; and DataFrameHandler collects selected fields into pandas or Polars tables. Handlers can also process events or call custom code.

See Result Handlers for built-in handlers, custom handlers, and sharing handlers across optimizations.

How can I inspect or export result data?

Result objects include the evaluated variables, objective and constraint values, gradient information, and optional metadata. To select fields, label them, or export them as pandas or Polars data, see Working with Results.

Where do progress messages and optimizer files go?

ropt progress messages use Python's standard logging module. The optimizer backend can also report its own progress with backend.verbose. optimizer.output_dir gives the optimizer a directory for its files. optimizer.stdout and optimizer.stderr capture the process output streams while the optimization runs, including output from evaluation code.

See Logging and Choosing an Optimizer for details.