How kvalevska/agtest-framework works
AGtest-framework is a desktop workbench for configuring and analyzing multi-objective optimization experiments. The principal workflow selects a registered problem and algorithm, runs a single experiment with generation updates, then presents Pareto and quality-metric results for inspection and spreadsheet export. Separate branches support finite Cartesian Multi experiments and per-metric trajectories. The README documents optional parallel evaluation and plot controls; sampled excerpts do not expose all GUI wiring, so uncertain internal calls are omitted.
Main entry points: app.py Main GUI · multi.py Multi runner · algorithms Optimizers · registry.py Problem registry
Components 16
Desktop interface
- Main GUI
src/pymoo_gui/app.py
Experiment domain
- Problem registry
src/pymoo_gui/problems/registry.py - Benchmarks
src/pymoo_gui/problems - Algorithm registry
src/pymoo_gui/algorithms/__init__.py - Optimizers
src/pymoo_gui/algorithms - Research MOEA base
src/pymoo_gui/algorithms/research_common.py - Parallel evaluation
src/pymoo_gui/parallel.py
Results and analysis
- Pareto plots
src/pymoo_gui/viz/pareto_dialogs.py - Quality metrics
src/pymoo_gui/metrics/quality.py - Metric export
src/pymoo_gui/metrics/export.py - Metric trajectories
src/pymoo_gui/viz/metric_trajectories.py
Batch execution
- Multi runner
src/pymoo_gui/multi.py
Other components
- Experimenter
- pymoo engine
- Platypus library
- XLSX files
Connections 21
- Experimenter → Main GUI: configures runs
- Main GUI → Problem registry: selects problem
- Problem registry → Benchmarks: creates problem
- Main GUI → Algorithm registry: selects algorithm
- Algorithm registry → Optimizers: creates optimizer
- Optimizers → Research MOEA base: reuses utilities
- Optimizers → pymoo engine: uses engine
- Optimizers → Platypus library: optional adapter
- Main GUI → Optimizers: runs experiment
- Main GUI → Parallel evaluation: optionally evaluates in parallel
- Main GUI → Pareto plots: updates visualization
- Main GUI → Quality metrics: computes generation metrics
- Main GUI → Metric trajectories: shows metric history
- Main GUI → Metric export: exports results
- Metric export → XLSX files: writes workbooks
- Experimenter → Multi runner: starts batch
- Multi runner → Problem registry: uses problem defaults
- Multi runner → Algorithm registry: uses algorithm defaults
- Multi runner → Optimizers: executes combinations
- Multi runner → Quality metrics: records metrics
- Multi runner → Metric export: exports results