Add file AGENTS.md

Author: ajnebro

AGENTS.md

@@ -0,0 +1,35 @@
+# AGENTS
+
+Guidance for humans and AI assistants collaborating on jMetalPy.
+
+## Scope
+- Use agents for refactors, tests, docs, experiment scripts, and small utilities.
+- Avoid delegating API changes, licensing, or research conclusions without maintainer sign-off.
+- Keep changes scoped; prefer incremental PRs over broad rewrites.
+
+## Safety and Privacy
+- Do not share credentials, private datasets, or unpublished results.
+- Assume no outbound network; request approval before downloads or package installs.
+- Avoid destructive commands (e.g., removing files, rewriting history). When in doubt, ask.
+- Work only inside the repository workspace.
+
+## Coding Expectations
+- Follow `CODING_GUIDELINES.md` (typing, docstrings, Python 3.11+, testing).
+- English-only identifiers and comments; use ASCII unless the file already uses Unicode.
+- Prefer clarity over cleverness; keep functions small and focused.
+
+## Tooling and Commands
+- Prefer `rg` for searches; summarize command output instead of dumping logs.
+- Run targeted `pytest`, `ruff`, and `mypy` for affected areas when feasible; note any tests not run.
+- Use `apply_patch` or minimal diffs; avoid auto-generated bulk changes unless requested.
+
+## Interaction Style
+- Be concise, reference paths with backticks (e.g., `src/module.py:42`), and summarize results.
+- Explain rationale for non-obvious choices; add brief comments only when code needs context.
+- Never revert user changes unless explicitly asked.
+
+## Pre-Submission Checklist
+- [ ] Changes comply with `CODING_GUIDELINES.md`.
+- [ ] Tests/lint/type checks run where relevant, with results noted.
+- [ ] No secrets or external data added; no destructive actions taken.
+- [ ] Summary of changes and affected files prepared for the reviewer.

README.md

@@ -13,6 +13,7 @@ A paper introducing jMetalPy is available at: https://doi.org/10.1016/j.swevo.20
 - [Installation](#installation)
 - [Usage](#hello-world-)
 - [Hyperparameter Tuning](#hyperparameter-tuning-)
+- [Agents](#agents)
 - [Features](#features)
 - [Changelog](#changelog)
 - [License](#license)
@@ -209,6 +210,10 @@ print(f"Best score: {result.best_score}")
 print(f"Best params: {result.best_params}")
 ```
 
+## Agents
+
+If you use AI assistants (e.g., Copilot, Codex) while working on this project, please follow the guidelines in [AGENTS.md](AGENTS.md).
+
 ### Available Example Configurations
 
 | File | Description |

examples/multiobjective/nsgaii/nsgaii_distance_based_archive_linf.py

@@ -2,9 +2,11 @@
 from jmetal.operator.crossover import SBXCrossover
 from jmetal.operator.mutation import PolynomialMutation
 from jmetal.problem import ZDT4
+from jmetal.problem.multiobjective.dtlz import DTLZ1
 from jmetal.util.archive import DistanceBasedArchive
 from jmetal.util.distance import DistanceMetric
 from jmetal.util.evaluator import SequentialEvaluatorWithArchive
+from jmetal.util.plotting import save_plt_to_file
 from jmetal.util.solution import (
     print_function_values_to_file,
     print_variables_to_file, read_solutions,
@@ -20,8 +22,8 @@
 
     problem.reference_front = read_solutions(filename="resources/reference_fronts/ZDT4.pf")
 
-    # Create distance-based archive with size 100 using L-infinity distance metric
-    archive = DistanceBasedArchive(maximum_size=100, metric=DistanceMetric.LINF)
+    # Create distance-based archive
+    archive = DistanceBasedArchive(maximum_size=100, metric=DistanceMetric.L2_SQUARED)
     evaluator = SequentialEvaluatorWithArchive(archive)
 
     max_evaluations = 25000
@@ -40,9 +42,14 @@
     # Get solutions from the archive instead of the algorithm result
     front = evaluator.get_archive().solution_list
 
-    # Save results to file with specific suffix
-    print_function_values_to_file(front, "FUN." + algorithm.label + ".LINF")
-    print_variables_to_file(front, "VAR." + algorithm.label + ".LINF")
+    # Save results to file
+    print_function_values_to_file(front, "FUN." + algorithm.label)
+    print_variables_to_file(front, "VAR." + algorithm.label)
+
+    # Save a PNG visualization of the front (and optional HTML if Plotly available)
+    png = save_plt_to_file(front, "FUN." + algorithm.label, out_dir='.', html_plotly=True)
+    print(f"Saved front plot to: {png}")
+
 
     print(f"Algorithm: {algorithm.get_name()}")
     print(f"Problem: {problem.name()}")

examples/multiobjective/nsgaii/nsgaii_standard_settings_with_crowding_distance_archive.py

@@ -44,11 +44,8 @@
     print_variables_to_file(front, "VAR." + algorithm.label)
 
     # Save a PNG visualization of the front (and optional HTML if Plotly available)
-    try:
-        png = save_plt_to_file(front, "FUN." + algorithm.label, out_dir='.', html_plotly=True)
-        print(f"Saved front plot to: {png}")
-    except Exception as e:
-        print(f"Warning: could not generate front plot: {e}")
+    png = save_plt_to_file(front, "FUN." + algorithm.label, out_dir='.', html_plotly=True)
+    print(f"Saved front plot to: {png}")
 
     print(f"Algorithm: {algorithm.get_name()}")
     print(f"Problem: {problem.name()}")

examples/multiobjective/nsgaii/nsgaii_standard_settings_with_distance_based_archive.py

@@ -62,6 +62,11 @@
     except Exception as e:
         print(f"Warning: could not generate front plot: {e}")
 
+    # Save a PNG visualization of the front (and optional HTML if Plotly available)
+    png = save_plt_to_file(front, "FUN." + algorithm.label, out_dir='.', html_plotly=True)
+    print(f"Saved front plot to: {png}")
+
+
     print(f"Algorithm: {algorithm.get_name()}")
     print(f"Problem: {problem.name()}")
     print(f"Archive size: {len(front)} solutions")

src/jmetal/tuning/cli/parallel.py

@@ -194,8 +194,6 @@ def main():
     seed = args.seed if args.seed is not None else config.seed
     population_size = args.population_size if args.population_size is not None else config.population_size
 
-    args = parser.parse_args()
-    
     # Build extra arguments to pass to workers
     extra_args = [
         "--trials", str(n_trials),

src/jmetal/tuning/tuning_parallel.py

@@ -385,12 +385,21 @@ def main():
         problems = None
 
     # Apply config values with CLI overrides
-    algorithm = args.algorithm if args.algorithm != "NSGAII" or config is None else config.algorithm
-    total_trials = args.trials if config is None else config.n_trials
-    max_evaluations = args.evaluations if config is None else config.n_evaluations
-    sampler = args.sampler if config is None else config.sampler
-    seed = args.seed if config is None else config.seed
-    population_size = args.population_size if config is None else config.population_size
+    # CLI arguments take precedence over config file
+    if config is not None:
+        algorithm = args.algorithm if args.algorithm != "NSGAII" else config.algorithm
+        total_trials = args.trials if args.trials != NUMBER_OF_TRIALS else config.n_trials
+        max_evaluations = args.evaluations if args.evaluations != TRAINING_EVALUATIONS else config.n_evaluations
+        sampler = args.sampler if args.sampler != "tpe" else config.sampler
+        seed = args.seed if args.seed != SEED else config.seed
+        population_size = args.population_size if args.population_size != POPULATION_SIZE else config.population_size
+    else:
+        algorithm = args.algorithm
+        total_trials = args.trials
+        max_evaluations = args.evaluations
+        sampler = args.sampler
+        seed = args.seed
+        population_size = args.population_size
 
     # Create observers based on argument
     # Note: plot observer only enabled on worker 0 to show global progress

src/jmetalpy.egg-info/PKG-INFO

@@ -105,6 +105,8 @@ A paper introducing jMetalPy is available at: https://doi.org/10.1016/j.swevo.20
 ### Table of Contents
 - [Installation](#installation)
 - [Usage](#hello-world-)
+- [Hyperparameter Tuning](#hyperparameter-tuning-)
+- [Agents](#agents)
 - [Features](#features)
 - [Changelog](#changelog)
 - [License](#license)
@@ -193,6 +195,128 @@ plot_front.plot(front, label='NSGAII-ZDT1', filename='NSGAII-ZDT1', format='png'
 
 <img src=docs/source/_static/NSGAII-ZDT1.png width=450 alt="Pareto front approximation">
 
+## Hyperparameter Tuning 🎛️
+
+jMetalPy includes an Optuna-based hyperparameter tuning system that allows you to automatically find optimal algorithm configurations.
+
+### Quick Start
+
+```python
+from jmetal.tuning import tune
+
+# Basic tuning with default settings
+result = tune("NSGAII", n_trials=50)
+print(f"Best parameters: {result.best_params}")
+```
+
+### Using YAML Configuration Files
+
+The recommended way to configure tuning experiments is through YAML files:
+
+```bash
+# Run with a configuration file
+python -m jmetal.tuning.cli.sequential --config examples/tuning/configs/default.yaml
+
+# Override specific settings via CLI
+python -m jmetal.tuning.cli.sequential --config examples/tuning/configs/quick_demo.yaml --trials 30
+```
+
+### Example YAML Configurations
+
+**Quick Demo** (`examples/tuning/configs/quick_demo.yaml`):
+```yaml
+algorithm: NSGAII
+n_trials: 20
+n_evaluations: 5000
+
+problems:
+  - name: ZDT1
+    reference_front: ZDT1.pf
+  - name: ZDT2
+    reference_front: ZDT2.pf
+```
+
+**SBX-Only Configuration** (`examples/tuning/configs/sbx_only.yaml`):
+```yaml
+algorithm: NSGAII
+n_trials: 100
+n_evaluations: 10000
+
+parameter_space:
+  offspring_population_size:
+    values: [50, 100, 150, 200]
+  
+  crossover:
+    type: sbx  # Only explore SBX crossover
+    probability: {min: 0.8, max: 1.0}
+    distribution_index: {min: 5.0, max: 100.0}
+  
+  mutation:
+    type: polynomial  # Only explore polynomial mutation
+    probability_factor: {min: 0.5, max: 2.0}
+    distribution_index: {min: 5.0, max: 100.0}
+```
+
+**Narrow Search Space** (`examples/tuning/configs/narrow_search.yaml`):
+```yaml
+algorithm: NSGAII
+n_trials: 50
+
+parameter_space:
+  crossover:
+    probability: {min: 0.9, max: 1.0}
+    distribution_index: {min: 10.0, max: 30.0}
+  mutation:
+    probability_factor: {min: 0.8, max: 1.2}
+```
+
+### Parallel Tuning
+
+For large-scale tuning experiments, use parallel execution with PostgreSQL:
+
+```bash
+# Launch 4 parallel workers
+python -m jmetal.tuning.cli.parallel --config examples/tuning/configs/default.yaml --workers 4
+
+# With progress observer
+python -m jmetal.tuning.cli.parallel --config my_config.yaml --workers 8 --observer progress
+```
+
+### Programmatic Usage with Custom Config
+
+```python
+from jmetal.tuning import tune, TuningConfig
+
+# Load configuration from YAML
+config = TuningConfig.from_yaml("my_config.yaml")
+
+# Run tuning with config
+result = tune(
+    algorithm=config.algorithm,
+    problems=config.get_problems_as_tuples(),
+    n_trials=config.n_trials,
+    n_evaluations=config.n_evaluations,
+    parameter_space=config.parameter_space,
+)
+
+print(f"Best score: {result.best_score}")
+print(f"Best params: {result.best_params}")
+```
+
+## Agents
+
+If you use AI assistants (e.g., Copilot, Codex) while working on this project, please follow the guidelines in [AGENTS.md](AGENTS.md).
+
+### Available Example Configurations
+
+| File | Description |
+|------|-------------|
+| `default.yaml` | Standard ZDT benchmark with 100 trials |
+| `quick_demo.yaml` | Fast demo with 20 trials, 2 problems |
+| `sbx_only.yaml` | SBX crossover + polynomial mutation only |
+| `narrow_search.yaml` | Narrow parameter ranges for fine-tuning |
+| `dtlz_3obj.yaml` | DTLZ problems with 3 objectives |
+
 ## Features
 The current release of jMetalPy (v1.9.0) contains the following components:
 

src/jmetalpy.egg-info/SOURCES.txt

@@ -68,7 +68,9 @@ src/jmetal/problem/singleobjective/knapsack.py
 src/jmetal/problem/singleobjective/tsp.py
 src/jmetal/problem/singleobjective/unconstrained.py
 src/jmetal/tuning/__init__.py
+src/jmetal/tuning/config.py
 src/jmetal/tuning/tuning.py
+src/jmetal/tuning/tuning_config.py
 src/jmetal/tuning/tuning_parallel.py
 src/jmetal/tuning/_legacy/__init__.py
 src/jmetal/tuning/_legacy/nsgaii_optuna_tuning.py
@@ -84,10 +86,6 @@ src/jmetal/tuning/algorithms/nsgaii.py
 src/jmetal/tuning/cli/__init__.py
 src/jmetal/tuning/cli/parallel.py
 src/jmetal/tuning/cli/sequential.py
-src/jmetal/tuning/config/__init__.py
-src/jmetal/tuning/config/defaults.py
-src/jmetal/tuning/config/paths.py
-src/jmetal/tuning/config/problems.py
 src/jmetal/tuning/metrics/__init__.py
 src/jmetal/tuning/metrics/indicators.py
 src/jmetal/tuning/metrics/reference_fronts.py