add examples

Author: codelion

examples/algotune/affine_transform_2d/config.yaml

@@ -0,0 +1,110 @@
+# Configuration for affine_transform_2d task - Optimized Gemini Flash 2.5
+# Achieved 1.64x AlgoTune Score with these settings
+
+# General settings
+max_iterations: 100
+checkpoint_interval: 10
+log_level: "INFO"
+random_seed: 42
+diff_based_evolution: true  # Best for Gemini models
+max_code_length: 10000
+
+# LLM Configuration
+llm:
+  api_base: "https://openrouter.ai/api/v1"
+  models:
+    - name: "openai/o4-mini"
+      weight: 1.0
+  
+  temperature: 0.4  # Optimal (better than 0.2, 0.6, 0.8)
+  max_tokens: 16000  # Optimal context
+  timeout: 150
+  retries: 3
+
+# Prompt Configuration - Optimal settings
+prompt:
+  system_message: |
+    SETTING:
+    You're an autonomous programmer tasked with solving a specific problem. You are to use the commands defined below to accomplish this task. Every message you send incurs a cost—you will be informed of your usage and remaining budget by the system.
+    You will be evaluated based on the best-performing piece of code you produce, even if the final code doesn't work or compile (as long as it worked at some point and achieved a score, you will be eligible).
+    Apart from the default Python packages, you have access to the following additional packages:
+     - cryptography
+     - cvxpy
+     - cython
+     - dace
+     - dask
+     - diffrax
+     - ecos
+     - faiss-cpu
+     - hdbscan
+     - highspy
+     - jax
+     - networkx
+     - numba
+     - numpy
+     - ortools
+     - pandas
+     - pot
+     - psutil
+     - pulp
+     - pyomo
+     - python-sat
+     - pythran
+     - scikit-learn
+     - scipy
+     - sympy
+     - torch
+    Your primary objective is to optimize the `solve` function to run as as fast as possible, while returning the optimal solution.
+    You will receive better scores the quicker your solution runs, and you will be penalized for exceeding the time limit or returning non-optimal solutions.
+
+    Below you find the description of the task you will have to solve. Read it carefully and understand what the problem is and what your solver should do.
+
+    You are an expert programmer specializing in signal_processing algorithms. Your task is to improve the affine_transform_2d algorithm implementation with baseline comparison.
+
+    The problem description is:
+    2D Affine Transform
+
+    Apply a 2D affine transformation to an input image (2D array). The transformation is defined by a 2x3 matrix which combines rotation, scaling, shearing, and translation. This task uses cubic spline interpolation (order=3) and handles boundary conditions using the 'constant' mode (padding with 0).
+
+    Focus on improving the solve method to correctly handle the input format and produce valid solutions efficiently. Your solution will be compared against the reference AlgoTune baseline implementation to measure speedup and correctness.
+  num_top_programs: 3      # Best balance
+  num_diverse_programs: 2  # Best balance
+  include_artifacts: true  # +20.7% improvement
+
+# Database Configuration
+database:
+  population_size: 1000
+  archive_size: 100
+  num_islands: 4
+  
+  # Selection parameters - Optimal ratios
+  elite_selection_ratio: 0.1   # 10% elite
+  exploration_ratio: 0.3       # 30% exploration
+  exploitation_ratio: 0.6      # 60% exploitation
+  
+  # NO feature_dimensions - let it use defaults based on evaluator metrics
+  feature_bins: 10
+  
+  # Migration parameters
+  migration_interval: 20
+  migration_rate: 0.1  # Better than 0.2
+
+# Evaluator Configuration
+evaluator:
+  timeout: 200
+  max_retries: 3
+  
+  # Cascade evaluation
+  cascade_evaluation: true
+  cascade_thresholds: [0.5, 0.8]
+  
+  # Parallel evaluations
+  parallel_evaluations: 1
+
+# AlgoTune task-specific configuration
+algotune:
+  num_trials: 5
+  data_size: 100
+  timeout: 300
+  num_runs: 3
+  warmup_runs: 1