optimize-function

Optimize a Function — hermes-lab Quick-Start Example

Optimize the Rosenbrock function f(x,y) = (1-x)² + 100·(y-x²)² using hermes-lab burst mode. No API keys, no external agents — runs entirely locally in under 2 minutes.

The global minimum is 0 at (x=1, y=1). The lab maximizes the negated value, so the best possible score is 0.

Files

• search_space.json — parameter bounds for x ∈ [-5, 5] and y ∈ [-5, 5]
• evaluate.py — reads train_config.json, evaluates Rosenbrock, writes metrics.json
• spec.yaml — experiment specification for labctl create

Run it (from repo root)

# 1. Point the lab at a scratch data directory
export HERMES_LAB_DATA_ROOT=./demo-data

# 2. Initialize the lab
python3 scripts/labctl.py init

# 3. Register the experiment
python3 scripts/labctl.py create examples/optimize-function/spec.yaml

# 4. Run 20 random-search iterations
python3 scripts/labctl.py burst optimize-rosenbrock --strategy random -n 20

# 5. Check results
python3 scripts/labctl.py status

After the burst you will see the best score found and its (x, y) values. The closer the score is to 0 the better — that means x ≈ 1 and y ≈ 1.

Try other strategies

# Bayesian optimization (if optuna is installed)
python3 scripts/labctl.py burst optimize-rosenbrock --strategy bayesian -n 30

# More iterations
python3 scripts/labctl.py burst optimize-rosenbrock --strategy random -n 100

Clean up

rm -rf ./demo-data