"""
Minimal example: Physics-Informed Bayesian Optimization in ~30 lines.

Demonstrates the core workflow:
1. Define a physics model
2. Define the parameter space
3. Create a designer and suggest experiments
"""

import torch
from physics_informed_bo.experiment.parameter_space import ParameterSpace
from physics_informed_bo.experiment.designer import ExperimentDesigner


# Physics model: simple Arrhenius-like kinetics
def physics_model(X):
    temp, conc = X[:, 0], X[:, 1]
    return torch.exp(-5000 / temp) * conc**0.5


# Parameter space
space = ParameterSpace()
space.add_continuous("temperature", 300, 600, units="K")
space.add_continuous("concentration", 0.1, 10.0, units="mol/L")

# Some initial observations
X_init = torch.tensor([[350.0, 1.0], [450.0, 5.0], [500.0, 3.0]])
y_init = torch.tensor([0.1, 0.8, 0.6])

# Create designer with physics + data
designer = ExperimentDesigner(
    parameter_space=space,
    physics_fn=physics_model,
    initial_data=(X_init, y_init),
)

# Suggest next experiments
next_experiments = designer.suggest(n=3)
print("Suggested experiments:")
for exp in space.to_dict(next_experiments):
    print(f"  Temperature={exp['temperature']:.1f} K, Concentration={exp['concentration']:.2f} mol/L")

# After running experiments, update the designer
# designer.update(X_new, y_new)