ax.dev, GitHub readme, and loop tutorial examples are ignoring minimize kwarg

[sgbaird]

@jwatchorn pointed out to me that the example from https://ax.dev/, the GitHub readme, and the Loop API tutorial are ignoring the minimize kwarg and maximizing it instead (note these are loop API examples). I assume this is related to the deprecation of the minimize kwarg in the context of the Service API.

See this reproducer: https://colab.research.google.com/drive/1iEUgDgFO7K_B6fYSSf9YSFOacCaKvk7u?usp=sharing

from ax import optimize
best_parameters, best_values, experiment, model = optimize(
      parameters=[
        {
          "name": "x1",
          "type": "range",
          "bounds": [-10.0, 10.0],
        },
        {
          "name": "x2",
          "type": "range",
          "bounds": [-10.0, 10.0],
        },
      ],
      # Booth function
      evaluation_function=lambda p: (p["x1"] + 2*p["x2"] - 7)**2 + (2*p["x1"] + p["x2"] - 5)**2,
      minimize=True,
  )
print(best_parameters)
# {'x1': -10.0, 'x2': -10.0}
print(best_values)
# ({'objective': 2593.8262505563807}, {'objective': {'objective': 10.51208881969219}})

[bernardbeckerman]

Thanks for this great writeup with repro as usual @sgbaird! Taking a look now.

[bernardbeckerman]

This is a bug! And the fix is up : ) (link) Thanks for this detailed report - made it super easy to find + fix.