define step-size for parameter range

When creating experiment, how do I go about creating a parameter with a range but with discrete intervals / step-size?

ax_client.create_experiment(
    name="hartmann_test_experiment",
    parameters=[
        {
            "name": "x1",
            "type": "range",
            "bounds": [10, 1000], # with step size 10
            "value_type": "int",
            "log_scale": False,
        },
       {
            "name": "x2",
            "type": "range",
            "bounds": [0, 1000], # with step size 20
            "value_type": "int",
            "log_scale": False,
        },
        ...

[danielcohenlive]

Thanks for the question, @hududed! It looks like we don’t have that feature currently, but I can offer 2 workarounds:

1. Use an integer range parameter and multiply the value by a constant on evaluation
2. Use a choice parameter and set is_ordered to True Hope that helps!

[Balandat]

Use a choice parameter and set is_ordered to True

I would only recommend this if there are relatively few values and if the values are unevenly spaced. In all other cases you should use an integer parameter and scale things up to the right multiple on your end of the code. This will make it easier for Ax to build a good surrogate and optimize the problem well.