CurrinExp Prior

In this example we show how to use HyperMapper's prior-guided pptimization to optimize a CurrinExp function. We look for minimizing the value of this function given two parameters x1, x2.

The Objective Function

The CurrinExp objective function can be defined as:

factor1 = 1 - math.exp(-1/(2*x2))
factor2 = 2300*x1*x1*x1 + 1900*x1*x1 + 2092*x1 + 60
factor3 = 100*x1*x1*x1 + 500*x1*x1 + 4*x1 + 20
y_value = factor1*factor2/factor3
y_energy = x1 + x2

An example of this code can be found in prior_currinexp.py.

The JSON Configuration File

In this example, we will inject priors into HyperMapper's optimization. Since the CurrinExp function has its optimum at X = [0, 1], we use decay and exponential priors for the input parameters. The json configuration file for this example is:

{
    "application_name": "currinexp",
    "optimization_objectives": ["Value"],
    "optimization_method": "prior_guided_optimization",
    "optimization_iterations": 20,
    "design_of_experiment":{
        "number_of_samples": 3
    },
    "input_parameters" : {
        "x1": {
            "parameter_type" : "real",
            "values" : [0, 1],
            "prior" : "decay"
        },
        "x2": {
            "parameter_type" : "real",
            "values" : [0.00001, 1],
            "prior" : "exponential"
        }
    }
}

Note that we add two new fields to the json:

• We tell HyperMapper to use the prior-guided optimization method with "optimization_method": "prior_guided_optimization"
• We tell HyperMapper the priors for the input parameters with "prior": "decay" and "prior": "exponential".

You can find this json in currinexp_scenario.json. See here for more information on how to use prior-guided optimization and the supported prior types.

Run HyperMapper

Remember to add the scripts folder to your $PYTHONPATH. In order to run this example, we use:

cd $HYPERMAPPER_HOME
python3 example_scenarios/synthetic/currinexp/currinexp.py

An example of stdout output can be found here.

The result of this script is a csv file called currinexp_output_dse_samples.csv.