params.go

package bayesopt

import (
	"math"
	"math/rand"
)

// SampleTries is the number of tries a sample function should try before
// truncating the samples to the boundaries.
var SampleTries = 1000

// Param represents a parameter that can be optimized.
type Param interface {
	// GetName returns the name of the parameter.
	GetName() string
	// GetMax returns the maximum value.
	GetMax() float64
	// GetMin returns the minimum value.
	GetMin() float64
	// Sample returns a random point within the bounds. It doesn't have to be
	// uniformly distributed.
	Sample() float64
}

var _ Param = UniformParam{}

// LinearParam is a UniformParam. Deprecated.
type LinearParam = UniformParam

// UniformParam is a uniformly distributed parameter between Max and Min.
type UniformParam struct {
	Name     string
	Max, Min float64
}

// GetName implements Param.
func (p UniformParam) GetName() string {
	return p.Name
}

// GetMax implements Param.
func (p UniformParam) GetMax() float64 {
	return p.Max
}

// GetMin implements Param.
func (p UniformParam) GetMin() float64 {
	return p.Min
}

// Sample implements Param.
func (p UniformParam) Sample() float64 {
	return rand.Float64()*(p.Max-p.Min) + p.Min
}

var _ Param = NormalParam{}

// NormalParam is a normally distributed parameter with Mean and StdDev.
// The Max and Min parameters use discard sampling to find a point between them.
// Set them to be math.Inf(1) and math.Inf(-1) to disable the bounds.
type NormalParam struct {
	Name         string
	Max, Min     float64
	Mean, StdDev float64
}

// GetName implements Param.
func (p NormalParam) GetName() string {
	return p.Name
}

// GetMax implements Param.
func (p NormalParam) GetMax() float64 {
	return p.Max
}

// GetMin implements Param.
func (p NormalParam) GetMin() float64 {
	return p.Min
}

// Sample implements Param.
func (p NormalParam) Sample() float64 {
	return truncateSample(p, func() float64 {
		return rand.NormFloat64()*p.StdDev + p.Mean
	})
}

var _ Param = ExponentialParam{}

// ExponentialParam is an exponentially distributed parameter between 0 and in
// the range (0, +math.MaxFloat64] whose rate parameter (lambda) is Rate and
// whose mean is 1/lambda (1).
// The Max and Min parameters use discard sampling to find a point between them.
// Set them to be math.Inf(1) and math.Inf(-1) to disable the bounds.
type ExponentialParam struct {
	Name     string
	Max, Min float64
	Rate     float64
}

// GetName implements Param.
func (p ExponentialParam) GetName() string {
	return p.Name
}

// GetMax implements Param.
func (p ExponentialParam) GetMax() float64 {
	return p.Max
}

// GetMin implements Param.
func (p ExponentialParam) GetMin() float64 {
	return p.Min
}

// Sample implements Param.
func (p ExponentialParam) Sample() float64 {
	return truncateSample(p, func() float64 {
		return rand.ExpFloat64() / p.Rate
	})
}

func truncateSample(p Param, f func() float64) float64 {
	max := p.GetMax()
	min := p.GetMin()

	var sample float64
	for i := 0; i < SampleTries; i++ {
		sample = f()
		if sample >= min && sample <= max {
			return sample
		}
	}
	return math.Min(math.Max(sample, min), max)
}

// RejectionParam samples from Param and then uses F to decide whether or not to
// reject the sample. This is typically used with a UniformParam. F should
// output a value between 0 and 1 indicating the proportion of samples this
// point should be accepted. If F always outputs 0, Sample will get stuck in an
// infinite loop.
type RejectionParam struct {
	Param

	F func(x float64) float64
}

// Sample implements Param.
func (p RejectionParam) Sample() float64 {
	for {
		x := p.Param.Sample()
		y := p.F(x)
		if rand.Float64() < y {
			return x
		}
	}
}