1

I am writing a probability model that will take any number of "Outcome"s, then when a number is rolled and passed into the model the correct "Outcome" will be returned;

Essentially the logic is a map of outcomes with indexes representing that outcomes particular weight.

Outcome One 25% Outcome Two 25% Outcome Three 50%

these values will translate to;

outcomes := make(map[int]Outcome)
outcomes[25] = Outcome{"Outcome One", 25}
outcomes[50] = Outcome{"Outcome One", 25}
outcomes[100] = Outcome{"Outcome One", 50}

I have a function that then takes an input lets say 10, and loops around outcomes until the index is larger than the input;

Expected

input: 10, output: Outcome{"Outcome One", 25}
input: 30, output: Outcome{"Outcome Two", 25}
input: 60, output: Outcome{"Outcome Two", 50}

However in my unit tests for an input of 10, I get a combination of "Outcome One" and "Outcome Two", I think the problem lies within my for loop.

ProbabilityMatrix_test.go

var outcome1 = Outcome{"Outcome One", 25}
var outcome2 = Outcome{"Outcome Two", 25}
var outcome3 = Outcome{"Outcome Three", 50}

probabilityMatrix := ProbabilityMatrix{}
probabilityMatrix.SetUp()

probabilityMatrix.AddOutcome(outcome1)
probabilityMatrix.AddOutcome(outcome2)
probabilityMatrix.AddOutcome(outcome3)

outcome := probabilityMatrix.RollA(10)

if outcome != outcome1 {
    t.Errorf("%s", probabilityMatrix.Delimiters)
    t.Errorf("incorrect outcome, got %s, expected %s", outcome.Name, outcome1.Name)
}

The below code returns Outcome One, around 75% of the time (correct), and Outcome Two 25%

package RealisticTemperatureGenerator

type Outcome struct {
    Name        string
    Probability int
}

type ProbabilityMatrix struct {
    Delimiters     map[int]Outcome
    DefaultOutcome Outcome
    Total          int
}

func (pm *ProbabilityMatrix) SetUp() {
    pm.Delimiters = make(map[int]Outcome)
    pm.Total = 0
}

func (pm *ProbabilityMatrix) AddOutcome(outcome Outcome) {

    pm.DefaultOutcome = outcome
    currentIndex := outcome.Probability + pm.Total
    pm.Delimiters[currentIndex] = outcome
    pm.Total = currentIndex
}

func (pm *ProbabilityMatrix) RollA(input int) Outcome {
  return pm.WalkDelimiters(input)
}

// Problem Possibly here
func (pm ProbabilityMatrix) WalkDelimiters(input int) Outcome {

  for key, _ := range pm.Delimiters {
      if pm.Delimiters[key].Probability >= input {
        return pm.Delimiters[key]
      }
  }
  return pm.DefaultOutcome
}
John Mack
  • 195
  • 1
  • 1
  • 6
  • Also see related: [Golang map in order range loop](https://stackoverflow.com/questions/39450120/golang-map-in-order-range-loop/39450454#39450454) – icza Jun 06 '18 at 09:34
  • And if you are intensive on look-ups & want the result really fast without iterating then you should add a key-value pair for each of the percentage values in the map. That should take care of it esp as your inputs are always integers. – Ravi R Jun 06 '18 at 09:34

1 Answers1

2

When looping over maps in golang the order of the returned elements is random. This is the reason for the inconsistent behavior. See "Iteration order" in the official blog: https://blog.golang.org/go-maps-in-action

If you want a stable order, you have to maintain the keys in another structure:

keys := []int{25, 50, 100}

for _, key := range keys {
    if pm.Delimiters[key].Probability >= input {
        return pm.Delimiters[key]
    }
}
mbuechmann
  • 5,413
  • 5
  • 27
  • 40