lotus/chain/messagepool/block_proba.go
Jakub Sztandera f35555964d
Better "optimal selection
Signed-off-by: Jakub Sztandera <kubuxu@protocol.ai>
2020-08-13 14:08:10 +02:00

81 lines
1.5 KiB
Go

package messagepool
import (
"math"
"sync"
)
var noWinnersProbCache []float64
var noWinnersProbOnce sync.Once
func noWinnersProb() []float64 {
noWinnersProbOnce.Do(func() {
poissPdf := func(x float64) float64 {
const Mu = 5
lg, _ := math.Lgamma(x + 1)
result := math.Exp((math.Log(Mu) * x) - lg - Mu)
return result
}
out := make([]float64, 0, MaxBlocks)
for i := 0; i < MaxBlocks; i++ {
out = append(out, poissPdf(float64(i)))
}
noWinnersProbCache = out
})
return noWinnersProbCache
}
func binomialCoefficient(n, k float64) float64 {
if k > n {
return math.NaN()
}
r := 1.0
for d := 1.0; d <= k; d++ {
r *= n
r /= d
n -= 1
}
return r
}
func (mp *MessagePool) blockProbabilities(tq float64) []float64 {
noWinners := noWinnersProb() // cache this
p := 1 - tq
binoPdf := func(x, trials float64) float64 {
// based on https://github.com/atgjack/prob
if x > trials {
return 0
}
if p == 0 {
if x == 0 {
return 1.0
}
return 0.0
}
if p == 1 {
if x == trials {
return 1.0
}
return 0.0
}
coef := binomialCoefficient(trials, x)
pow := math.Pow(p, x) * math.Pow(1-p, trials-x)
if math.IsInf(coef, 0) {
return 0
}
return coef * pow
}
out := make([]float64, 0, MaxBlocks)
for place := 0; place < MaxBlocks; place++ {
var pPlace float64
for otherWinners, pCase := range noWinners {
pPlace += pCase * binoPdf(float64(place), float64(otherWinners+1))
}
out = append(out, pPlace)
}
return out
}