Method midpoint32

coder/arith/encoding.go:12–14

(lo, hi uint32)

func NewEncoder() *Encoder { return &Encoder{lo: 0, hi: ^uint32(0)} }

func (p P) midpoint32(lo, hi uint32) uint32 {
	return lo + uint32(uint64(hi-lo)*uint64(p)>>probBits)
}

func (enc *Encoder) Encode(bit uint, prob P) {
	x := prob.midpoint32(enc.lo, enc.hi)