init: v1.0.0

2026-05-27 23:03:00 +08:00
commit 8d97f750eb
466 changed files with 80067 additions and 0 deletions
@@ -0,0 +1,142 @@
+package entropy
+
+import (
+	"fmt"
+	"math"
+)
+
+func MarkovTests(data []byte, verbose bool) float64 {
+	// 将data转换为0,1比特串表示
+	epsilon := make([]byte, 0, len(data)*8)
+	for _, b := range data {
+		epsilon = append(epsilon, (b>>7)&1)
+		epsilon = append(epsilon, (b>>6)&1)
+		epsilon = append(epsilon, (b>>5)&1)
+		epsilon = append(epsilon, (b>>4)&1)
+		epsilon = append(epsilon, (b>>3)&1)
+		epsilon = append(epsilon, (b>>2)&1)
+		epsilon = append(epsilon, (b>>1)&1)
+		epsilon = append(epsilon, (b>>0)&1)
+	}
+
+	// double markov_test(byte* data, long len, const int verbose, const char *label){
+	var i, C_0, C_1, C_00, C_10 int64
+	var H_min, tmp_min_entropy, P_0, P_1, P_00, P_01, P_10, P_11, entEst float64
+
+	C_0 = 0
+	C_00 = 0
+	C_10 = 0
+	dataLen := int64(len(epsilon))
+	// get counts for unconditional and transition probabilities
+	for i = 0; i < dataLen-1; i++ {
+		if epsilon[i] == 0 {
+			C_0++
+			if epsilon[i+1] == 0 {
+				C_00++
+			}
+		} else if epsilon[i+1] == 0 {
+			C_10++
+		}
+	}
+
+	//C_0 is now  the number of 0 bits from S[0] to S[len-2]
+
+	C_1 = dataLen - 1 - C_0 //C_1 is the number of 1 bits from S[0] to S[len-2]
+
+	//Note that P_X1 = C_X1 / C_X = (C_X - C_X0)/C_X = 1.0 - C_X0/C_X = 1.0 - P_X0
+	if C_0 > 0 {
+		P_00 = float64(C_00) / float64(C_0)
+		P_01 = 1.0 - P_00
+	} else {
+		P_00 = 0.0
+		P_01 = 0.0
+	}
+
+	if C_1 > 0 {
+		P_10 = float64(C_10) / float64(C_1)
+		P_11 = 1.0 - P_10
+	} else {
+		P_10 = 0.0
+		P_11 = 0.0
+	}
+
+	// account for the last symbol
+	if epsilon[dataLen-1] == 0 {
+		C_0++
+	}
+	//C_0 is now  the number of 0 bits from S[0] to S[len-1]
+
+	P_0 = float64(C_0) / float64(dataLen)
+	P_1 = 1.0 - P_0
+
+	if verbose {
+		fmt.Printf("P_0 = %.17g\n", P_0)
+		fmt.Printf("P_1 = %.17g\n", P_1)
+		fmt.Printf("P_{0,0} = %.17g\n", P_00)
+		fmt.Printf("P_{0,1} = %.17g\n", P_01)
+		fmt.Printf("P_{1,0} = %.17g\n", P_10)
+		fmt.Printf("P_{1,1} = %.17g\n", P_11)
+	}
+
+	H_min = 128.0
+
+	//In the next block, note that if P_0X > 0.0, then P_0 > 0.0
+	//and similarly if P_1X > 0.0, then P_1 > 0.0
+
+	// Sequence 00...0
+	if P_00 > 0.0 {
+		tmp_min_entropy = -math.Log2(P_0) - 127.0*math.Log2(P_00)
+		if tmp_min_entropy < H_min {
+			H_min = tmp_min_entropy
+		}
+	}
+
+	// Sequence 0101...01
+	if (P_01 > 0.0) && (P_10 > 0.0) {
+		tmp_min_entropy = -math.Log2(P_0) - 64.0*math.Log2(P_01) - 63.0*math.Log2(P_10)
+		if tmp_min_entropy < H_min {
+			H_min = tmp_min_entropy
+		}
+	}
+
+	// Sequence 011...1
+	if (P_01 > 0.0) && (P_11 > 0.0) {
+		tmp_min_entropy = -math.Log2(P_0) - math.Log2(P_01) - 126.0*math.Log2(P_11)
+		if tmp_min_entropy < H_min {
+			H_min = tmp_min_entropy
+		}
+	}
+
+	// Sequence 100...0
+	if (P_10 > 0.0) && (P_00 > 0.0) {
+		tmp_min_entropy = -math.Log2(P_1) - math.Log2(P_10) - 126.0*math.Log2(P_00)
+		if tmp_min_entropy < H_min {
+			H_min = tmp_min_entropy
+		}
+	}
+
+	// Sequence 1010...10
+	if (P_10 > 0.0) && (P_01 > 0.0) {
+		tmp_min_entropy = -math.Log2(P_1) - 64.0*math.Log2(P_10) - 63.0*math.Log2(P_01)
+		if tmp_min_entropy < H_min {
+			H_min = tmp_min_entropy
+		}
+	}
+
+	// Sequence 11...1
+	if P_11 > 0.0 {
+		tmp_min_entropy = -math.Log2(P_1) - 127.0*math.Log2(P_11)
+		if tmp_min_entropy < H_min {
+			H_min = tmp_min_entropy
+		}
+	}
+
+	entEst = math.Min(H_min/128.0, 1.0)
+
+	if verbose {
+		fmt.Printf("p_max = %.17g\n", math.Pow(2.0, -H_min))
+		fmt.Printf("min entropy = %.17g\n", entEst)
+	}
+
+	return entEst
+}