From 57c573975353fd6d84d80e8479e514cc1e2248cd Mon Sep 17 00:00:00 2001
From: Gheorghita Mutu <gheorghitamutu@gmail.com>
Date: Sat, 31 Aug 2024 22:39:58 +0300
Subject: [PATCH] [Entropy] + add Renyi entropy (untested) (2)

---
 GViewCore/src/Entropy/Entropy.cpp | 44 ++++++++++++++++++++++++++-----
 1 file changed, 37 insertions(+), 7 deletions(-)

diff --git a/GViewCore/src/Entropy/Entropy.cpp b/GViewCore/src/Entropy/Entropy.cpp
index 31c6b12c..f794efe7 100644
--- a/GViewCore/src/Entropy/Entropy.cpp
+++ b/GViewCore/src/Entropy/Entropy.cpp
@@ -2,28 +2,58 @@
 
 #include <math.h>
 
+constexpr uint32 MAX_NUMBER_OF_BYTES = 256;
+
 namespace GView::Entropy
 {
-double ShannonEntropy(const BufferView& buffer)
+void SetFrequencies(const BufferView& buffer, std::array<char, MAX_NUMBER_OF_BYTES>& frequency)
 {
-    char frequency[256]{};
-
     // Count frequency of each byte in the buffer
     for (uint32 i = 0; i < buffer.GetLength(); i++) {
         const auto c = buffer[i];
         frequency[c]++;
     }
+}
 
-    // Calculate entropy
+double ShannonEntropy_private(const BufferView& buffer, std::array<char, MAX_NUMBER_OF_BYTES>& frequency)
+{
     double entropy = 0.0;
-    for (const auto& value : frequency) {
-        if (value == 0) {
+    for (auto f : frequency) {
+        if (f == 0) {
             continue;
         }
-        double probability = static_cast<double>(value) / buffer.GetLength();
+        double probability = static_cast<double>(f) / buffer.GetLength();
         entropy -= probability * log2(probability);
     }
 
     return entropy; // max log2(n) = 8
 }
+
+double ShannonEntropy(const BufferView& buffer)
+{
+    std::array<char, MAX_NUMBER_OF_BYTES> frequency{};
+    SetFrequencies(buffer, frequency);
+    return ShannonEntropy_private(buffer, frequency);
+}
+
+double RenyiEntropy(const BufferView& buffer, double alpha)
+{
+    std::array<char, MAX_NUMBER_OF_BYTES> frequency{};
+    SetFrequencies(buffer, frequency);
+
+    if (alpha == 1.0) {
+        return ShannonEntropy_private(buffer, frequency);
+    }
+
+    double sum = 0.0;
+    for (auto f : frequency) {
+        double probability = static_cast<double>(f) / buffer.GetLength();
+        if (probability > 0) {
+            sum += pow(probability, alpha);
+        }
+    }
+
+    // Convert to bits if using log base e
+    return ((1.0 / (1.0 - alpha)) * log(sum)) / log(2);
+}
 } // namespace GView::Entropy