more stable sampling K from N

include/LightGBM/utils/random.h

@@ -5,6 +5,7 @@
 
 #include <random>
 #include <vector>
+#include <set>
 
 namespace LightGBM {
 
@@ -65,13 +66,13 @@ public:
   inline std::vector<int> Sample(int N, int K) {
     std::vector<int> ret;
     ret.reserve(K);
-    if (K > N || K < 0) {
+    if (K > N || K <= 0) {
       return ret;
     } else if (K == N) {
       for (int i = 0; i < N; ++i) {
         ret.push_back(i);
       }
-    } else if (K > N / 2) {
+    } else if (K > 1 && K > (N / std::log2(K))) {
       for (int i = 0; i < N; ++i) {
         double prob = (K - ret.size()) / static_cast<double>(N - i);
         if (NextFloat() < prob) {
@@ -79,15 +80,15 @@ public:
         }
       }
     } else {
-      int min_step = 1;
-      int avg_step = N / K;
-      int max_step = 2 * avg_step - min_step;
-      int start = -1;
-      for (int i = 0; i < K; ++i) {
-        int step = NextShort(min_step, max_step + 1);
-        start += step;
-        if (start >= N) { break; }
-        ret.push_back(start);
+      std::set<int> sample_set;
+      while (sample_set.size() < K) {
+        int next = RandInt32() % N;
+        if (sample_set.count(next) == 0) {
+          sample_set.insert(next);
+        }
+      }
+      for (auto iter = sample_set.begin(); iter != sample_set.end(); ++iter) {
+        ret.push_back(*iter);
       }
     }
     return ret;