Adding random seed reset to improve stability of tests

mldrugdiscovery · Feb 1, 2018 · dd4978b · dd4978b
1 parent ce1061f
commit dd4978b
Show file tree

Hide file tree

Showing 6 changed files with 35 additions and 16 deletions.
diff --git a/similarity_search/apps/tune_vptree.cc b/similarity_search/apps/tune_vptree.cc
@@ -113,7 +113,7 @@ void RunExper(unsigned AddRestartQty,
 
   LOG(LIB_INFO) << "We are going to tune parameters for " << MethodName;
 
-  static  thread_local auto&          engine(GET_RANDOM_GENERATOR);
+  static  thread_local auto&          engine(getRandomGenerator());
   static  std::normal_distribution<>  normGen(0.0f, log(FullFactor));
 
   AnyParamManager pmgr(IndexParams);

diff --git a/similarity_search/include/utils.h b/similarity_search/include/utils.h
@@ -33,6 +33,7 @@
 #include <random>
 #include <climits>
 #include <stdexcept>
+#include <memory>
 
 #include "idtype.h"
 
@@ -65,7 +66,6 @@ typedef SSIZE_T ssize_t;
 
 #define FIELD_DELIMITER ':'
 
-#define GET_RANDOM_GENERATOR    getRandomGenerator<std::mt19937>()
 
 namespace similarity {
 
@@ -75,6 +75,8 @@ using std::stringstream;
 
 using namespace std;
 
+typedef std::mt19937 RandomGeneratorType;
+
 
 const char* GetFileName(const char* fullpath);
 
@@ -84,18 +86,29 @@ bool DoesFileExist(const char *filename);
 
 inline bool DoesFileExist(const string &filename) { return DoesFileExist(filename.c_str()); }
 
-extern int randomSeed;
 
 /*
- * Random number generation is thread safe when respective
- * objects are not shared among threads. So, we will keep one
- * random number generator per thread.
+ * 1. Random number generation is thread safe when respective
+ *    objects are not shared among threads. So, we will keep one
+ *    random number generator per thread.
+ * 2. There is a default seed to initialize all random generators.
+ * 3. However, sometimes we may want to reset the random number generator
+ *    within a working thread (i.e., this would be only a thread-specific change).
+ *    In particular, this is needed to improve reproducibility of integration tests.
  */ 
-template <class RandGenType>
-inline RandGenType & getRandomGenerator() {
-    static thread_local RandGenType gen(randomSeed);
+extern int                                                defaultRandomSeed;
+extern thread_local std::unique_ptr<RandomGeneratorType>  randomGen;
+
+inline void resetRandomGenerator(int newRandomSeed) {
+  randomGen.reset(new RandomGeneratorType(newRandomSeed));
+}
+
+inline RandomGeneratorType& getRandomGenerator() {
+    if (!randomGen) {
+      resetRandomGenerator(defaultRandomSeed);
+    }
 
-    return gen;
+    return *randomGen;
 }
 
 // random 32-bit integer number
@@ -108,7 +121,7 @@ inline int32_t RandomInt() {
     // thread_local is static by default, but let's keep it static for clarity
     static thread_local std::uniform_int_distribution<int32_t> distr(0, std::numeric_limits<int32_t>::max());
 
-    return distr(GET_RANDOM_GENERATOR); 
+    return distr(getRandomGenerator()); 
 }
 
 template <class T>
@@ -122,7 +135,7 @@ inline T RandomReal() {
     // thread_local is static by default, but let's keep it static for clarity
     static thread_local std::uniform_real_distribution<T> distr(0, 1);
 
-    return distr(GET_RANDOM_GENERATOR); 
+    return distr(getRandomGenerator()); 
 }
 
 void RStrip(char* str);

diff --git a/similarity_search/src/init.cc b/similarity_search/src/init.cc
@@ -29,13 +29,15 @@
 #include "logging.h"
 
 #include <iostream>
+#include <memory>
 
 namespace similarity {
 
-int randomSeed = 0;
+int                                                 defaultRandomSeed = 0;
+thread_local std::unique_ptr<RandomGeneratorType>   randomGen;
 
 void initLibrary(int seed, LogChoice choice, const char* pLogFile) {
-  randomSeed = seed;
+  defaultRandomSeed = seed;
 
   std::ios_base::sync_with_stdio(false);
   InitializeLogger(choice, pLogFile);

diff --git a/similarity_search/src/randproj_util.cc b/similarity_search/src/randproj_util.cc
@@ -33,7 +33,7 @@ template <class dist_t> void initRandProj(size_t nSrcDim, size_t nDstDim,
                                          bool bDoOrth,
                                          vector<vector<dist_t>>& projMatr) {
   // Static is thread-safe in C++-11
-  static thread_local auto&           randGen(GET_RANDOM_GENERATOR);
+  static thread_local auto&           randGen(getRandomGenerator());
   static  std::normal_distribution<>  normGen(0.0f, 1.0f);
 
   // 1. Create normally distributed vectors

diff --git a/similarity_search/src/searchoracle.cc b/similarity_search/src/searchoracle.cc
@@ -206,7 +206,7 @@ void PolynomialPruner<dist_t>::SetIndexTimeParams(AnyParamManager& pmgr) {
     unsigned exp_left = 0, exp_right = 0;
 
 
-    static thread_local auto&           randGen(GET_RANDOM_GENERATOR);
+    static thread_local auto&           randGen(getRandomGenerator());
     static  std::normal_distribution<>  normGen(0.0f, log(fullFactor));
 
 

diff --git a/similarity_search/test/test_integr_util.h b/similarity_search/test/test_integr_util.h
@@ -337,6 +337,10 @@ size_t RunTestExper(const vector<MethodTestCase>& vTestCases,
     }
   }
 
+  // For better reproducibility, let's reset
+  // random number generators. 
+  defaultRandomSeed = 0; // Will affect any new threads
+  resetRandomGenerator(defaultRandomSeed); // Affects only the current thread
 
   for (int TestSetId = 0; TestSetId < config.GetTestSetToRunQty(); ++TestSetId) {
     config.SelectTestSet(TestSetId);