nearly full reproducibility with a given seed #57

mldrugdiscovery · Jan 7, 2018 · c409c1d · c409c1d
1 parent b47b22a
commit c409c1d
Show file tree

Hide file tree

Showing 17 changed files with 39 additions and 32 deletions.
diff --git a/python_bindings/nmslib.cc b/python_bindings/nmslib.cc
@@ -358,7 +358,7 @@ PYBIND11_PLUGIN(nmslib) {
   py::module nmslibLogger = logging.attr("getLogger")("nmslib");
   setGlobalLogger(new PythonLogger(nmslibLogger));
 
-  initLibrary(LIB_LOGCUSTOM, NULL);
+  initLibrary(0 /* seed */, LIB_LOGCUSTOM, NULL);
 
   py::module m(module_name, "Bindings for Non-Metric Space Library (NMSLIB)");
 

diff --git a/scripts/test_run.sh b/scripts/test_run.sh
@@ -114,11 +114,7 @@ function do_run {
 # Methods that may create an index (at least for some spaces)
 do_run 0 "napp" " -c numPivot=512,numPivotIndex=64 " 0 "-t numPivotSearch=40 -t numPivotSearch=42 -t numPivotSearch=44 -t numPivotSearch=46 -t numPivotSearch=48" "napp_${SPACE}.index"
 do_run 1 "sw-graph" " -c NN=10 " 0 " -t efSearch=10 -t efSearch=20 -t efSearch=40 -t efSearch=80 -t efSearch=160 -t efSearch=240" "sw-graph_${SPACE}.index"
-if [ "$SPACE" = "l2" -o "$SPACE" = "cosinesimil" ] ; then 
-  do_run 1 "hnsw"     " -c M=10 " 1 " -t efSearch=10 -t efSearch=20 -t efSearch=40 -t efSearch=80 -t efSearch=160 -t efSearch=240" "hnsw_${SPACE}.index"
-else
-  do_run 1 "hnsw"     " -c M=10 " 0 " -t efSearch=10 -t efSearch=20 -t efSearch=40 -t efSearch=80 -t efSearch=160 -t efSearch=240" 
-fi
+do_run 1 "hnsw"     " -c M=10 " 1 " -t efSearch=10 -t efSearch=20 -t efSearch=40 -t efSearch=80 -t efSearch=160 -t efSearch=240" "hnsw_${SPACE}.index"
 
 # Methods that do not support creation of an index
 do_run 1 "vptree" " -c tuneK=$K,bucketSize=50,desiredRecall=0.99,chunkBucket=1   0 " ""

diff --git a/similarity_search/apps/bench_distfunc.cc b/similarity_search/apps/bench_distfunc.cc
@@ -1971,7 +1971,7 @@ using namespace similarity;
 int main(int argc, char* argv[]) {
     string LogFile;
     if (argc == 2) LogFile = argv[1];
-    initLibrary(LogFile.empty() ? LIB_LOGSTDERR:LIB_LOGFILE, LogFile.c_str());
+    initLibrary(0, LogFile.empty() ? LIB_LOGSTDERR:LIB_LOGFILE, LogFile.c_str());
 
     int nTest  = 0;
 

diff --git a/similarity_search/apps/bench_projection.cc b/similarity_search/apps/bench_projection.cc
@@ -250,7 +250,7 @@ int main(int argc, char *argv[]) {
     LOG(LIB_FATAL) << "Failed to parse cmd arguments";
   }
 
-  initLibrary(logFile.empty() ? LIB_LOGSTDERR:LIB_LOGFILE, logFile.c_str());
+  initLibrary(0, logFile.empty() ? LIB_LOGSTDERR:LIB_LOGFILE, logFile.c_str());
 
   LOG(LIB_INFO) << "Program arguments are processed";
 

diff --git a/similarity_search/apps/knn_stat.cc b/similarity_search/apps/knn_stat.cc
@@ -377,7 +377,7 @@ int main(int argc, char *argv[]) {
     LOG(LIB_FATAL) << "Failed to parse cmd arguments";
   }
 
-  initLibrary(logFile.empty() ? LIB_LOGSTDERR:LIB_LOGFILE, logFile.c_str());
+  initLibrary(0, logFile.empty() ? LIB_LOGSTDERR:LIB_LOGFILE, logFile.c_str());
 
   LOG(LIB_INFO) << "Program arguments are processed";
 

diff --git a/similarity_search/apps/main.cc b/similarity_search/apps/main.cc
@@ -566,7 +566,7 @@ int main(int ac, char* av[]) {
                         "then you have to specify the gold-standard cache file!");
     }
 
-    initLibrary(LogFile.empty() ? LIB_LOGSTDERR:LIB_LOGFILE, LogFile.c_str());
+    initLibrary(0, LogFile.empty() ? LIB_LOGSTDERR:LIB_LOGFILE, LogFile.c_str());
 
     LOG(LIB_INFO) << "Program arguments are processed";
 

diff --git a/similarity_search/apps/report_intr_dim.cc b/similarity_search/apps/report_intr_dim.cc
@@ -137,7 +137,7 @@ int main(int argc, char* argv[]) {
       THROW_RUNTIME_ERR(err);
     }
 
-    initLibrary(LIB_LOGSTDERR);
+    initLibrary(0, LIB_LOGSTDERR);
 
     if (DIST_TYPE_INT == distType) {
       TestSpace<int>(

diff --git a/similarity_search/apps/test_clust.cc b/similarity_search/apps/test_clust.cc
@@ -290,7 +290,7 @@ int main(int argc, char* argv[]) {
                                 SampleDistQty
   );
 
-  initLibrary(LogFile.empty() ? LIB_LOGSTDERR:LIB_LOGFILE, LogFile.c_str());
+  initLibrary(0, LogFile.empty() ? LIB_LOGSTDERR:LIB_LOGFILE, LogFile.c_str());
 
   ToLower(DistType);
   ToLower(ClustType);

diff --git a/similarity_search/apps/tune_vptree.cc b/similarity_search/apps/tune_vptree.cc
@@ -474,7 +474,7 @@ int main(int ac, char* av[]) {
                        IndexParams,
                        QueryTimeParams);
 
-  initLibrary(LogFile.empty() ? LIB_LOGSTDERR:LIB_LOGFILE, LogFile.c_str());
+  initLibrary(0, LogFile.empty() ? LIB_LOGSTDERR:LIB_LOGFILE, LogFile.c_str());
 
   ToLower(DistType);
 

diff --git a/similarity_search/include/init.h b/similarity_search/include/init.h
@@ -26,7 +26,7 @@
 
 namespace similarity {
 
-  void initLibrary(LogChoice choice = LIB_LOGNONE, const char*pLogFile = NULL);
+  void initLibrary(int seed = 0, LogChoice choice = LIB_LOGNONE, const char*pLogFile = NULL);
 }
 
 #endif
diff --git a/similarity_search/include/method/hnsw.h b/similarity_search/include/method/hnsw.h
@@ -481,8 +481,8 @@ namespace similarity {
 
         int getRandomLevel(double revSize)
         {
-            std::uniform_real_distribution<double> distribution(0.0, 1.0);
-            double r = -log(distribution(*generator)) * revSize;
+            // RandomReal is thread-safe
+            float r = -log(RandomReal<float>()) * revSize;
             return (int)r;
         }
 
@@ -513,7 +513,6 @@ namespace similarity {
 
         //
     private:
-        std::unique_ptr<std::default_random_engine> generator;
         size_t M_;
         size_t maxM_;
         size_t maxM0_;

diff --git a/similarity_search/include/utils.h b/similarity_search/include/utils.h
@@ -84,21 +84,32 @@ bool DoesFileExist(const char *filename);
 
 inline bool DoesFileExist(const string &filename) { return DoesFileExist(filename.c_str()); }
 
-inline int RandomInt() {
-    // Static is thread-safe in C++ 11
-    static random_device rdev;
-    static mt19937 gen(rdev());
-    static std::uniform_int_distribution<int> distr(0, std::numeric_limits<int>::max());
+extern int randomSeed;
+// random 32-bit integer number
+inline int32_t RandomInt() {
+    /*
+     * Random number generation is thread safe when respective
+     * objects are not shared among threads. So, we will keep one
+     * random number generator per thread.
+    */ 
+    // thread_local is static by default, but let's keep it static for clarity
+    static thread_local mt19937 gen(randomSeed);
+    static thread_local std::uniform_int_distribution<int32_t> distr(0, std::numeric_limits<int32_t>::max());
 
     return distr(gen); 
 }
 
 template <class T>
+// random real number from 0 (inclusive) to 1 (exclusive)
 inline T RandomReal() {
-    // Static is thread-safe in C++ 11
-    static random_device rdev;
-    static mt19937 gen(rdev());
-    static std::uniform_real_distribution<T> distr(0, 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.
+    */ 
+    // thread_local is static by default, but let's keep it static for clarity
+    static thread_local mt19937 gen(randomSeed);
+    static thread_local std::uniform_real_distribution<T> distr(0, 1);
 
     return distr(gen); 
 }

diff --git a/similarity_search/src/init.cc b/similarity_search/src/init.cc
@@ -32,7 +32,11 @@
 
 namespace similarity {
 
-void initLibrary(LogChoice choice, const char* pLogFile) {
+int randomSeed = 0;
+
+void initLibrary(int seed, LogChoice choice, const char* pLogFile) {
+  randomSeed = seed;
+
   std::ios_base::sync_with_stdio(false);
   InitializeLogger(choice, pLogFile);
   initSpaces();

diff --git a/similarity_search/src/method/hnsw.cc b/similarity_search/src/method/hnsw.cc
@@ -158,8 +158,6 @@ namespace similarity {
     {
         AnyParamManager pmgr(IndexParams);
 
-        generator.reset(new std::default_random_engine(100));
-
         pmgr.GetParamOptional("M", M_, 16);
 
         // Let's use a generic algorithm by default!

diff --git a/similarity_search/src/utils.cc b/similarity_search/src/utils.cc
@@ -40,7 +40,6 @@
 
 namespace similarity {
 
-
 const char* GetFileName(const char* fullpath) {
   for (int i = strlen(fullpath) - 1; i >= 0; --i) {
     if (fullpath[i] == '\\' || fullpath[i] == '/') {

diff --git a/similarity_search/test/bunit.cc b/similarity_search/test/bunit.cc
@@ -92,7 +92,7 @@ int TestRunner::RunAllTests() {
 int main(int argc, char *argv[]) {
   std::string LogFile;
   if (argc == 2) LogFile = argv[1];
-  similarity::initLibrary(LogFile.empty() ? LIB_LOGSTDERR:LIB_LOGFILE, LogFile.c_str());
+  similarity::initLibrary(0, LogFile.empty() ? LIB_LOGSTDERR:LIB_LOGFILE, LogFile.c_str());
 
   return similarity::TestRunner::Instance().RunAllTests();
 }
diff --git a/similarity_search/test/test_integr.cc b/similarity_search/test/test_integr.cc
@@ -308,7 +308,7 @@ int main(int ac, char* av[]) {
   string LogFile;
   if (ac == 2) LogFile = av[1];
 
-  initLibrary(LogFile.empty() ? LIB_LOGSTDERR:LIB_LOGFILE, LogFile.c_str());
+  initLibrary(0, LogFile.empty() ? LIB_LOGSTDERR:LIB_LOGFILE, LogFile.c_str());
 
   WallClockTimer timer;
   timer.reset();