faiss 0.1.2 → 0.1.3

data/vendor/faiss/tests/test_ivfpq_codec.cpp

@@ -7,6 +7,7 @@
 
 #include <cstdio>
 #include <cstdlib>
+#include <random>
 
 #include <gtest/gtest.h>
 
@@ -51,8 +52,10 @@ double eval_codec_error (long ncentroids, long m, const std::vector<float> &v)
 TEST(IVFPQ, codec) {
 
     std::vector <float> database (nb * d);
+    std::mt19937 rng;
+    std::uniform_real_distribution<> distrib;
     for (size_t i = 0; i < nb * d; i++) {
-        database[i] = drand48();
+        database[i] = distrib(rng);
     }
 
     double err0 = eval_codec_error(16, 8, database);

data/vendor/faiss/tests/test_ivfpq_indexing.cpp

@@ -8,6 +8,7 @@
 
 #include <cstdio>
 #include <cstdlib>
+#include <random>
 
 #include <gtest/gtest.h>
 
@@ -39,13 +40,14 @@ TEST(IVFPQ, accuracy) {
     // index that gives the ground-truth
     faiss::IndexFlatL2 index_gt (d);
 
-    srand48 (35);
+    std::mt19937 rng;
+    std::uniform_real_distribution<> distrib;
 
     { // training
 
         std::vector <float> trainvecs (nt * d);
         for (size_t i = 0; i < nt * d; i++) {
-            trainvecs[i] = drand48();
+            trainvecs[i] = distrib(rng);
         }
         index.verbose = true;
         index.train (nt, trainvecs.data());
@@ -55,7 +57,7 @@ TEST(IVFPQ, accuracy) {
 
         std::vector <float> database (nb * d);
         for (size_t i = 0; i < nb * d; i++) {
-            database[i] = drand48();
+            database[i] = distrib(rng);
         }
 
         index.add (nb, database.data());
@@ -69,7 +71,7 @@ TEST(IVFPQ, accuracy) {
 
         std::vector <float> queries (nq * d);
         for (size_t i = 0; i < nq * d; i++) {
-            queries[i] = drand48();
+            queries[i] = distrib(rng);
         }
 
         std::vector<faiss::Index::idx_t> gt_nns (nq);

data/vendor/faiss/tests/test_lowlevel_ivf.cpp

@@ -5,12 +5,14 @@
  * LICENSE file in the root directory of this source tree.
  */
 
+#include <cinttypes>
 #include <cstdio>
 #include <cstdlib>
 
 #include <memory>
 #include <vector>
 #include <thread>
+#include <random>
 
 #include <gtest/gtest.h>
 
@@ -23,6 +25,7 @@
 #include <faiss/IVFlib.h>
 #include <faiss/VectorTransform.h>
 
+
 using namespace faiss;
 
 namespace {
@@ -44,12 +47,15 @@ size_t nq = 200;
 
 int k = 10;
 
+std::mt19937 rng;
+
 
 std::vector<float> make_data(size_t n)
 {
     std::vector <float> database (n * d);
+    std::uniform_real_distribution<> distrib;
     for (size_t i = 0; i < n * d; i++) {
-        database[i] = drand48();
+        database[i] = distrib(rng);
     }
     return database;
 }
@@ -273,8 +279,9 @@ std::vector<uint8_t> make_data_binary(size_t n)
 {
 
     std::vector <uint8_t> database (n * nbit / 8);
+    std::uniform_int_distribution<> distrib;
     for (size_t i = 0; i < n * d; i++) {
-        database[i] = lrand48();
+        database[i] = distrib(rng);
     }
     return database;
 }
@@ -376,7 +383,7 @@ void test_lowlevel_access_binary (const char *index_key) {
 
         printf("new before reroder: [");
         for (int j = 0; j < k; j++)
-            printf("%ld,%d ", I[j], D[j]);
+            printf("%" PRId64 ",%d ", I[j], D[j]);
         printf("]\n");
 
         // re-order heap
@@ -384,10 +391,10 @@ void test_lowlevel_access_binary (const char *index_key) {
 
         printf("ref: [");
         for (int j = 0; j < k; j++)
-            printf("%ld,%d ", I_ref[j], D_ref[j]);
+            printf("%" PRId64 ",%d ", I_ref[j], D_ref[j]);
         printf("]\nnew: [");
         for (int j = 0; j < k; j++)
-            printf("%ld,%d ", I[j], D[j]);
+            printf("%" PRId64 ",%d ", I[j], D[j]);
         printf("]\n");
 
         // check that we have the same results as the reference search

data/vendor/faiss/tests/test_merge.cpp

@@ -7,6 +7,7 @@
 
 #include <cstdio>
 #include <cstdlib>
+#include <random>
 
 #include <gtest/gtest.h>
 
@@ -70,11 +71,13 @@ struct CommonData {
 
     CommonData(): database (nb * d), queries (nq * d), ids(nb), quantizer (d) {
 
+        std::mt19937 rng;
+        std::uniform_real_distribution<> distrib;
         for (size_t i = 0; i < nb * d; i++) {
-            database[i] = drand48();
+            database[i] = distrib(rng);
         }
         for (size_t i = 0; i < nq * d; i++) {
-            queries[i] = drand48();
+            queries[i] = distrib(rng);
         }
         for (int i = 0; i < nb; i++) {
             ids[i] = 123 + 456 * i;
@@ -111,7 +114,7 @@ int compare_merged (faiss::IndexShards *index_shards, bool shift_ids,
                    shift_ids);
         }
 
-        index_shards->sync_with_shard_indexes();
+        index_shards->syncWithSubIndexes();
     } else {
         std::vector<const faiss::InvertedLists *> lists;
         faiss::IndexIVF *index0 = nullptr;

data/vendor/faiss/tests/test_ondisk_ivf.cpp

@@ -7,6 +7,7 @@
 
 #include <cstdio>
 #include <cstdlib>
+#include <random>
 
 #include <omp.h>
 
@@ -69,8 +70,10 @@ TEST(ONDISK, make_invlists) {
 
     {
         std::vector<uint8_t> code(32);
+        std::mt19937 rng;
+        std::uniform_real_distribution<> distrib;
         for (int i = 0; i < nadd; i++) {
-            double d = drand48();
+            double d = distrib(rng);
             int list_no = int(nlist * d * d); // skewed distribution
             int * ar = (int*)code.data();
             ar[0] = i;
@@ -183,8 +186,10 @@ TEST(ONDISK, make_invlists_threaded) {
 
     std::vector<int> list_nos (nadd);
 
+    std::mt19937 rng;
+    std::uniform_real_distribution<> distrib;
     for (int i = 0; i < nadd; i++) {
-        double d = drand48();
+        double d = distrib(rng);
         list_nos[i] = int(nlist * d * d); // skewed distribution
     }
 

data/vendor/faiss/tests/test_pairs_decoding.cpp

@@ -10,6 +10,7 @@
 
 #include <memory>
 #include <vector>
+#include <random>
 
 #include <gtest/gtest.h>
 
@@ -37,11 +38,14 @@ size_t nb = 8000;
 // nb of queries
 size_t nq = 200;
 
+std::mt19937 rng;
+
 std::vector<float> make_data(size_t n)
 {
     std::vector <float> database (n * d);
+    std::uniform_real_distribution<> distrib;
     for (size_t i = 0; i < n * d; i++) {
-        database[i] = drand48();
+        database[i] = distrib(rng);
     }
     return database;
 }

data/vendor/faiss/tests/test_params_override.cpp

@@ -10,6 +10,7 @@
 
 #include <memory>
 #include <vector>
+#include <random>
 
 #include <gtest/gtest.h>
 
@@ -19,6 +20,7 @@
 #include <faiss/AutoTune.h>
 #include <faiss/IVFlib.h>
 
+
 using namespace faiss;
 
 namespace {
@@ -35,13 +37,15 @@ size_t nb = 1000;
 // nb of queries
 size_t nq = 200;
 
+std::mt19937 rng;
 
 
 std::vector<float> make_data(size_t n)
 {
     std::vector <float> database (n * d);
+    std::uniform_real_distribution<> distrib;
     for (size_t i = 0; i < n * d; i++) {
-        database[i] = drand48();
+        database[i] = distrib(rng);
     }
     return database;
 }
@@ -155,8 +159,9 @@ TEST(TPO, IVFFlatPP) {
 
 std::vector<uint8_t> make_data_binary(size_t n) {
     std::vector <uint8_t> database (n * d / 8);
+    std::uniform_int_distribution<> distrib;
     for (size_t i = 0; i < n * d / 8; i++) {
-        database[i] = lrand48();
+        database[i] = distrib(rng);
     }
     return database;
 }

data/vendor/faiss/tests/test_sliding_ivf.cpp

@@ -9,6 +9,7 @@
 #include <cstdlib>
 
 #include <memory>
+#include <random>
 #include <vector>
 
 #include <gtest/gtest.h>
@@ -19,6 +20,7 @@
 #include <faiss/clone_index.h>
 #include <faiss/IVFlib.h>
 
+
 using namespace faiss;
 
 typedef Index::idx_t idx_t;
@@ -47,8 +49,10 @@ int window_size = 10;
 std::vector<float> make_data(size_t n)
 {
     std::vector <float> database (n * d);
+    std::mt19937 rng;
+    std::uniform_real_distribution<> distrib;
     for (size_t i = 0; i < n * d; i++) {
-        database[i] = drand48();
+        database[i] = distrib(rng);
     }
     return database;
 }
@@ -92,8 +96,10 @@ void make_index_slices (const Index* trained_index,
 
         auto xb = make_data(nb * d);
         std::vector<faiss::Index::idx_t> ids (nb);
+        std::mt19937 rng;
+        std::uniform_int_distribution<> distrib;
         for (int j = 0; j < nb; j++) {
-            ids[j] = lrand48();
+            ids[j] = distrib(rng);
         }
         index->add_with_ids (nb, xb.data(), ids.data());
     }
@@ -192,7 +198,7 @@ int test_sliding_invlists (const char *index_key) {
         // will be deleted by the index
         index_ivf->replace_invlists (ci, true);
 
-        printf ("   nb invlists = %ld\n", ils.size());
+        printf ("   nb invlists = %zd\n", ils.size());
 
         auto new_res = search_index (index.get(), xq.data());
 
@@ -208,7 +214,7 @@ int test_sliding_invlists (const char *index_key) {
             if (ref_res[j] != new_res[j])
                 ndiff++;
         }
-        printf("  nb differences: %ld / %ld\n",
+        printf("  nb differences: %zd / %zd\n",
                ndiff, ref_res.size());
         EXPECT_EQ (ref_res, new_res);
     }

data/vendor/faiss/tutorial/cpp/1-Flat.cpp

@@ -7,39 +7,45 @@
 
 #include <cstdio>
 #include <cstdlib>
+#include <random>
 
 #include <faiss/IndexFlat.h>
 
+// 64-bit int
+using idx_t = faiss::Index::idx_t;
 
 int main() {
     int d = 64;                            // dimension
     int nb = 100000;                       // database size
     int nq = 10000;                        // nb of queries
 
+    std::mt19937 rng;
+    std::uniform_real_distribution<> distrib;
+
     float *xb = new float[d * nb];
     float *xq = new float[d * nq];
 
     for(int i = 0; i < nb; i++) {
         for(int j = 0; j < d; j++)
-            xb[d * i + j] = drand48();
+            xb[d * i + j] = distrib(rng);
         xb[d * i] += i / 1000.;
     }
 
     for(int i = 0; i < nq; i++) {
         for(int j = 0; j < d; j++)
-            xq[d * i + j] = drand48();
+            xq[d * i + j] = distrib(rng);
         xq[d * i] += i / 1000.;
     }
 
     faiss::IndexFlatL2 index(d);           // call constructor
     printf("is_trained = %s\n", index.is_trained ? "true" : "false");
     index.add(nb, xb);                     // add vectors to the index
-    printf("ntotal = %ld\n", index.ntotal);
+    printf("ntotal = %zd\n", index.ntotal);
 
     int k = 4;
 
     {       // sanity check: search 5 first vectors of xb
-        long *I = new long[k * 5];
+        idx_t *I = new idx_t[k * 5];
         float *D = new float[k * 5];
 
         index.search(5, xb, k, D, I);
@@ -48,7 +54,7 @@ int main() {
         printf("I=\n");
         for(int i = 0; i < 5; i++) {
             for(int j = 0; j < k; j++)
-                printf("%5ld ", I[i * k + j]);
+                printf("%5zd ", I[i * k + j]);
             printf("\n");
         }
 
@@ -65,7 +71,7 @@ int main() {
 
 
     {       // search xq
-        long *I = new long[k * nq];
+        idx_t *I = new idx_t[k * nq];
         float *D = new float[k * nq];
 
         index.search(nq, xq, k, D, I);
@@ -74,14 +80,14 @@ int main() {
         printf("I (5 first results)=\n");
         for(int i = 0; i < 5; i++) {
             for(int j = 0; j < k; j++)
-                printf("%5ld ", I[i * k + j]);
+                printf("%5zd ", I[i * k + j]);
             printf("\n");
         }
 
         printf("I (5 last results)=\n");
         for(int i = nq - 5; i < nq; i++) {
             for(int j = 0; j < k; j++)
-                printf("%5ld ", I[i * k + j]);
+                printf("%5zd ", I[i * k + j]);
             printf("\n");
         }
 

data/vendor/faiss/tutorial/cpp/2-IVFFlat.cpp

@@ -8,28 +8,33 @@
 #include <cstdio>
 #include <cstdlib>
 #include <cassert>
+#include <random>
 
 #include <faiss/IndexFlat.h>
 #include <faiss/IndexIVFFlat.h>
 
+using idx_t = faiss::Index::idx_t;
 
 int main() {
     int d = 64;                            // dimension
     int nb = 100000;                       // database size
     int nq = 10000;                        // nb of queries
 
+    std::mt19937 rng;
+    std::uniform_real_distribution<> distrib;
+
     float *xb = new float[d * nb];
     float *xq = new float[d * nq];
 
     for(int i = 0; i < nb; i++) {
         for(int j = 0; j < d; j++)
-            xb[d * i + j] = drand48();
+            xb[d * i + j] = distrib(rng);
         xb[d * i] += i / 1000.;
     }
 
     for(int i = 0; i < nq; i++) {
         for(int j = 0; j < d; j++)
-            xq[d * i + j] = drand48();
+            xq[d * i + j] = distrib(rng);
         xq[d * i] += i / 1000.;
     }
 
@@ -38,15 +43,14 @@ int main() {
     int k = 4;
 
     faiss::IndexFlatL2 quantizer(d);       // the other index
-    faiss::IndexIVFFlat index(&quantizer, d, nlist, faiss::METRIC_L2);
-    // here we specify METRIC_L2, by default it performs inner-product search
+    faiss::IndexIVFFlat index(&quantizer, d, nlist);
     assert(!index.is_trained);
     index.train(nb, xb);
     assert(index.is_trained);
     index.add(nb, xb);
 
     {       // search xq
-        long *I = new long[k * nq];
+        idx_t *I = new idx_t[k * nq];
         float *D = new float[k * nq];
 
         index.search(nq, xq, k, D, I);
@@ -54,7 +58,7 @@ int main() {
         printf("I=\n");
         for(int i = nq - 5; i < nq; i++) {
             for(int j = 0; j < k; j++)
-                printf("%5ld ", I[i * k + j]);
+                printf("%5zd ", I[i * k + j]);
             printf("\n");
         }
 
@@ -64,7 +68,7 @@ int main() {
         printf("I=\n");
         for(int i = nq - 5; i < nq; i++) {
             for(int j = 0; j < k; j++)
-                printf("%5ld ", I[i * k + j]);
+                printf("%5zd ", I[i * k + j]);
             printf("\n");
         }