HNSW

Append LAL graph

TestHNSW/HNSWDemo/Program.cs

@@ -1,6 +1,7 @@
 using HNSWDemo.Tests;
 using System;
-using ZeroLevel.Services.Web;
+using System.IO;
+using ZeroLevel.HNSW;
 
 namespace HNSWDemo
 {
@@ -8,12 +9,29 @@ namespace HNSWDemo
     {
         static void Main(string[] args)
         {
-            var uri = new Uri("https://hack33d.ru/bpla/upload.php?path=128111&get=0J/QuNC70LjQv9C10L3QutC+INCS0LvQsNC00LjQvNC40YAg0JzQuNGF0LDQudC70L7QstC40Yc7MDQuMDkuMTk1NCAoNjYg0LvQtdGCKTvQnNC+0YHQutC+0LLRgdC60LDRjzsxMjgxMTE7TEFfUkVaVVM7RkxZXzAy");
-            var parts = UrlUtility.ParseQueryString(uri.Query);
-            new AutoClusteringMNISTTest().Run();
-            //new HistogramTest().Run();
+            
+
+            new LALTest().Run();
+            // new AutoClusteringMNISTTest().Run();
+            // new AccuracityTest().Run();
             Console.WriteLine("Completed");
             Console.ReadKey();
         }
+
+        static int GetC(string file)
+        {
+            var name = Path.GetFileNameWithoutExtension(file);
+            var index = name.IndexOf("_M");
+            if (index > 0)
+            {
+                index = name.IndexOf("_", index + 2);
+                if (index > 0)
+                {
+                    var num = name.Substring(index + 1, name.Length - index - 1);
+                    return int.Parse(num);
+                }
+            }
+            return -1;
+        }
     }
 }

TestHNSW/HNSWDemo/Properties/launchSettings.json

@@ -0,0 +1,9 @@
+{
+  "profiles": {
+    "HNSWDemo": {
+      "commandName": "Project",
+      "hotReloadEnabled": false,
+      "nativeDebugging": false
+    }
+  }
+}

TestHNSW/HNSWDemo/Tests/AccuracityTest.cs

@@ -11,7 +11,7 @@ namespace HNSWDemo.Tests
         : ITest
     {
         private static int K = 200;
-        private static int count = 3000;
+        private static int count = 10000;
         private static int testCount = 500;
         private static int dimensionality = 128;
 

TestHNSW/HNSWDemo/Tests/HistogramTest.cs

@@ -1,5 +1,6 @@
 using System;
 using System.Drawing;
+using System.IO;
 using System.Linq;
 using ZeroLevel.HNSW;
 
@@ -10,12 +11,23 @@ namespace HNSWDemo.Tests
     {
         private static int Count = 3000;
         private static int Dimensionality = 128;
-        private static int Width = 3000;
-        private static int Height = 3000;
+        private static int Width = 2440;
+        private static int Height = 1920;
 
         public void Run()
         {
-            var vectors = VectorUtils.RandomVectors(Dimensionality, Count);
+            Create(Dimensionality, @"D:\hist");
+            // Process.Start("explorer", $"D:\\hist{Dimensionality.ToString("D3")}.jpg");
+
+            /* for (int i = 12; i < 512; i++)
+             {
+                 Create(i, @"D:\hist");
+             }*/
+        }
+
+        private void Create(int dim, string output)
+        {
+            var vectors = VectorUtils.RandomVectors(dim, Count);
             var world = SmallWorld.CreateWorld<float[]>(NSWOptions<float[]>.Create(8, 16, 200, 200, Metrics.L2Euclidean));
             world.AddItems(vectors);
 
@@ -29,7 +41,7 @@ namespace HNSWDemo.Tests
             var max = histogram.Bounds[threshold];
             var R = (max + min) / 2;
 
-            DrawHistogram(histogram, @"D:\hist.jpg");
+            DrawHistogram(histogram, Path.Combine(output, $"hist{dim.ToString("D3")}.jpg"));
         }
 
         static void DrawHistogram(Histogram histogram, string filename)

TestHNSW/HNSWDemo/Tests/LALTest.cs

@@ -0,0 +1,101 @@
+using HNSWDemo.Model;
+using System;
+using System.Collections.Generic;
+using System.IO;
+using System.Linq;
+using ZeroLevel.HNSW;
+
+namespace HNSWDemo.Tests
+{
+    internal class LALTest
+        : ITest
+    {
+        private const int count = 5000;
+        private const int testCount = 100;
+        private const int dimensionality = 128;
+
+        public void Run()
+        {
+            var moda = 3;
+            var persons = Person.GenerateRandom(dimensionality, count);
+            var samples = new Dictionary<int, List<(float[], Person)>>();
+            var options = NSWOptions<float[]>.Create(6, 8, 100, 100, Metrics.Cosine);
+
+            foreach (var p in persons)
+            {
+                var c = (int)Math.Abs(p.Item2.Number.GetHashCode() % moda);
+                if (samples.ContainsKey(c) == false) samples.Add(c, new List<(float[], Person)>());
+                samples[c].Add(p);
+            }
+
+
+            var worlds = new SplittedLALGraph();
+            var mappers = new HNSWMappers<long>(l => (int)Math.Abs(l.GetHashCode() % moda));
+
+            var worlds_dict = new Dictionary<int, SmallWorld<float[]>>();
+            var maps_dict = new Dictionary<int, HNSWMap<long>>();
+
+            foreach (var p in samples)
+            {
+                var c = p.Key;
+                if (worlds_dict.ContainsKey(c) == false)
+                {
+                    worlds_dict.Add(c, new SmallWorld<float[]>(options));
+                }
+                if (maps_dict.ContainsKey(c) == false)
+                {
+                    maps_dict.Add(c, new HNSWMap<long>());
+                }
+                var w = worlds_dict[c];
+                var m = maps_dict[c];
+                var ids = w.AddItems(p.Value.Select(i => i.Item1));
+
+                for (int i = 0; i < ids.Length; i++)
+                {
+                    m.Append(p.Value[i].Item2.Number, ids[i]);
+                }
+            }
+
+            var name = Guid.NewGuid().ToString();
+            foreach (var p in samples)
+            {
+                var c = p.Key;
+                var w = worlds_dict[c];
+                var m = maps_dict[c];
+
+                using (var s = File.Create(name))
+                {
+                    w.Serialize(s);
+                }
+                using (var s = File.OpenRead(name))
+                {
+                    var l = LALGraph.FromHNSWGraph<float[]>(s);
+                    worlds.Append(l, c);
+                }
+                File.Delete(name);
+                mappers.Append(m, c);
+            }
+
+            var entries = new long[10];
+            for (int i = 0; i < entries.Length; i++)
+            {
+                entries[i] = persons[DefaultRandomGenerator.Instance.Next(0, persons.Count - 1)].Item2.Number;
+            }
+
+            var contexts = mappers.CreateContext(null, entries);
+            var result = worlds.KNearest(10, contexts);
+
+            Console.WriteLine("Entries:");
+            foreach (var n in entries)
+            {
+                Console.WriteLine($"\t{n}");
+            }
+
+            Console.WriteLine("Extensions:");
+            foreach (var n in mappers.ConvertIdsToFeatures(result))
+            {
+                Console.WriteLine($"\t[{n}]");
+            }
+        }
+    }
+}

TestPipeLine/Consumer/Program.cs

@@ -7,6 +7,15 @@ namespace Consumer
     {
         static void Main(string[] args)
         {
+            IConfiguration conf = Configuration.Create();
+            conf.Append("ServiceName", "Test consumer");
+            conf.Append("ServiceKey", "test.consumer");
+            conf.Append("ServiceType", "Destination");
+            conf.Append("ServiceGroup", "Test");
+            conf.Append("Version", "1.0.0.1");
+            conf.Append("discovery", "127.0.0.1:5012");
+            Configuration.Save(conf);
+
             Bootstrap.Startup<ConsumerService>(args)
                 .EnableConsoleLog(LogLevel.FullStandart)
                 .UseDiscovery()

TestPipeLine/Processor/Program.cs

@@ -7,6 +7,15 @@ namespace Processor
     {
         static void Main(string[] args)
         {
+            IConfiguration conf = Configuration.Create();
+            conf.Append("ServiceName", "Test processor");
+            conf.Append("ServiceKey", "test.processor");
+            conf.Append("ServiceType", "Core");
+            conf.Append("ServiceGroup", "Test");
+            conf.Append("Version", "1.0.0.1");
+            conf.Append("discovery", "127.0.0.1:5012");
+            Configuration.Save(conf);
+
             Bootstrap.Startup<ProcessorService>(args)
                 .EnableConsoleLog(LogLevel.FullStandart)
                 .UseDiscovery()

TestPipeLine/Source/Program.cs

@@ -6,6 +6,15 @@ namespace Source
     {
         static void Main(string[] args)
         {
+            IConfiguration conf = Configuration.Create();
+            conf.Append("ServiceName", "Test source");
+            conf.Append("ServiceKey", "test.source");
+            conf.Append("ServiceType", "Sources");
+            conf.Append("ServiceGroup", "Test");
+            conf.Append("Version", "1.0.0.1");
+            conf.Append("discovery", "127.0.0.1:5012");
+            Configuration.Save(conf);
+
             Bootstrap.Startup<SourceService>(args)
                 .EnableConsoleLog(ZeroLevel.Logging.LogLevel.FullStandart)
                 .UseDiscovery()

TestPipeLine/Watcher/Program.cs

@@ -6,6 +6,15 @@ namespace Watcher
     {
         static void Main(string[] args)
         {
+            IConfiguration conf = Configuration.Create();
+            conf.Append("ServiceName", "Watcher");
+            conf.Append("ServiceKey", "test.watcher");
+            conf.Append("ServiceType", "System");
+            conf.Append("ServiceGroup", "Test");
+            conf.Append("Version", "1.0.0.1");
+            conf.Append("discovery", "127.0.0.1:5012");
+            Configuration.Save(conf);
+
             Bootstrap.Startup<WatcherService>(args)
                 .UseDiscovery()
                 .Run()

ZeroLevel.HNSW/Services/HNSWMap.cs

@@ -1,4 +1,5 @@
 using System.Collections.Generic;
+using System.IO;
 using ZeroLevel.Services.Serialization;
 
 namespace ZeroLevel.HNSW
@@ -12,6 +13,7 @@ namespace ZeroLevel.HNSW
         private Dictionary<TFeature, int> _map;
         private Dictionary<int, TFeature> _reverse_map;
 
+        public int this[TFeature feature] => _map.GetValueOrDefault(feature);
         public HNSWMap(int capacity = -1)
         {
             if (capacity > 0)
@@ -27,6 +29,14 @@ namespace ZeroLevel.HNSW
             }
         }
 
+        public HNSWMap(Stream stream)
+        {
+            using (var reader = new MemoryStreamReader(stream))
+            {
+                Deserialize(reader);
+            }
+        }
+
         public void Append(TFeature feature, int vectorId)
         {
             _map[feature] = vectorId;

ZeroLevel.HNSW/Services/HNSWMappers.cs

@@ -0,0 +1,75 @@
+using System;
+using System.Collections.Generic;
+
+namespace ZeroLevel.HNSW
+{
+    public class HNSWMappers<TFeature>
+    {
+        private readonly IDictionary<int, HNSWMap<TFeature>> _mappers = new Dictionary<int, HNSWMap<TFeature>>();
+        private readonly Func<TFeature, int> _bucketFunction;
+        public HNSWMappers(Func<TFeature, int> bucketFunction)
+        {
+            _bucketFunction = bucketFunction;
+        }
+
+        public void Append(HNSWMap<TFeature> map, int c)
+        {
+            _mappers.Add(c, map);
+        }
+
+        public IEnumerable<TFeature> ConvertIdsToFeatures(IEnumerable<int> ids)
+        {
+            foreach (var map in _mappers)
+            {
+                foreach (var feature in map.Value.ConvertIdsToFeatures(ids))
+                {
+                    yield return feature;
+                }
+            }
+        }
+
+        public IDictionary<int, SearchContext> CreateContext(IEnumerable<TFeature> activeNodes, IEnumerable<TFeature> entryPoints)
+        {
+            var actives = new Dictionary<int, List<int>>();
+            var entries = new Dictionary<int, List<int>>();
+            if (activeNodes != null)
+            {
+                foreach (var node in activeNodes)
+                {
+                    var c = _bucketFunction(node);
+                    if (_mappers.ContainsKey(c))
+                    {
+                        if (actives.ContainsKey(c) == false)
+                        {
+                            actives.Add(c, new List<int>());
+                        }
+                        actives[c].Add(_mappers[c][node]);
+                    }
+                }
+            }
+            if (entryPoints != null)
+            {
+                foreach (var entryPoint in entryPoints)
+                {
+                    var c = _bucketFunction(entryPoint);
+                    if (_mappers.ContainsKey(c))
+                    {
+                        if (entries.ContainsKey(c) == false)
+                        {
+                            entries.Add(c, new List<int>());
+                        }
+                        entries[c].Add(_mappers[c][entryPoint]);
+                    }
+                }
+            }
+            var result = new Dictionary<int, SearchContext>();
+            foreach (var pair in _mappers)
+            {
+                var active = actives.GetValueOrDefault(pair.Key);
+                var entry = entries.GetValueOrDefault(pair.Key);
+                result.Add(pair.Key, new SearchContext().SetActiveNodes(active).SetEntryPointsNodes(entry));
+            }
+            return result;
+        }
+    }
+}

ZeroLevel.HNSW/Services/LAL/LALGraph.cs

@@ -0,0 +1,105 @@
+using System.Collections.Generic;
+using System.IO;
+using System.Linq;
+using ZeroLevel.Services.Serialization;
+
+namespace ZeroLevel.HNSW
+{
+    public class LALGraph
+    {
+        private readonly LALLinks _links = new LALLinks();
+
+        private LALGraph() { }
+        public static LALGraph FromLALGraph(Stream stream)
+        {
+            var l = new LALGraph();
+            l.Deserialize(stream);
+            return l;
+        }
+
+        public static LALGraph FromHNSWGraph<TItem>(Stream stream)
+        {
+            var l = new LALGraph();
+            l.DeserializeFromHNSW<TItem>(stream);
+            return l;
+        }
+
+        public IEnumerable<int> KNearest(int k, SearchContext context)
+        {
+            var v = new VisitedBitSet(_links.Count, 1);
+            var C = new Queue<int>();
+            var W = new List<int>();
+            var entryPoints = context.EntryPoints;
+
+            do
+            {
+                foreach (var ep in entryPoints)
+                {
+                    var neighboursIds = _links.FindNeighbors(ep);
+                    for (int i = 0; i < neighboursIds.Length; ++i)
+                    {
+                        C.Enqueue(neighboursIds[i]);
+                    }
+                    v.Add(ep);
+                }
+                // run bfs
+                while (C.Count > 0)
+                {
+                    // get next candidate to check and expand
+                    var toExpand = C.Dequeue();
+                    if (context.IsActiveNode(toExpand))
+                    {
+                        if (W.Count < k)
+                        {
+                            W.Add(toExpand);
+                            if (W.Count > k)
+                            {
+                                var loser_id = DefaultRandomGenerator.Instance.Next(0, W.Count - 1);
+                                W.RemoveAt(loser_id);
+                            }
+                        }
+                    }
+                }
+                entryPoints = W.Select(id => id).ToList();
+            }
+            while (W.Count < k && entryPoints.Any());
+            C.Clear();
+            v.Clear();
+            return W;
+        }
+
+        public void Deserialize(Stream stream)
+        {
+            using (var reader = new MemoryStreamReader(stream))
+            {
+                _links.Deserialize(reader); // deserialize only base layer and skip another
+            }
+        }
+
+        public void DeserializeFromHNSW<TItem>(Stream stream)
+        {
+            using (var reader = new MemoryStreamReader(stream))
+            {
+                reader.ReadInt32(); //  EntryPoint
+                reader.ReadInt32(); //  MaxLayer
+
+                int count = reader.ReadInt32(); // Vectors count
+                for (int i = 0; i < count; i++)
+                {
+                    reader.ReadCompatible<TItem>(); //  Vector
+                }
+
+                reader.ReadInt32(); //  countLayers
+                _links.Deserialize(reader); // deserialize only base layer and skip another
+            }
+        }
+
+        public void Serialize(Stream stream)
+        {
+            using (var writer = new MemoryStreamWriter(stream))
+            {
+                _links.Serialize(writer);
+            }
+        }
+    }
+}

ZeroLevel.HNSW/Services/LAL/LALLinks.cs

@@ -0,0 +1,79 @@
+using System.Collections.Concurrent;
+using System.Collections.Generic;
+using System.Linq;
+using ZeroLevel.Services.Serialization;
+
+namespace ZeroLevel.HNSW
+{
+    internal class LALLinks
+    {
+        private ConcurrentDictionary<int, int[]> _set = new ConcurrentDictionary<int, int[]>();
+        internal IDictionary<int, int[]> Links => _set;
+
+        private readonly int[] _empty = new int[0];
+        internal int Count => _set.Count;
+
+        public LALLinks()
+        {
+        }
+
+        internal IEnumerable<(int, int)> FindLinksForId(int id)
+        {
+            if (_set.ContainsKey(id))
+            {
+                return _set[id].Select(v => (id, v));
+            }
+            return Enumerable.Empty<(int, int)>();
+        }
+
+        internal int[] FindNeighbors(int id)
+        {
+            if (_set.ContainsKey(id))
+            {
+                return _set[id];
+            }
+            return _empty;
+        }
+
+        internal IEnumerable<(int, int)> Items()
+        {
+            return _set
+                .SelectMany(pair => _set[pair.Key]
+                    .Select(v => (pair.Key, v)));
+        }
+
+        public void Dispose()
+        {
+            _set.Clear();
+            _set = null;
+        }
+        public void Serialize(IBinaryWriter writer)
+        {
+            writer.WriteInt32(_set.Count);
+            foreach (var record in _set)
+            {
+                writer.WriteInt32(record.Key);
+                writer.WriteCollection(record.Value);
+            }
+        }
+
+        public void Deserialize(IBinaryReader reader)
+        {
+            _set.Clear();
+            _set = null;
+            var count = reader.ReadInt32();
+            _set = new ConcurrentDictionary<int, int[]>(1, count);
+
+            for (int i = 0; i < count; i++)
+            {
+                var id = reader.ReadInt32();
+                var links_count = reader.ReadInt32();
+                _set[id] = new int[links_count];
+                for (int l = 0; l < links_count; l++)
+                {
+                    _set[id][l] = reader.ReadInt32();
+                }
+            }
+        }
+    }
+}

ZeroLevel.HNSW/Services/LAL/SplittedLALGraph.cs

@@ -0,0 +1,29 @@
+using System.Collections.Generic;
+
+namespace ZeroLevel.HNSW
+{
+    public class SplittedLALGraph
+    {
+        private readonly IDictionary<int, LALGraph> _graphs = new Dictionary<int, LALGraph>();
+
+        public void Append(LALGraph graph, int c)
+        {
+            _graphs.Add(c, graph);
+        }
+
+        public IEnumerable<int> KNearest(int k, IDictionary<int, SearchContext> contexts)
+        {
+            var partial_k = 1 + (k / _graphs.Count);
+            var result = new List<int>();
+            foreach (var graph in _graphs)
+            {
+                var context = contexts[graph.Key];
+                if (context.EntryPoints != null)
+                {
+                    result.AddRange(graph.Value.KNearest(partial_k, context));
+                }
+            }
+            return result;
+        }
+    }
+}

ZeroLevel.HNSW/Services/LinksSet.cs

@@ -70,24 +70,15 @@ namespace ZeroLevel.HNSW
             _set.Clear();
             _set = null;
         }
-
-        private const int HALF_LONG_BITS = 32;
         public void Serialize(IBinaryWriter writer)
         {
-            writer.WriteBoolean(false); // true - set with weights
-            var count = _set.Sum(pair => pair.Value.Count);
-            writer.WriteInt32(count);
+            writer.WriteInt32(_set.Count);
             foreach (var record in _set)
             {
-                var id = record.Key;
-                foreach (var r in record.Value)
-                {
-                    var key = (((long)(id)) << HALF_LONG_BITS) + r;
-                    writer.WriteLong(key);
-                }
+                writer.WriteInt32(record.Key);
+                writer.WriteCollection(record.Value);
             }
         }
-
         public void Deserialize(IBinaryReader reader)
         {
             if (reader.ReadBoolean() != false)
@@ -100,16 +91,13 @@ namespace ZeroLevel.HNSW
             _set = new ConcurrentDictionary<int, HashSet<int>>();
             for (int i = 0; i < count; i++)
             {
-                var key = reader.ReadLong();
-
-                var id1 = (int)(key >> HALF_LONG_BITS);
-                var id2 = (int)(key - (((long)id1) << HALF_LONG_BITS));
-
-                if (!_set.ContainsKey(id1))
+                var id = reader.ReadInt32();
+                var links_count = reader.ReadInt32();
+                _set[id] = new HashSet<int>(links_count);
+                for (var l = 0; l < links_count; l++)
                 {
-                    _set[id1] = new HashSet<int>();
+                    _set[id].Add(reader.ReadInt32());
                 }
-                _set[id1].Add(id2);
             }
         }
     }

ZeroLevel.HNSW/SmallWorldFactory.cs

@@ -4,7 +4,9 @@ namespace ZeroLevel.HNSW
 {
     public static class SmallWorld
     {
-        public static SmallWorld<TItem> CreateWorld<TItem>(NSWOptions<TItem> options) => new SmallWorld<TItem>(options);
-        public static SmallWorld<TItem> CreateWorldFrom<TItem>(NSWOptions<TItem> options, Stream stream) => new SmallWorld<TItem>(options, stream);
+        public static SmallWorld<TItem> CreateWorld<TItem>(NSWOptions<TItem> options) 
+            => new SmallWorld<TItem>(options);
+        public static SmallWorld<TItem> CreateWorldFrom<TItem>(NSWOptions<TItem> options, Stream stream) 
+            => new SmallWorld<TItem>(options, stream);
     }
 }

ZeroLevel.HNSW/ZeroLevel.HNSW.csproj

@@ -3,8 +3,41 @@
   <PropertyGroup>
     <TargetFramework>net6.0</TargetFramework>
     <Platforms>AnyCPU;x64</Platforms>
+    <PlatformTarget>x64</PlatformTarget>
+    <DebugType>full</DebugType>
+    <Version>1.0.0.1</Version>
+    <Company>ogoun</Company>
+    <Authors>Ogoun</Authors>
+    <Copyright>Copyright Ogoun 2022</Copyright>
+    <PackageProjectUrl>https://github.com/ogoun/Zero/wiki</PackageProjectUrl>
+    <PackageIcon>zero.png</PackageIcon>
+    <RepositoryUrl>https://github.com/ogoun/Zero</RepositoryUrl>
+    <RepositoryType>git</RepositoryType>
   </PropertyGroup>
 
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|AnyCPU'">
+    <Optimize>False</Optimize>
+  </PropertyGroup>
+
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Debug|x64'">
+    <Optimize>False</Optimize>
+  </PropertyGroup>
+
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|AnyCPU'">
+    <Optimize>True</Optimize>
+  </PropertyGroup>
+
+  <PropertyGroup Condition="'$(Configuration)|$(Platform)'=='Release|x64'">
+    <Optimize>True</Optimize>
+  </PropertyGroup>
+
+  <ItemGroup>
+    <None Include="..\zero.png">
+      <Pack>True</Pack>
+      <PackagePath>\</PackagePath>
+    </None>
+  </ItemGroup>
+
   <ItemGroup>
     <PackageReference Include="System.Numerics.Vectors" Version="4.5.0" />
   </ItemGroup>

ZeroLevel/ZeroLevel.csproj

@@ -13,12 +13,12 @@
     <PackageLicenseUrl></PackageLicenseUrl>
     <PackageIconUrl></PackageIconUrl>
     <RepositoryUrl>https://github.com/ogoun/Zero</RepositoryUrl>
-    <RepositoryType>GitHub</RepositoryType>
+    <RepositoryType>git</RepositoryType>
     <Version>3.3.5.7</Version>
     <FileVersion>3.3.5.7</FileVersion>
     <Platforms>AnyCPU;x64;x86</Platforms>
     <PackageIcon>zero.png</PackageIcon>
-    <DebugType>none</DebugType>
+    <DebugType>full</DebugType>
     <ErrorReport>none</ErrorReport>
     <ApplicationIcon>zero.ico</ApplicationIcon>
   </PropertyGroup>