lsh 0.3.2-java → 0.4.0-java

data/lib/lsh/index.rb

@@ -111,9 +111,9 @@ module LSH
 
     def hash(vector, projection, bias = true)
       hash = []
-      projection.each do |random_vector|
-        dot_product = similarity(vector, random_vector)
-        window = storage.parameters[:window]
+      dot_products = (vector * projection).row(0).to_a
+      window = storage.parameters[:window]
+      dot_products.each do |dot_product|
         if window == Float::INFINITY # Binary LSH
           if dot_product >= 0
             hash << 1
@@ -128,6 +128,15 @@ module LSH
       hash
     end
 
+    def random_vector(dim)
+      MathUtil.random_gaussian_matrix(1, dim)
+    end
+
+    def random_vector_unit(dim)
+      r = random_vector(dim)
+      r /= MathUtil.norm(r)
+    end
+
     def array_to_hash(array)
       return array.hash
       # Derives a 28 bit hash value from an array of integers
@@ -150,20 +159,7 @@ module LSH
     end
 
     def generate_projection(dim, k)
-      vectors = []
-      k.times do |i|
-        vectors << random_vector(dim)
-      end
-      vectors
-    end
-
-    def random_vector_unit(dim)
-      r = random_vector(dim)
-      r /= MathUtil.norm(r)
-    end
-
-    def random_vector(dim)
-      MathUtil.random_gaussian_vector(dim)
+      MathUtil.random_gaussian_matrix(dim, k)
     end
 
     def similarity(v1, v2)

data/lib/lsh/math_util_gsl.rb

@@ -27,20 +27,24 @@ module LSH
       @@gsl_random.uniform
     end
 
-    def self.random_gaussian_vector(dim)
-      @@gsl_random.gaussian(1, dim)
+    def self.zeros(k, l)
+      GSL::Matrix.alloc(k, l)
     end
 
-    def self.zeros(dim)
-      GSL::Vector.alloc(dim)
+    def self.random_gaussian_vector(dim)
+      @@gsl_random.gaussian(1, dim)
     end
 
     def self.random_gaussian_matrix(k, l)
-      GSL::Matrix.randn(k, l)
+      matrix = zeros(k, l)
+      (0..(k - 1)).each do |i|
+        matrix.set_row(i, random_gaussian_vector(l))
+      end
+      matrix
     end
 
     def self.dot(v1, v2)
-      v1 * v2.col
+      (v1 * v2.transpose)[0,0]
     end
 
     def self.norm(v)
@@ -49,7 +53,7 @@ module LSH
 
     def self.uniq(vs)
       # Can't use uniq as
-      # [ v, JSON.parse(v.to_json) ].uniq.size == 2 with GSL
+      # [ v, JSON.parse(v.to_json, :create_additions => true) ].uniq.size == 2 with GSL
       results = []
       vs.each { |v| results << v unless results.member? v }
       results
@@ -61,7 +65,7 @@ end
 
 module GSL
 
-  class Vector
+  class Matrix
 
     def to_json(*a)
       {

data/lib/lsh/math_util_jblas.rb

@@ -24,16 +24,16 @@ module LSH
       JBLAS.rand[0,0]
     end
 
-    def self.random_gaussian_vector(dim)
-      JBLAS.randn(1, dim)
-    end
-
     def self.random_gaussian_matrix(k, l)
       JBLAS.randn(k, l)
     end
 
-    def self.zeros(dim)
-      JBLAS::DoubleMatrix.new(1, dim)
+    def self.random_gaussian_vector(l)
+      random_gaussian_matrix(1, l)
+    end
+
+    def self.zeros(k, l)
+      JBLAS::DoubleMatrix.new(k, l)
     end
 
     def self.dot(v1, v2)
@@ -68,6 +68,10 @@ module JBLAS
       from_array(o['data']).t
     end
 
+    def size
+      [ rows, columns ]
+    end
+
   end
 
 end

data/lib/lsh/storage/redis_backend.rb

@@ -68,9 +68,7 @@ module LSH
         # (too slow to serialize and store in Redis for
         # large number of dimensions/projections)
         projections.each_with_index do |projection, i|
-          projection.each_with_index do |vector, j|
-            vector.save(File.join(@data_dir, 'projections', "vector_#{i}_#{j}.dat"))
-          end
+          projection.save(File.join(@data_dir, 'projections', "projection_#{i}.dat"))
         end
       end
 
@@ -79,13 +77,9 @@ module LSH
         @projections ||= (
           projections = []
           parameters[:number_of_independent_projections].times do |i|
-            vectors = []
-            parameters[:number_of_random_vectors].times do |j|
-              v = MathUtil.zeros(parameters[:dim])
-              v.load(File.join(@data_dir, 'projections', "vector_#{i}_#{j}.dat"))
-              vectors << v
-            end
-            projections << vectors
+            m = MathUtil.zeros(parameters[:dim], parameters[:number_of_random_vectors])
+            m.load(File.join(@data_dir, 'projections', "projection_#{i}.dat"))
+            projections << m
           end
           projections
         )
@@ -119,7 +113,7 @@ module LSH
       end
 
       def load_vector(hash)
-        vector = MathUtil.zeros(parameters[:dim])
+        vector = MathUtil.zeros(1, parameters[:dim])
         vector.load(File.join(@data_dir, hash+'.dat'))
         vector
       end

data/lib/lsh/web.rb

@@ -23,7 +23,7 @@ module LSH
       mime_type = (params[:mime_type] || 'application/json')
       if mime_type == 'application/json'
         t0 = Time.now
-        vector = JSON.parse(params[:data])
+        vector = JSON.parse(params[:data], :create_additions => true)
         result_vectors = index.query(vector, params[:radius] || 0)
         results = []
         if params[:include] == 'id'
@@ -43,7 +43,7 @@ module LSH
         mime_type = (params[:mime_type] || 'application/json')
         if mime_type == 'application/json'
           t0 = Time.now
-          vector = JSON.parse(params[:data])
+          vector = JSON.parse(params[:data], :create_additions => true)
           results = index.query_ids_by_vector(vector, params[:radius] || 0)
           content_type :json
           { "time" => Time.now - t0, "results" => results }.to_json
@@ -66,7 +66,7 @@ module LSH
       mime_type = (params[:mime_type] || 'application/json')
       if mime_type == 'application/json'
         t0 = Time.now
-        vector = JSON.parse(params[:data])
+        vector = JSON.parse(params[:data], :create_additions => true)
         index.add(vector, params[:id])
         content_type :json
         { "time" => Time.now - t0, "status" => "indexed" }.to_json

metadata

@@ -1,15 +1,15 @@
 --- !ruby/object:Gem::Specification
 name: lsh
 version: !ruby/object:Gem::Version
+  version: 0.4.0
   prerelease:
-  version: 0.3.2
 platform: java
 authors:
 - Yves Raimond
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2013-02-04 00:00:00.000000000 Z
+date: 2013-03-18 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: jblas-ruby