np.linalg.norm

Author: Oceania2018

src/TensorFlowNET.Core/APIs/tf.linalg.cs

@@ -39,6 +39,9 @@ public Tensor diag(Tensor diagonal, string name = null)
             public Tensor matmul(Tensor a, Tensor b)
                 => math_ops.matmul(a, b);
 
+            public Tensor norm(Tensor a, string ord = "euclidean", Axis axis = null, string name = null)
+                => ops.norm(a, ord: ord, axis: axis, name: name);
+
             public Tensor batch_matmul(Tensor x, Tensor y, bool adj_x = false, bool adj_y = false, string name = null)
                 => math_ops.batch_matmul(x, y, adj_x: adj_x, adj_y: adj_y, name: name);
 

src/TensorFlowNET.Core/Binding.Util.cs

@@ -166,6 +166,8 @@ public static int len(object a)
                     return arr.Count;
                 case IEnumerable enumerable:
                     return enumerable.OfType<object>().Count();
+                case Axis axis:
+                    return axis.size;
                 case Shape arr:
                     return arr.ndim;
             }

src/TensorFlowNET.Core/NumPy/Implementation/LinearAlgebraImpl.cs

@@ -1,14 +1,26 @@
 using System;
 using System.Collections.Generic;
 using System.Text;
+using static Tensorflow.Binding;
 
 namespace Tensorflow.NumPy
 {
     public class LinearAlgebraImpl
     {
+        [AutoNumPy]
         public NDArray lstsq(NDArray a, NDArray b, string rcond = "warn")
+            => new NDArray(tf.linalg.lstsq(a, b));
+
+        [AutoNumPy]
+        public NDArray norm(NDArray a, Axis axis = null)
         {
-            return a;
+            if (a.dtype.is_integer())
+            {
+                var float_a = math_ops.cast(a, dtype: tf.float32);
+                return new NDArray(tf.linalg.norm(float_a, axis: axis));
+            }
+                
+            return new NDArray(tf.linalg.norm(a, axis: axis));
         }
     }
 }

src/TensorFlowNET.Core/Operations/linalg_ops.cs

@@ -52,6 +52,21 @@ public Tensor matrix_solve_ls(Tensor matrix, Tensor rhs,
             return _composite_impl(matrix, rhs, l2_regularizer: l2_regularizer);
         }
 
+        public Tensor norm(Tensor tensor, string ord = "euclidean", Axis axis = null, string name = null, bool keepdims = true)
+        {
+            var is_matrix_norm = axis != null && len(axis) == 2;
+            return tf_with(ops.name_scope(name, default_name: "norm", tensor), scope =>
+            {
+                if (is_matrix_norm)
+                    throw new NotImplementedException("");
+                var result = math_ops.sqrt(math_ops.reduce_sum(tensor * math_ops.conj(tensor), axis, keepdims: true));
+
+                if(!keepdims)
+                    result = array_ops.squeeze(result, axis);
+                return result;
+            });
+        }
+
         Tensor _composite_impl(Tensor matrix, Tensor rhs, Tensor l2_regularizer = null)
         {
             Shape matrix_shape = matrix.shape.dims.Skip(matrix.shape.ndim - 2).ToArray();

src/TensorFlowNET.Core/Operations/nn_impl.py.cs

@@ -109,6 +109,16 @@ public static (Tensor, Tensor) moments(Tensor x,
             });
         }
 
+        public static Tensor normalize(Tensor tensor, string ord = "euclidean", Axis axis = null, string name = null)
+        {
+            return tf_with(ops.name_scope(name, "normalize", tensor), scope =>
+            {
+                var norm = tf.linalg.norm(tensor, ord: ord, axis: axis, name: name);
+                var normalized = tensor / norm;
+                return normalized;
+            });
+        }
+
         public static Tensor batch_normalization(Tensor x,
             Tensor mean,
             Tensor variance,

test/TensorFlowNET.UnitTest/NumPy/LinearAlgebra.Test.cs

@@ -19,5 +19,13 @@ public void lstsq()
         {
 
         }
+
+        [TestMethod]
+        public void norm()
+        {
+            var x = np.arange(9) - 4;
+            var y = x.reshape((3, 3));
+            var norm = np.linalg.norm(y);
+        }
     }
 }