TensorFlow 张量操作的实现

import tensorflow as tf

# 创建标量
scalar = tf.constant(5)
# 创建向量
vector = tf.constant([1, 2, 3])
# 创建矩阵
matrix = tf.constant([[1, 2], [3, 4]])
# 创建3维张量
tensor = tf.constant([[[1, 2], [3, 4]], [[5, 6], [7, 8]]])

# 从Python列表创建
tensor_from_list = tf.convert_to_tensor([1, 2, 3])

# 从Numpy数组创建
import numpy as np
array = np.array([[1, 2], [3, 4]])
tensor_from_np = tf.convert_to_tensor(array)

# 生成全零张量
zeros = tf.zeros([2, 3])

# 生成全一张量
ones = tf.ones([3, 2])

# 生成随机正态分布张量
randn = tf.random.normal([2, 2], mean=0.0, stddev=1.0)

# 生成均匀分布张量
randu = tf.random.uniform([3, 3], minval=0, maxval=1)

a = tf.constant([[1, 2], [3, 4]])
b = tf.constant([[5, 6], [7, 8]])

# 逐元素加法
add = tf.add(a, b)

# 逐元素乘法
mul = tf.multiply(a, b)

# 矩阵乘法
matmul = tf.matmul(a, b)

# 张量求和
sum_all = tf.reduce_sum(a)
sum_axis0 = tf.reduce_sum(a, axis=0)
sum_axis1 = tf.reduce_sum(a, axis=1)

# 张量平均值
mean = tf.reduce_mean(a)

# 张量最大值
max_val = tf.reduce_max(a)

tensor = tf.constant([[1, 2], [3, 4], [5, 6]])

# 获取张量形状
shape = tensor.shape

# 改变张量形状
reshaped = tf.reshape(tensor, [2, 3])

# 转置张量
transposed = tf.transpose(tensor)

# 扩展维度
expanded = tf.expand_dims(tensor, axis=0)

# 压缩维度
squeezed = tf.squeeze(expanded)

tensor = tf.constant([[1, 2, 3], [4, 5, 6], [7, 8, 9]])

# 获取单个元素
elem = tensor[1, 2]  # 获取第2行第3列的元素

# 获取行切片
row_slice = tensor[1:, :]  # 获取第2行及以后的所有行

# 获取列切片
col_slice = tensor[:, 1]  # 获取第2列

# 使用步长切片
strided_slice = tensor[::2, ::2]  # 每隔2个元素取一个

a = tf.constant([[1, 2, 3]])  # 形状(1,3)
b = tf.constant([[4], [5], [6]])  # 形状(3,1)

# 广播加法
c = a + b  # 结果形状(3,3)

tensor = tf.constant([[1, 2, 3], [4, 5, 6]])

# 沿轴0求和
sum0 = tf.reduce_sum(tensor, axis=0)  # [5,7,9]

# 沿轴1求最大值
max1 = tf.reduce_max(tensor, axis=1)  # [3,6]

# 计算逻辑与
logical = tf.reduce_all(tensor > 3)  # False

# 计算均值
mean = tf.reduce_mean(tensor)  # 3.5

a = tf.constant([[1, 2], [3, 4]])
b = tf.constant([[5, 6], [7, 8]])

# 沿轴0拼接
concat0 = tf.concat([a, b], axis=0)

# 沿轴1拼接
concat1 = tf.concat([a, b], axis=1)

# 张量分割
split0 = tf.split(a, num_or_size_splits=2, axis=0)
split1 = tf.split(a, num_or_size_splits=[1, 1], axis=1)

tensor = tf.constant([[3, 1, 4], [1, 5, 9]])

# 排序
sorted_values, sorted_indices = tf.sort(tensor, direction='DESCENDING')

# argsort
argsort = tf.argsort(tensor)

# top-k
top_k_values, top_k_indices = tf.math.top_k(tensor, k=2)

# 张量收集
tensor = tf.constant([[0, 1, 2], [3, 4, 5]])
indices = tf.constant([0, 1])
gathered = tf.gather(tensor, indices)  # 收集第0行和第1行

# 张量分散
updates = tf.constant([10, 20])
scattered = tf.tensor_scatter_nd_update(tensor, [[0, 0], [1, 1]], updates)

# 张量条件操作
cond = tf.where(tensor > 3, tensor, tf.zeros_like(tensor))  # 大于3保留原值，否则设为0

x = tf.Variable(3.0)
with tf.GradientTape() as tape:
    y = x ** 2 + 2 * x + 1
dy_dx = tape.gradient(y, x)  # 2x + 2 = 8

# 张量转Numpy数组
tensor = tf.constant([[1, 2], [3, 4]])
numpy_array = tensor.numpy()

# Numpy数组转张量
new_tensor = tf.convert_to_tensor(numpy_array)